Home

- ePrints Soton

1. eh FIGURE A 2 Photovoltaic module schematic 160 Appendix A Module schematics PSL iva za LT VOUT V SHDN LX SS LHOIB OE3 0 Del ie ku UNOLEDOES O um A I Tra mE N Gei al UE 23 3 N Hei ech FR Ss D3 E JP1 a WOL a Sdt 50 ER 38 7 Se 53 Y N 3 Onge el E FIGURE A 2 Vibration module schematic Appendix A Module schematics 161 V VOUT SHDN LX t e LBI VFB 3 LBO GND D2 P BAT754 o 100u FIGURE A 4 Mains module schematic RES c3 DATA E E WW 1M R H2 TT e ORTUNLEEOE D ES H4 ZS e UHC B NES 0 c2 Io US H Maan e je Ju 2 U4 IRLMIL250 S PTS n gt gt 1 eyo TS9 1 E DS2502 FIGURE A 5 Supercapacitor module schematic 162 Appendix A Module schematics Foi Iva A GND VBATT EN e IONEALOLE3 0 ae ONEA4DLE3 0 H1 M NI M g9 YACIOH VVWWSC Oe yi veo gz ad vve b OSl 1S 7 19 Wi q yas E 2 FR a g T WL gt La a oa Sx WL vo
2. Lead acid NiCd NiMH Lithium Specific energy Wh kg 30 35 75 150 Energy density Wh L 90 100 240 400 Nominal voltage V 2 0 1 2 1 2 4 0 Open circuit voltage V 2 1 1 29 1 4 4 1 Operating voltage V 2 0 1 8 1 25 1 00 1 25 1 10 4 0 3 0 End voltage V 1 75 1 0 1 0 3 0 Self discharge month 4 8 15 20 15 25 2 Calendar life years 2 8 2 5 2 5 2 Cycle life cycles 250 500 300 700 300 600 1000 Discharge profile Flat Very flat Flat Sloping Advantages Long life Good low T High E High E on float T high rate density density independent Limitations Cant store Memory effect Moderate cost Low rate discharged low cycle life a At room temperature 25 C TABLE 2 2 Rechargeable battery types and characteristics adapted from 8 16 Chapter 2 Background Energy Sensing and Wireless Communication 2 2 4 Supercapacitors Supercapacitors also known as ultracapacitors electrochemical capacitors or electric double layer capacitors offer a much higher energy density than aluminium electrolytic capacitors but retain many of their other characteristics Unlike conventional capaci tors which have two plates separated by a dielectric supercapacitors have two layers of the same substrate typically activated carbon which store charge separated by a very thin material Their operation is more complex than conventional capacitors but they loosely follow the standard capacitor equation
3. Technology and Applications International Forum Proceedings Albuquerque New Mexico January 2000 L Mateu and F Moll Review of energy harvesting techniques and applications for microelectronics Proceedings of SPIE The International Society for Optical Engineering 5837 Part 1 359 373 2005 P D Mitcheson E M Yeatman G K Rao A S Holmes and T C Green Energy harvesting from human and machine motion for wireless electronic devices Proceedings of the IEEE 96 9 1457 86 2008 N H Reich W G J H M v Sark E A Alsema S Y Kan Silvester A H v d Heide R W Lof and R E I Schropp Weak light performance and spectral response of different solar cell types In Twentieth European Photovoltaic Solar Energy Conference Proceedings of Barcelona Spain 2005 J F Randall and J Jacot Is AM1 5 applicable in practice Modelling eight photovoltaic materials with respect to light intensity and two spectra Renewable Energy 28 12 1851 64 2003 SANYO Semiconductor Co Ltd Amorphous silicon solar cells amorphous pho tosensors http semicon sanyo com en pamph_pdf_e EP120B pdf Novem ber 2007 Last accessed May 2010 Schott Solar GmbH ASI OEM Indoor Solar Modules http www schott com photovoltaic english products oem_products 2010 Last accessed May 2010 S Roundy Energy scavenging for wireless sensor nodes with a focus on vibration to electricity conversion PhD Thesis
4. lt 4 uajuoy voyeuoju 1 eee e 1 Hh oat Empty Batte FIGURE 2 22 Priority balancing in IDEALS reproduced from which shows how the energy priority is balanced against the packet priority shown in Figure It should be noted that the range of priority values in this case EP 0 5 and PP 1 5 is largely an arbitrary choice but the EP must start at 0 and PP at 1 with both the EP and PP having the same maximum value While the figure refers only to battery powered nodes the scheme can also be applied to any form of energy store and is highly suited to wireless sensor nodes that harvest energy from their environment and can therefore have a rapidly changing energy status In this way the system will always operate to ensure that important messages can traverse the network dropping fewer important messages where necessary to conserve energy and hence sustain the operation of the network Merrett also makes the distinction between relative thresholds and absolute thresholds This allows networks comprised of nodes with differently configured energy subsystems to cooperate nodes either set their EP by the absolute amount of energy stored or by proportionally how full their energy stores are 2 6 3 Achieving energy awareness The energy adaptive algorithms described in Section 2 6 2 must be able to monitor the energy status of the node At its very simplest the system should monitor the amount
5. Input displacement m Spring displacement m Abbreviations uC Microcontroller ADC Analogue to Digital Converter AEASN Adaptive Energy Aware Sensor Networks AM1 5 Air Mass 1 5 AMR Automatic Meter Reading ANSI American National Standards Institute ASI Amorphous Silicon CEDS Component Electronic Data Sheet CENS Center for Embedded Networked Sensing CHI Common Hardware Interface DIF Data Information Fusion DTC Defence Technology Centre EAN Energy Analysis Layer ECO Energy Control Layer EEDS Energy Electronic Data Sheet EMA Energy Management Architecture EP Energy Priority EPROM Electrically Programmable ROM EPSRC Engineering and Physical Sciences Research Council ESD Electronic Systems and Devices GPS Global Positioning System XXV xxvi ABBREVIATIONS HAL Hardware Adaptation Abstraction Layer HEDS Health Electronic Data Sheet HeH Hybrid Energy Harvester HIL Hardware Interface Layer HPL Hardware Presentation Layer HVAC Heating Ventilation and Air Conditioning I O Input Output IDEALS Information manageD Energy aware ALgorithm for Sensor networks IEEE Institute of Electrical and Electronics Engineers IrDA Infrared Data Association ISA International Society of Automation ISM Industrial Scientific and Medical KTN Knowledge Transfer Network LAN Local Area Network LPM Low Power Mode MIT Massachusetts Institute of Technology NiCd Nickel Cadmium NiMH Nickel Metal Hydride PC Personal Computer P
6. the onewireWriteByte function e onewireReadBlock reads a block of data of a specified length from the 1 Wire EPROM memory e onewireReadByte reads a single byte of data from the 1 Wire EPROM memory e onewireReadBit reads one bit from the 1 Wire bus It is generally called by the onewireReadByte function e onewireUpdateCRCByte permits a CRC to be generated from the data read from the 1 Wire device This can the be compared to the CRC generated by the device itself in order to check for communication errors 5 3 3 Performance costs of 1 Wire communications As discussed in Section 4 3 3 1 Wire ICs are two terminal devices which have one pin connected to ground with the remaining pin being used for both data and power The DS2502 device used in this project requires a 5kQ pull up resistor on the bus line 128 Chapter 5 Case Study Deployment in a Prototype System Software As covered earlier the design of the system means that the 1 Wire devices draw no quiescent current when the system is not actively communicating with energy devices i e when it is sitting in address 0 At other times when the device is not actively engaged in communications the input load current is typically 5uA At times when the microcontroller needs to pull the bus low there is clearly an amount of current dissipated through the 5kQ resistor which is proportional to the regulated voltage for example with a 3V regulated vol
7. Beg Gok Pek bo std wa 131 cua 132 5 6 Code to interface with Simple Packet Protocol communication stach 134 5 7 Code to interface with SimpliciTI communication stack 135 o 144 MEA eee eee ee eS 145 XV Declaration of Authorship I Alexander Stewart Weddell declare that the thesis entitled A Comprehensive Scheme for Reconfigurable Energy Aware Wireless Sensor Nodes and the work presented in the thesis are both my own and have been generated by me as the result of my own original research I confirm that e this work was done wholly or mainly while in candidature for a research degree at this University e where any part of this thesis has previously been submitted for a degree or any other qualification at this University or any other institution this has been clearly stated e where I have consulted the published work of others this is always clearly at tributed e where I have quoted from the work of others the source is always given With the exception of such quotations this thesis is entirely my own work e I have acknowledged all main sources of help e where the thesis is based on work done by myself jointly with others I have made clear exactly what was done by others and what I have contributed myself e parts of this work have been published as listed in Section 1 6 of the thesis Signed Date xvii Acknowledgements I would like to sincerely thank Nick Harris and Neil Whi
8. G Tolle K Whitehouse and D Culler Trio enabling sustainable and scalable outdoor wireless sensor net work deployments The Fifth International Conference on Information Processing in Sensor Networks pages 407 15 2006 8 D Linden and T B Reddy editors Handbook of Batteries McGraw Hill New York 3rd edition 2002 9 S Jacobs Utility meter operating 20 years on original lithium battery Metering International 3 1 pp 2004 191 192 BIBLIOGRAPHY 10 11 12 13 14 15 16 17 18 19 20 21 22 Tadiran Batteries GmbH Lithium batteries technical brochure http www tadiranbatteries de eng downloads lbr06eng pdf October 2008 Last ac cessed May 2010 D Rakhmatov S Vrudhula and D A Wallach A model for battery lifetime analysis for organizing applications on a pocket computer IEEE Transactions on Very Large Scale Integration VLSI Systems 11 6 1019 30 2003 Maxim Dallas Rechargeable Batteries Basics Pitfalls and Safe Recharging Practices 2005 Application Note AN3501 J M Tarascon and M Armand Issues and challenges facing rechargeable lithium batteries Nature 414 6861 359 67 2001 F Simjee and P H Chou Everlast long life supercapacitor operated wireless sensor node ISLPED 06 Proceedings of the 2006 International Symposium on Low Power Electronics and Design pages 197 202 2006 X Jiang J Polastre and D Culler Per
9. In this application the ecoUpdateEnergy function is called to check the energy status of the node and update the EP value Dependent on the EP value if it is different from EP_EMPTY or EP_UNKNOWN the system will sense the temperature and transmit it Another application layer function appGetData is called which obtains the data from the sensing stack and formats it into a message to be transmitted through the SimpliciTI protocol stack The EP value then dictates the period that the node will sleep for In this implementation the maximum sleep period is 60s and the minimum is ls Provided that a minimum 16s reporting interval can be supported by the power source on the node the system will continually operate and will generally settle into one of the modes or oscillate between two of them given a constant power input alternatively if no power is harvested by the system the device s reporting frequency will decline gracefully as the stored energy is depleted 146 Chapter 5 Case Study Deployment in a Prototype System Software 5 6 2 Energy related performance As stated above a number of mathematical functions have been implemented which allow the energy status of the node to be computed The most resource intensive cal culation is that for the photovoltaic module which involves a substantial amount of floating point arithmetic This was taken as a worst case example to evaluate the time taken to estimate the powe
10. Options for overvoltage protection circuits a uses a MOSFET to dis connect the energy harvester and thus limit the voltage while b simply uses a linear regulator the voltage detector and limits its maximum output voltage to the rating of the voltage regulator This circuit is recommended for use on the multiplexer module but the other circuit configurations are used on energy modules Voltage Regulator Voltage Detector Vino oVout FIGURE 4 4 Combined undervoltage and overvoltage protection circuit The suggested circuit has been used in the prototype multiplexer module in this case study Figure shows the results of testing of this functionality It can be observed that when the supply voltage rises above 2V it is tracked by the output voltage until it is regulated to 3V The output voltage then falls to zero when the supply voltage drops below 2V This operation is achieved using a Torex XC61C 2 0V CMOS output voltage detector and a Torex XC6215 LDO 3 0V linear regulator The quiescent power consumption of the voltage detector has been verified at below 1A with the voltage regulator consuming a similar amount of quiescent current The figure shows that the voltage drop through the regulator is also negligible 4 3 6 Power and energy estimation Some methods for power monitoring were introduced in Section 3 6 6 In the prototype modules produced in this case study closed circuit current measurement and open
11. Primary battery 01001011 Rechargeable battery 01001100 Supercapacitor 10000100 Mains module 10001000 PV module 10001100 Vibration electromagnetic harvester 10010000 Vibration piezoelectric harvester 10010100 Vibration electrostatic harvester 10011000 Thermoelectric harvester 10011100 Wind harvester TABLE 3 3 Examples of module type identifiers 3 3 3 Hardware and interrogation method As discussed in Section 2 4 2 TEDS is most commonly implemented with Maxim Dallas 1 wire EPROMs 1 wire devices communicate using a timing specific protocol and re quire only two connections ground and a combined data and power pin The capa bilities of the 1 wire protocol mean that it is possible for a large number of devices to share the same single wire bus The addressing of devices in this configuration is not straightforward as the master has to go through an initial search process to identify all devices on the line in order to determine their serial numbers when it has done this it will still be uncertain about the actual physical location of each device In most appli cations this does not matter however in this situation it is essential to know which device is connected to which specific port on the multiplexer module in order that it may be monitored and managed individually using the port number as the address In this system this problem is mitigated by t
12. buffer or otherwise supply energy to the multiplexer module in order that it can reliably and efficiently supply power to the microcontroller Given that the multiplexer module will generally operate with an unregulated voltage of up to 4 5V it is expected that energy modules will also self regulate their outputs to this maximum Each module is effectively able to operate autonomously and must function in a stable manner without external control Care must be taken when designing energy modules to ensure that they will allow the system to start from cold A further detail is that the energy module should not be damaged by not being connected to the multiplexer module i e they should not rely on their outputs being connected in order to maintain safe operation Therefore as a general rule e Energy harvesters and mains electricity modules should be designed to initialise supplying energy to the system by default e Fast response buffers such as supercapacitors should on system start up default to permitting both charge and discharge e Slow response buffers such as rechargeable batteries should by default be per mitted to discharge but may support a manual override facility e g using push buttons e Primary energy sources such as non rechargeable batteries will normally have a push button control to allow the system to cold start on initial installation then be managed by the microcontroller Their default s
13. 158 EEN 159 A 3 Vibration module schematic 2 a a a e a 160 ge ceed aire he e eT hid RRR Boe ee et AE 161 A 5 Supercapacitor module schematic 20200000 161 Ap Battery module schematic 2 0 0 0 20000 eee ee ee 162 List of Tables ana bu hee oe 13 e 15 neon celeron 19 2 4 List of vibration oource e 21 at 31 2 6 Basic TEDS content as defined in IEEE 1451 4 2004 33 2 7 Thermistor Template Summary 0 0 000502 e ee 34 3 1 Outline format for the Energy Electronic Data Sheet 64 3 2 Module class code 64 3 3 Examples of module type identifiers 65 A 67 o AE 69 8 7 Energy Priority Levels css gh Peano e a heh ee bee 75 3 8 Simplified discharge of alkaline AA cell 79 4 1 Interface pins from multiplexer module address 0 104 4 2 Interface pins from multiplexer module address 7 104 4 3 Interface pins from photovoltaic module 106 4 4 Parameters found for Indoor PV Module 109 4 5 Interface pins from vibration module o o o o es 112 4 6 Interface pins from wind module 112 HE 113 4 8 Interface pins from mains module lt 113 BEEN 115 ee ee 116 EES 117 Eos eee 132 ifs Od dea Bate A 133 DEENEN 163 erent 163 e St 164 SES 164 xiii Listings ee ee E 130 oo Gene ae Red Ee eee be ee 130 oe far tee SS Sc ee Rt ce es 131
14. 1M 1 FIGURE 4 6 Example of voltage measurement circuit with resistor values permitting measurement of the open circuit voltage from the operational amplifier can be considered to be a low impedance buffered replica of the divided voltage The measured voltage from the voltage divider is dependent on the standard voltage divider equation Vmeas Ra 4 1 V Ri R et The total impedance presented to the device in question can simply be expressed as Rio Ri R2 4 2 Therefore with knowledge of these equations the impedance of R and Ra can be set at appropriate values to ensure that the divided voltage is within the acceptable range for all possible values of the measurand For this purpose v will represent the maximum raw input voltage and w will be the maximum voltage detectable by the microcontroller ADC input Therefore for this application Ra e E 4 3 R Ez Sg Ge Methods for determining the state of charge or energy stored in a supercapacitor Sec tion and battery Section 3 6 5 have been introduced The above scheme is applicable to these applications For supercapacitors open circuit voltage is normally used to estimate the amount of energy stored for batteries the voltage across a known load is normally tested The necessary parameters to convert this into an estimate of stored energy are held on the electronic data sheet on each module 102 Chapter 4 Case Study Deployment in a
15. 1cm and has been demonstrated powering a wireless sensing and transmission system A drawback of this generator is its fragility and the intricate processes required in its manufacture for example the coil has 2 800 turns using 12um diameter copper wire While the manufacture of electromagnetic vibration energy harvesters is generally in tricate and expensive their major strength is that they are relatively low impedance sources that produce moderate voltages that can be efficiently rectified and used to power electronic devices Chapter 2 Background Energy Sensing and Wireless Communication 23 Mk3 Generator A Generator B System FIGURE 2 8 A miniaturised electromagnetic microgenerator a the total volume of the generator is approximately 1cm and b it has been incorporated into a wireless microsystem reproduced from 36 Piezoelectric Piezoelectric materials have two complementary properties they deform when subjected to an electric field or produce electrical charge when mechanically deformed The second property is of greatest interest in this application as it provides the basis for electrical power generation from vibration Resonant structures can be fabricated and the periodic deformation of the piezoelectric material causes electrical charge to be produced during each oscillation Piezoelectric generators have the advantage that their design analysis and fabrication is straightforward It is relatively e
16. 31 propriate type of cell is important especially in low light or artificially lit environments Developments have been made in recent years in the area of vibration energy harvest ing commercial devices which harvest vibration from the environment and convert it into electrical energy are now emerging in the marketplace Present technologies are sensitive to specific vibration frequencies so the technologies are typically applicable to deployment on machines working from a mains power supply or a at a fixed frequency Thermoelectric devices are also being developed but are perhaps the least mature ma jor technology covered in this section of the thesis due to the difficulty of maintaining a suitable temperature difference across the thermocouple and of using this to generate a sufficient voltage Other technologies have also been discussed in this section including the exploitation of fluid flow or wind power and various forms of power from human heat and movement Wireless energy transfer from induction or radio frequencies has also been considered but the range of such technologies is limited and most rely on the use of a dedicated high power transmitter to transfer power to the receiver Power Source Power Buffering Voltage Commercially pW Zem Required Regulation Available Solar outside 15 000 Usually Maybe Yes Solar inside 10 Yes Maybe Yes Temperature 40 Usually Maybe Yes Human Power 330 Y
17. 31 The MPWiNodeX architecture and deployment 54 xl xii LIST OF FIGURES 3 1 A modular energy subsystem connected to a sensor node 60 EECHER 66 LaRoche BE EE 70 eege 70 rere rere 75 SE 77 3 7 Typical discharge profile of Duracell MN1500 alkaline cet 79 3 8 A combined stach 81 3 9 A basic template stack 82 A TN 83 pias E a ee a 85 4 1 Bistable multivibrator s a a 97 4 2 Undervoltage protection rout 98 DEENEN 99 4 4 Combined undervoltage and overvoltage protection circuit 99 4 5 Test of combined undervoltage and overvoltage protection circuit 100 4 6 Example of voltage measurement rout 101 ig eG oe ee ee ee ee das 102 4 8 Simplified schematic of multiplexer module showing data connections 103 4 9 Simplified schematic of multiplexer module showing power connections 103 Gab endear nd 106 4 11 The developed PV energy harvesting circuit 108 4 12 SPICE model for small PV cellsl e 108 ee 109 eegen 111 4 15 Circuit board for the vibration model 111 at eee BA De OS ac Bo ed 113 EE 114 EE 117 E TADA UA A 118 se Othe geet da di dle 123 E E eae os 124 5 3 Set up for EEDS on energy module to be programmed 126 dete etre 129 5 5 A detailed combined stack oa aoa a a a 134 5 6 The SimpliciTI communication stach 135 Bae pae E ce 140 A ea ee e 143 5 9 Annotated Hyperterminal output from second test 148 EE
18. EEDS on the multiplexer module is also accessed through address 7 the provision of the multiplexer module EEDS facility on address 7 rather than address 0 means that the one wire bus does not consume power while the system rests in its default address 0 state 4 4 3 Additional features Additional circuitry includes pull down resistors for the address lines and a pull up resistor for the 1 wire bus Diode clamping is provided by an array of NXP BAT754 Schottky diodes which prevent the outputs from the multiplexer module from exceeding the limits of the supply lines of the microcontroller As shown in the photograph three push switches are located on the top edge of the board these are an artefact of an earlier version of the scheme where all four management lines were of mixed purpose and the push buttons were used to trigger interrupts on the microcontroller they were connected to the low power monitoring lines of the system on address 0 The direction and type of each of the four management lines is now mandated measurement control measurement and 2x control to remove the potential for dangerous conflict when energy modules are exchanged during node operation The facility is currently provided to trigger an interrupt when the unregulated voltage passes 2 7V by the use of an additional voltage detector and MOSFET arrangement This is an arbitrary selection but is used to show the ability of the system to
19. Figure around 25 of the light unit is obscured by photovoltaic cells 18 This type of parasitic energy harvesting is undesirable as it directly impacts on the performance of the device it is attached to in this case substantially reducing the amount of light radiating from the light fitting A rather more useful report on the subject of low power photovoltaic energy harvesting is by Leder et al with a basic circuit for storing energy in a supercapacitor and quick start up for the circuit provided by charging a smaller capacitor directly through a linear regulator 119 Dondi et al present a model for a photovoltaic cell and a method of tracking the maximum power point of a large cell by using a smaller secondary cell to provide a voltage reference 120 Texas Instruments have recently released a version of their eZ430 RF2500 which is powered from a photovoltaic module and is capable of operating indoors 121 FIGURE 2 29 Photovoltaic cells harvest energy from a ceiling mounted fluorescent unit but obscure much of the light reproduced from Hande et al 2 9 3 Systems integrating multiple energy resources Few projects have incorporated multiple energy resources onto a single node An early example is PUMA 122 developed at the University of California Irvine which uses Chapter 2 Background Energy Sensing and Wireless Communication 53 power routing switches to connect parts of a sensor node to energy harvestin
20. I O Lite HAL Interface with physical hardware no formal PHY Ka E E e oO E E a Ku Communication Hardware Any TI transceiver hardware sub 1GHz or 2 45GHz FIGURE 5 6 The SimpliciTI communication stack adapted from 72 5 4 3 Sensing stack The structure of the sensing stack was introduced in Section A basic sensing stack has been developed to interface with the on chip temperature sensor The workings of this system are similar to the energy stack The PYS layer interfaces with the ADC to obtain an analogue reading from the temperature sensor This is translated by the SPR layer which accounts for offset and gradient through pre determined values that must be derived for each individual device The final temperature reading is presented by the SEV layer to the shared application layer As no electronic data sheet format has been developed for the sensing hardware it was necessary to hard code the device models into the stack software Clearly in future iterations of the system it will be desirable to be able to connect a range of sensors to the system and for them to self identify their 136 Chapter 5 Case Study Deployment in a Prototype System Software interface and calibration coefficients in a similar way to the energy hardware 5 4 4 Energy stack An energy stack compliant with the structure introduced in Section has been developed and is described in full in Section The energy stack has
21. North America and 2 45GHz worldwide The 2 45GHz band has been widely adopted and offers data rates up to 250kbit s and a transmission range of up to 100 metres Since the release of IEEE 802 15 4 2003 there have been two revisions to the standard Additional functionality has to be added to 802 15 4 to deliver routing network manage ment and application support To this end the ZigBee Alliance has defined standards for embedding wireless communications capabilities 66 ZigBee compliant devices are capable of star tree or mesh network topologies potentially taking several hops to route data from one sensor through other nodes capable of routing to a central data collector The ZigBee stack is shown in Figure and is intended to be based on cheaper hardware than Bluetooth with a simpler software stack and much lower energy requirement Many of the energy savings are derived from faster handshaking with Zig Bee devices able to join a network and make a transmission in a very short time period Other 802 15 4 based protocols such as MiWi 67 and WirelessHART 68 have been developed but do not have the capabilities support or level of acceptance of ZigBee Wireless local area networks wireless LANs or WLANs are defined by the IEEE 802 11 WiFi standard first released in 1997 and its amendments 69 802 11g 70 a recent international amendment to the original standard operates at 2 45GHz and permits data rates of up to 54
22. Seattle 2008 Mid Technology Corporation SEH25w Piezoelectric Vibration amp Solar Energy Harvester http www mide com products volture seh25w php 2009 Last accessed May 2010 D Kraemer L Hu A Muto X Chen G Chen and M Chiesa Photovoltaic thermoelectric hybrid systems a general optimization methodology Applied Physics Letters 92 24 243503 1 June 2008 Watlow Ltd Watlow sensors http www watlow co uk products sensors 2004 Last accessed May 2010 IEEE Standards Association IEEE standard for a smart transducer interface for sensors and actuators mixed mode communication protocols and transducer electronic data sheet TEDS formats IEEE Std 1451 4 2004 2004 National Instruments Corporation An Overview of IEEE 1451 4 Transducer Elec tronic Data Sheets Technical report National Instruments 2004 D Potter Smart plug and play sensors Instrumentation amp Measurement Maga zine IEEE 5 1 28 30 Mar 2002 196 BIBLIOGRAPHY 59 60 61 62 63 64 65 66 67 68 69 70 H M Willey One cheap network topology 1 wire bus Embedded Systems Pro gramming 14 1 59 76 2001 S Bandari C Santiago H S Mohammed and J Schmalzel Component elec tronic datasheets in ISHM pages 106 109 Houston TX United States 2006 J L Schmalzel F Figueroa J A Morris and S A Mandayam A road map for integrated systems health manageme
23. This is an important development as it means that sensor nodes no longer need to be developed for specific energy hardware and appropriate energy devices can be attached to the sensor node at the time of system deployment The scheme also permits the energy devices to be changed after deployment and allows the microcontroller to detect these changes and adapt the node s operation accordingly The energy demands of the implemented system are conservative provided that the Chapter 6 Conclusions and Future Work 153 update frequency of the energy status is kept low The quiescent power draw of each module has been measured as of the order of a few microwatts and the multiplexer module which enables the scheme also draws similar levels of power Therefore in a milliwatt scale sensor node which was the intended application of this technology the quiescent power consumed by the hardware enabling this scheme is dependent on the actual hardware connected typically a few tens of microwatts Notably it has been observed that a single Schottky diode in a 1mA current path causes an approximate 9 efficiency loss transistors were used wherever possible but in some applications it was necessary to use diodes for the purposes of device protection Many of the diodes would be required in any energy harvesting system so this should not be considered to be a drawback that is specific to this scheme rather it is a reminder that careful design is
24. Wireless Identification and Sensing Plat form has been shown to work with a UHF RF ID reader which has a 4W effective radiated power The WISP complete with its 15cm antenna is shown in Figure 2 16a and the voltage obtained from the harvesting subsystem is shown in Figure 2 16b The same team have also demonstrated a system which is capable of harvesting energy from broadcast television signals The device shown in Figure was shown to harvest 604 W sufficient to power a digital thermometer hygrometer when located 4km from a transmitter broadcasting at 960kW at 674 680MHz The system features a 5dBi log periodic antenna which had to be manually oriented to face the transmitter with line of sight While both developments are interesting the limited power outputs and the requirement to be close to a dedicated transmitter or have direct line of sight with a broadcast station using a manually oriented antenna limits their range of applications 30 Chapter 2 Background Energy Sensing and Wireless Communication 37 46 Calculated Distance m 18 23 29 e Output v WISP Threshold Voltage 1 9 4 Uplink Packet Errors 5 Responses Per Query Voltage v Pi E D D S 2 4 6 8 Peak Recieved Power dBm a B FIGURE 2 16 Devices from Intel Research a the WISP device which is a passive RF ID tag with gt 1m range b its performance c a system harvesting energy from broadcast TV signals reprod
25. allows the EEDS for the multiplexer module to be read through address 7 and leaves the remaining addresses channels 1 through to 6 for other energy modules The presence of a multiplexer module EEDS on a channel other than 7 would indicate a configuration or hardware error 3 4 3 Interface format With the conventional method of searching a 1 wire bus outlined earlier in this appli cation it would still be unclear which module is connected to each multiplexer module socket hence making monitoring or control of energy modules through their correct sockets impossible In this system the 1 wire EEDS devices are connected through the multiplexer module in the same way as the control and measurement signals so that they can be accessed one module at a time by channel rather than serial number cor responding to the actual socket the modules are connected to on the multiplexer module The scheme shown in Table 3 4 is used to define the multiplexed address of each module Address Binary Channel Module Type Functionality 0 000 Multiplexer Low Power Monitoring 1 001 1 Energy module Module dependent 2 010 2 Energy module Module dependent 3 011 3 Energy module Module dependent 4 100 4 Energy module Module dependent 5 101 5 Energy module Module dependent 6 110 6 Energy module Module dependent 7 111 Multiplexer Active interrogation TABLE 3 4
26. and due to their thin insulator they will typically have a lower operating voltage They are relatively insensitive to charge discharge cycling and require no special charging circuitry 14 However their energy density remains an order of magnitude below those of conventional secondary batteries and their leakage or self discharge rates are also relatively high Superca pacitors have been incorporated into self powered sensing systems where they provide a cache between harvesting sources a secondary battery and a sensor node Jiang et al carried out experiments to determine the self discharge rates of a number of su percapacitors and found that the worst performing supercapacitor lost approximately 50 of its energy through leakage in 16 hours and that the capacitor size does not give an indication of likely self discharge rate There is however some doubt over whether the observed effects were attributable to leakage or the redistribution of charge within the supercapacitor A common range of supercapacitors is the Panasonic Gold series their technical guide has information on the modelling and behaviour of supercapacitors Wireless sensor nodes such as the those described in Section 2 7 1 draw around 30mA when fully active and communicating excluding the current draw of any additional sensors or peripher als The Panasonic HW series has a maximum operating voltage of 2 3V maximum operating temperature of 70 C and guaranteed li
27. bars if appropriate e Physical Sensing PYS interfaces with the physical sensor hardware through ADCs or digital communication ports obtaining raw measurements Shared Application Sensor Evaluation Presentation to application layer Sensor Processing Compensation for offset and gradient ing a E Ka E D 2 D E FIGURE 3 11 A Sensing Stack reproduced from 124 The focus of this project is on energy management and the development of the system to contain stacks for interfacing with sensing energy and communications hardware The main task of the sensing stack in this instance is to provide data to be transmitted by the system A basic sensing stack has been implemented in the prototype and is described in Section 86 Chapter 3 Development Towards a Reconfigurable Energy Subsystem 3 8 General system operation 3 8 1 Start up On initial system start up the EEDS on the multiplexer module is queried by the microcontroller on the sensor node This provides information including how many energy modules are supported by the multiplexer module The microcontroller then queries the other modules in the energy subsystem The important parameters are stored in the microcontroller s memory and hence it is not necessary for the microcontroller to regularly query the EEDS on the modules as a matter of course The default behaviour of systems on first installation is to allow the energy harvesting
28. charge up the supercapacitor module Once this store reaches approximately 2 1V the system connects the power supply to the CC2430 which then starts up and tests its voltage When this has reached a suitable level nominally 2 7V the micro controller then performs the first energy intensive tasks such as scanning its energy subsystem To deliver a near instant start up to the system the on button may be pressed on the primary battery module which will cause the system to receive power from the battery Once the microcontroller has taken control of the energy subsystem the microcontroller disconnects the primary battery in order to conserve the charge level on the cell Alternatively the mains adapter may be turned on which would also act to rapidly charge up the supercapacitor module However the microcontroller cannot act to turn off this supply as it is assumed to be a zero cost resource which should be taken ad vantage of whenever it is available The first scan of the energy subsystem is used to ascertain which sockets are occupied what types of device are present and their oper ating parameters From this initial scan the microcontroller can reach an estimate of the amount of energy stored by the system This data is stored in the microcontroller Chapter 4 Case Study Deployment in a Prototype System Hardware 119 memory so the EEDS of each module need only be scanned once 1 Wire activities are power intensive so
29. circuit voltage measurement have been used to estimate the amount of power being produced While in line current measurement current shunt would be desirable un fortunately due to practical reasons it is very difficult to use in this system For in line 100 Chapter 4 Case Study Deployment in a Prototype System Hardware Tek SL 2 CH2 1 00 M 25 0s FIGURE 4 5 Test of combined undervoltage and overvoltage protection circuit Grey line denotes raw voltage on energy multiplexer black line shows regulated voltage provided to microcontroller current measurement a resistor must be connected so as to interrupt the positive supply line the differential voltage across that resistor must then be measured This is problem atic as the resistor voltage must be small of around 12 to avoid affecting the efficiency of the system so the measured voltage will be small the real performance of micropower rail to rail operational amplifiers is also imperfect It would also be non trivial to decide which supply rail to drive the operational amplifier from if using the raw voltage on the module the output may be outside the acceptable input range for the microcontroller alternatively if driven from the regulated voltage it would be impossible to achieve operation with inputs up to the raw supply rail voltage on the module For this reason the alternative schemes of measuring the open circuit voltage or current th
30. curve as shown in Figure 3 7 and rep resented for the primary battery in the electronic data sheet in Table 5 1 The function accepts the measured voltage as an unsigned char input and returns the energy stored as an unsigned int in tens of microwatts eanMathExponential takes two inputs an exponent as a float and the number of terms to compute it to as an unsigned char The functions returns the Taylor expansion of e computed to the specified number of terms increasing number of terms improves its precision but impacts on performance and can result in overflow if taken to extremes The result is returned as a float eanMathFactorial takes an unsigned int as an input calculates its factorial and returns it as an unsigned long It is required by the other mathematical functions that carry out Taylor expansions eanMathPower takes two inputs a number as a float and a power as an unsigned char It returns the number raised to the power as a float e eanMultiplyRaise takes three inputs a number x as an unsigned int a mul tiplier A as an unsigned char and a power y as an unsigned char The function returns the result of AxY as an unsigned int eanCalculateComplex is a custom function used for calculating the power from a photovoltaic module It takes two inputs the measured voltage and the raw voltage on the multiplexer module both as unsigned char variables in order to estimate the nominal power from the ph
31. curves at various light levels This cell is optimised for indoor use and at lower light levels it has higher efficiency levels than alternative technologies As an example a module with dimensions 90mm x 72mm has a nominal power of 324uW at 200 Lux and 1 79mW at 1 000 Lux A challenge is to run the cell at its optimal operating point in order to maximise the obtained power Dynamic loads such as wireless sensor nodes cannot generally be run directly from the cell as this will result in a varying load impedance typically a secondary battery or storage capacitor is used to buffer energy for the system Intensity dependence Power density Power density E 200 Lux S 200 Lux E 100 Lux d ve Ts Current voltage characteristics V V 0 100 200 300 400 500 600 mV 0 100 200 300 400 500 600mV A Module B Intensity depen Cc Power density dence FIGURE 2 5 The appearance and operating characteristics of a Schott Solar ASI Indoor Photovoltaic Module reproduced from 28 Cells are reliant on sufficient levels of light penetrating the glass to reach the p n junction Over a 10 year deployment significant residue can be expected to build up on a sensor node package especially in an industrial environment This can significantly degrade the cell performance reducing the harvesting efficiency It must be ensured that the package can be cleaned easily or that the effects of dust and residue are minimised 2 3 3 Vibration energy harvestin
32. defining common interfaces and electronic data sheet templates for transduc ers This allows data acquisition systems to obtain scale and interpret transducer data without the need for manual configuration when installed A 1451 4 compliant system will normally consist of a smart sensor that is connected to a data acquisition system which interfaces through a network capable application processor over a network to a display device such as a PC An example of a typical smart sensor is from the Watlow INFOSENSE P transducer family which implements the IEEE 1451 4 standard As shown in Figure 2 18 the smart sensor has its Transducer Electronic Data Sheet TEDS data stored in a memory on the device s connector rather than the conventional approach of printing the information on a tag attached to the cable The process of installing identifying and calibrating the transducer is therefore automated and takes place as soon as the device is attached to the system In practice TEDS is a very useful feature for autonomous sensors as it adds a plug and play capability to conventional dumb sensors By allowing devices to store detailed operational interfacing and identification data about themselves it frees system installers from having to individually configure the interfaces between digital systems and analogue sensors This can reduce the deployment cost of sensors and increase reliability a INFOSENSE B INFOSENSE P F
33. device s to charge up the short term energy stores such as the supercapacitor modules Once the store voltage passes a threshold nominally 2 1V but ultimately dependent on the minimum operating voltage of the microcontroller the system connects the power supply to the microcontroller which then starts up and tests its voltage When this has reached a suitable level nominally 2 7V the microcontroller will perform the first energy intensive tasks such as scanning its energy subsystem The first scan of the energy subsystem is used to ascertain which sockets are occupied what types of device are present and their operating parameters From this initial scan the microcontroller can reach an estimate of the amount of energy stored by the system This data is stored in the microcontroller memory so the EEDS of each module need only be scanned once 1 Wire activities are energy intensive so it is undesirable to carry out unnecessary communications 3 8 2 Default operation In normal operation after the configuration phase the microcontroller periodically monitors the energy modules in order to gauge the energy status of the sensor node To do this the multiplexer address lines are set to the device of interest and the control and measurement lines are used to control the modules and obtain measurements For example with the photovoltaic module the measurement control line is used as an output to cause a transistor to disconnect the ce
34. eee aa at ee E Goo ee 53 210 Summary ae e we Be ek oe dee ae Pad be ee ee ae H 54 55 de AE A ee a a a eae Ba Shae 55 3 2 Design for reconfigurability 2 2 0 02002000002 ee eee 55 EE 55 bod Boe OR Gee Soe Be Soe HGRA ee eer aes 58 bg oes e ae gene EE EE Ae ee 60 3 2 4 Major challenges and strategy 00200004 61 3 3 Energy Electronic Data Sheet EEDS 63 ae Sow ow A Gua we A Adee da 63 3 3 2 Datasheet format aoaaa 64 3 3 3 Hardware and interrogation method 65 3 4 Common Hardware Interface CHI o 66 3 4 1 Overview and justification 2 0 0 2020 00 66 bei E aaa eee Sk ah ee Se Gg 67 3 4 3 Interface format 2 ee ee 67 3 4 4 Integration of multiple energy sources 69 CONTENTS vii EE 70 3 5 1 Energy multiplexer 2 020202 000 eae 70 padoe s eang ee ee a a 71 Eua ea A ee eh a ee Ge eG 72 SAE ME E ot de ea tea vee h 73 Peach e uti gS EE 73 3 6 1 Overview 73 3 0 2 Energy monitoring e Lei Dee ea eR RR 74 rara dada aa 6264 75 3 6 4 Supercapacitor state of charge and capacity 76 3 6 5 Battery state of charge and capacity o 78 3 6 6 Power monitoring e 80 3 7 Software structure 2 a 81 3 7 1 Overall software structure 2 2 2 2 0 200202002 eae 81 ENEE 82 ee ee dia da 85 wa bb be he ebb ae ae heehs Gada
35. energy aware algorithm is one proposed by Delin e al in which nodes enter 40 Chapter 2 Background Energy Sensing and Wireless Communication a sleep state when their stored energy drops below a threshold value and wake up again when they have been recharged sufficiently by the solar cell Cianci et al have used threshold rules to balance the workload between a group of nodes in a network Other algorithms use distributed processes to permit redundant nodes i e those nodes whose communications or sensing coverage is duplicated by other nodes to sleep to conserve energy 86 Merrett et al have developed schemes to discard messages of low importance in the interests of ensuring that high importance messages can traverse the network Their schemes are collectively known as IDEALS RMR Information manageD Energy aware ALgorithm for Sensor networks with Rule Managed Reporting Central to the concept is the classification of the energy status of the node by its Energy Priority EP and the importance of the message by its Packet Priority PP The complete system diagram for IDEALS RMR is shown in Figure 87 A Rules Stats Update Fulfilled New Packet Environmental Data Fulfilled Rules Rules with PP acket gt Environmental Data Sensor Processing Energy Analysis A A A A i A Seen Data Penis Data Set Data Profile of Energy Source Residual Energy in Store li Sensor Sensor Sensor
36. from the PYE and interpreting this as adjusted energy or power levels For example for a supercapacitor energy store the PYE will measure a voltage of 2 0V and the EAN uses the E CV 2 equation to convert this to an energy value 2 Measurement requests The layer will process measurement requests from the EAN and obtain values through the PYE Values may be interpreted with the help of models as above where necessary 3 Command execution Executes commands related to the energy subsystem on behalf of the EAN Directs the flow of energy by initiating switching of transistors etc by means of the PYE layer In summary the EAN layer mainly deals with the coordination of measurement oper ations and the interpretation of values from the PYE by means of device models The aim is to provide a common interface for the ECO so that the stack structure remains consistent independent of the detail of the energy subsystem Physical Energy PYE Layer The main task of the physical energy layer is to provide an interface with the node s physical energy hardware This includes the following tasks 1 Configuring inputs and outputs To set up the interface between the node hardware and its energy subsystem Examples are configuring ADC parameters or setting up microcontroller pins as inputs outputs 2 Obtaining values from inputs Obtaining raw values for parameters on input pins Through interfacing with ADC hardware and obtainin
37. higher resolution on the battery state of charge against excessive energy wastage in testing Lithium thionyl chloride cells in common with many other lithium primary battery chemistries feature a stable operating voltage of around 3 6V until they become depleted I0 Hence state of charge determination is non trivial and systems must estimate the amount of energy used to identify their energy status End of life identification for LiSOCL batteries however is possible With a pulsed discharge similar to that used to identify state of charge in alkaline batteries end of life can be identified up to 15 80 Chapter 3 Development Towards a Reconfigurable Energy Subsystem before the cut off voltage Otherwise continuous loading of the cell will only permit passive identification of end of life up to 3 before cut off The response of the cell can be sensitive to temperature variation so care must be taken to ensure that any larger voltage drops with pulsed discharges are not due to seasonal variations 3 6 6 Power monitoring While most decisions about the activity level of the node will use information about the energy stored in batteries or supercapacitors it is also useful to have a knowledge of the power profile of the system For example it may be useful to be able to predict when energy is likely to be harvested as well as the instantaneous amount of power being supplied Indeed it may also be useful to monitor individual ene
38. importance particularly in low light situations Randall and Jacot have analysed whether the industry standard test known as AM1 5 to measure the performance of solar cells is of use to designers where the intensity and spectral content of light sources is likely to differ greatly from ideal test conditions The difference between the performance of PV cells under filtered AM1 5 light and under fluorescent lighting has been analysed and some technologies exhibit significant performance changes as shown in Figure 2 4 Photovoltaic cells can be treated as voltage limited current sources and are characterised by their open circuit voltage Voc and short circuit current Isc along with other pa rameters As incident light levels drop short circuit current will typically decrease while Chapter 2 Background Energy Sensing and Wireless Communication 19 Condition Fluorescent Light Illuminance lux Condition Iluminance lux 5 _ Design stand a Very fine weather 100 000 120 000 SCH illuminated 1 000 Fine Weather 50 000 100 000 Office Conference room 300 600 Cloudy 10 000 50 000 Restaurants coffee shops Below 200 Rain 5 000 20 000 TABLE 2 3 Typical indoor and outdoor brightness levels reproduced from 27 c Si mc Si CIS ox a Si REF nN o J 110 4 100 4 ER 80 70 60 50 40 4 30 cell efficiency relative to STC 2
39. in a Prototype System Hardware 111 w LN LA N Li N ke nn ken Switching Supercapacitor Voltage V Diode o U 0 F un T T ER T TT T SI T T T H 0 5 10 15 20 25 30 35 40 45 50 55 60 Time minutes FIGURE 4 14 Comparison of MPP switching converter circuit against conventional diode only circuit Test carried out with 0 5F supercapacitors from the same batch and identical 1116929 PV cells under office lighting at 1pm on an October afternoon approximate brightness 1 100 lux capacitor is charged up before allowing energy to be output to the multiplexer module The connections between this module and the multiplexer module are shown in Table 4 5 FIGURE 4 15 Circuit board for the vibration module The nominal output power from the vibration energy harvester is estimated by the microcontroller by shutting down the step down converter and isolating its associated input capacitance A known load is then switched in across the output of the vibration generator and the voltage across this load is measured to determine the instantaneous power level The EEDS stores the characteristics of the energy harvester and the details of the known load In this case the known load is in fact a 66kQ voltage divider which feeds into the input of a unity gain buffer The voltage divider and unity gain buffer supply are both switched by the digital measurement input line meaning tha
40. includes a CC2430 system on chip device which incorporates an en hanced 8051 microcontroller and 802 15 4 2 45 GHz transceiver The CC2430 is an 8 bit 121 122 Chapter 5 Case Study Deployment in a Prototype System Software device that has a maximum clock speed of 32MHz and the particular device in use has 128kB of flash memory and 8kB of RAM It operates from a supply voltage of 2 0 3 6V The raw device has 21 general purpose I O pins eight of which are connected to the internal ADC It also has an internal 1 25V reference generator and an on chip temperature sensor Conversely the eZ430 RF2500 has an MSP430F 2274 microcontroller and separate non 802 15 4 compliant 2 45GHz radio transceiver The MSP430F2274 is a 16 bit device with a maximum clock speed of 16MHz 32kB of flash memory and 1kB of RAM Due to the restricted memory resources the MSP430 was the more challenging platform for which this system was developed and for which the most development effort was expended It operates from a supply voltage of 1 8 3 6V The raw device has 32 general purpose I O pins 12 of which are connected to the internal ADC The microcontroller also has a 1 5V internal reference generator and an on chip temperature sensor 5 2 2 Language environment and debugger Support for the CC2430 is provided by the IAR Embedded Workbench tool chain whereas the MSP430 is supported by Code Composer Studio from Texas Instruments Both environments
41. indicate when a voltage threshold has been passed Chapter 4 Case Study Deployment in a Prototype System Hardware 105 4 4 4 Start up and voltage regulation The maximum unregulated voltage supported by the multiplexer module is 4 5V and this is regulated down to a maximum of 3 0V by a LDO linear regulator A linear regulator was chosen for this design due to its simplicity and low quiescent power con sumption It supports currents of up to 150mA Undervoltage protection is provided by a 2 0V Maxim XC61C voltage detector having a CMOS output stage which feeds into the enable input of the linear regulator A tantalum 1004F capacitor provides a small low ESR buffer which compensates for the current spike when the module output is turned on and the microcontroller starts up The voltage regulation functionality of this module was shown earlier in Figure The start up performance of the system has been verified with both the TI eZ430 RF2500 and TI CC2430 microcontroller modules 4 4 5 Data multiplexing The prototype multiplexer module supports the connection of up to six energy modules through its RJ45 sockets It includes five Analog Devices ADG708 8 to 1 analogue switches four are used for the module management lines and one is used for 1 wire communications to the EEDS on each module The analogue switches are used for data rather than power so their on resistance is actually of little importance They have an enable pin which is sim
42. it is undesirable to carry out unnecessary communications Further detail about the software and system management operations and implemen tation of electronic data sheet functionality may be found in the description of the prototype system software in Chapter 5 4 6 3 Overall evaluation The quiescent current draw of the multiplexer module has been found to be in the region of 7 2uA The quiescent power consumption of the primary battery module was 1 8uA and the secondary battery module was 5 14A Therefore in standard operation the minimum quiescent current consumption of the energy subsystem with these components energy harvesting modules tend to draw little quiescent power when not generating energy is approximately 14 14A At 3V this equates to a quiescent power of around 424 W This is significantly less than the 1004W 10mW that this system is designed to operate with and thus can be considered to be a reasonable level of power consumption It has been shown that the presence of diodes in the path of the power supply lines have a major impact on the efficiency of the system with low forward voltage Schottky barrier diodes dropping 260mV equating to a 9 efficiency loss at 3V and 1mA For sources generating energy at lower voltages the effective efficiency loss will clearly be much greater In this project diodes were considered a necessary evil and were used sparingly where there was a risk of substantial amounts of ener
43. measurements of parameters such as temperature or vibration and communicate over a wireless medium A key benefit is that they can operate autonomously Nodes are commonly battery powered so that they can be deployed rapidly without the need to install a wired power supply however batteries must be changed when depleted and this can impose a costly maintenance re quirement Energy harvesting is an emerging field which offers the possibility for nodes to be powered indefinitely from environmental energy such as light vibration or tem perature difference The power generated from environmental energy is often limited and variable and nodes must be able to adapt their operation to take account of the power available There have been a number of demonstrations of wireless sensor nodes powered from harvested energy but existing demonstrators are tailored for specific types of energy resource constraining their use to applications with suitable energy availabil ity The existing interfaces between the energy hardware and the nodes embedded software is bespoke and limited to specific devices so it is impossible to exchange the energy hardware to adapt to differing energy availability The work described in this thesis delivers a comprehensive scheme for reconfigurable energy aware sensor nodes which overcomes the limitations of the existing systems and allows the energy hardware for sensor nodes to be connected together in a plug and play
44. microcontroller to infer how to interface with the energy hardware and interpret the measurements obtained i e trans late voltage measurements into estimates of instantaneous power or stored energy For example for a rechargeable battery the EEDS will indicate that the charging and dis charging of the device can be switched on or off via the digital interface and that the measurement control line connects the battery across a known load defined in the data sheet so that its state of charge can be estimated The state of charge of the battery can be inferred from a piecewise linear representation of the discharge curve by using the measured voltage across a known load From this along with information on the capacity of the battery the usable stored energy can be estimated by the microcontroller The class codes for energy devices are shown in Table It may be observed that energy modules are classified as energy stores or sources Stores include batteries pri mary and secondary capacitive storage devices and any other device whose primary function is to store rather than generate energy Conversely energy sources include energy harvesting devices such as devices generating energy from light vibration or temperature difference along with other energy sources such as mains electricity or wireless power transmission technologies The 00 classifier is intended for use only by multiplexer modules and class 11 is reserved f
45. necessary to ensure that the efficiency of these systems remains high An important note is that while the demonstrated architecture facilitates the connection of up to six energy modules the scheme would allow for cut down systems to be developed which could accommodate as few as one or two energy modules for systems with a single energy module the multiplexer would simply act as a through connector with a voltage regulation capability the only data sheet that would be needed on this system would be that of the energy module so the microcontroller could interface with that directly For simplified devices with two or more energy modules a simplified energy multiplexer module with fewer ports would be used This would deliver savings in terms of cost and most probably reduce the physical size of the device The fact that the bulk of the code developed for the prototype system has been written in C means that it is transferable to a range of platforms commonly used in wireless sensor nodes Two separate platforms have been used in the development of this scheme the CC2430 based on an 8 bit 8051 processor and the MSP430 based on a 16 bit RISC processor The simplified interface with the application layer means that devices are able to self manage their resources without the need to understand the intricacies of their energy subsystems Of course as the proposed scheme has been successfully demonstrated on low power highly resourc
46. non rechargeable lithium manganese coin cells such as the CR2430 Secondary batteries Recent developments in low self discharge NiMH cells prompted the choice of Sanyo Eneloop cells for this project The AAA size cells have a nominal 1 2V operating voltage and 800mAh capacity They are accommodated by the module developed for primary batteries with the addition of a number of components to allow recharging of the cells Their low self discharge rate and ease of recharging makes them especially suitable for this application Due to the small amounts of power available and the complexities of charging it was decided not to use lithium based rechargeable chemistries Supercapacitors Two types of supercapacitor were used in this project accommodated by the super capacitor module Two radial Panasonic Gold supercapacitors with nominal 1F 2 3V capacity can be connected in series or one CAP XX supercapacitor nominal 0 22F or 0 55F 4 5V capacity can be used Provision is also made for balancing resistors to be fitted These capacitor types were chosen as they are commonly used in energy harvesting systems and are of low cost and moderate capacity 4 3 Components and energy management circuits 4 3 1 Low power system components In this case study a number of components are required to provide the energy man agement and measurement functionality for the system Clearly as the node is running from sub milliwatt power sources the q
47. of deployment or during its operation In the deployment the node is demonstrated with energy harvested from vibration light and thermal differences with the energy being buffered in supercapacitors and rechargeable batteries A non rechargeable battery provides a last ditch back up for emergency use The system is shown to operate from a range of devices which are swappable during operation and remains aware of its energy status throughout Chapter 1 Introduction 7 In line with the requirements detailed earlier this thesis describes the development of processes to deliver reconfigurable efficient energy aware sensor nodes The main contributions of the research can thus be summarised 1 Electronic data sheets and hardware interfaces in order to deliver a recon figurable plug and play energy subsystem for wireless sensor nodes it has been necessary to define an electronic data sheet format so that operating parameters can be stored on energy related devices This is a progression from the established transducer electronic data sheet concept A common hardware interface which standardises the interface between the sensor node and its sensor hardware has been developed to enable this data sheet to be read and allow the node s energy status to be monitored 2 Embedded software development an embedded software structure has been implemented which splits energy and sensor management tasks from the communi cati
48. of energy or isolate parts of the system that are not in use and therefore reduce the power wasted by the system They also enable parts of the system to be switched on to take measurements or permit charging discharging of energy stores e PMOS transistor International Rectifier IRLML6401 This is a power MOS FET with a minimum 0 4V gate threshold and an on resistance of 0 050 with a Vas of 2 5V and a low drain current e NMOS transistor International Rectifier IRLML2502 Another power MOS FET with a minimum gate threshold voltage of 0 6V and an on resistance of 0 040 with a Vas of 2 5V and a low drain current e JFET N transistor NXP PMBFJ309 This has a minimum cut off voltage of 1V and typical drain source on state resistance of 50Q With a 500 resistance a 1mA current will lead to an approximate voltage drop of 150mV e Schottky diode NXP BAT754 This is a small signal Shottky barrier diode that is available in a number of configurations With a 1mA forward current it has a voltage drop of 260mV 96 Chapter 4 Case Study Deployment in a Prototype System Hardware Voltage regulation These components act to provide a regulated voltage to the microcontroller in the case of the multiplexer module or to the raw voltage on the multiplexer module in the case of the other modules They deliver the facility for under and over voltage protection e Voltage detector Torex XC61CC This series of CMOS voltage dete
49. operation and calibration of the transducer For example with the systems shown in Figure 2 18 only four calibration points are given on the tag while the TEDS can store the parameters of the complete calibration curve This results in more accurate interpretation of transducer data The template shown in Table 2 7 is for a thermistor In this example the information is useful to the processor as it provides the information for the raw output from the transducer to be converted into an accurate temperature value It also gives information about the physical interface with the analogue transducer Supporting hardware TEDS data is typically stored in an electrically programmable read only memory EPROM and the digital interface with smart transducers is based on the Maxim Dallas 1 Wire protocol 58 In this master slave interface the master i e the data acquisition unit supplies power and communicates over a single wire using a defined sequence of time based commands 59 A number of 1 Wire devices can be attached to the same 1 Wire bus and controlled individually by a single master At the time of writing 1 Wire EPROM devices were available from Maxim Dallas in sizes between 1kbit and 64kbit and are the only EPROM device mentioned in the IEEE 1451 4 2004 specifications 34 Chapter 2 Background Energy Sensing and Wireless Communication Property Description Access Bits Units TEMPLATE Template ID 8 Ele
50. recharge process 12 With modern cell constructions memory effects are now negligible NiCd cells have historically been most susceptible to this effect but the Handbook of Batteries states that most users may never experience low performance due to memory effect the use of the term memory effect persists since it is often used to explain low battery capacity that is attributable to other problems 8 The number of cycles that the battery can go through is dependent on its chemistry and other factors but modern cells generally Chapter 2 Background Energy Sensing and Wireless Communication 15 400 300 Smaller size 200 Energy density W h 11 100 Lighter weight 0 50 100 150 200 250 Energy density W h kgr FIGURE 2 2 Comparison of rechargeable battery types in terms of volumetric and gravimetric energy densities reproduced from Tarascon and Armand 13 have a lifetime of at least a few hundred cycles Another parameter rarely quoted but important to energy harvesting nodes is the coulometric efficiency This indicates the proportion of energy stored against the energy expended in charging A figure of 99 9 is typical for lithium ion cells but is somewhat lower for other battery chemistries and highly dependent on charging conditions 8 Some typical parameters for the most commonly used secondary battery types are shown in Table
51. seconds and the microcontroller enters its low power sleep state A number of other definitions are also provided which define simple properties such as custom data types and macros for efficiently bit masking and processing data The hardware abstraction layer also hosts one wire interface functions which are described separately in Section 5 3 Effectively this layer permits the energy stack to be coded in a device agnostic fashion by removing the detail of interfacing with the physical device It allows the rest of the modules to be written in plain C which means that it can be compiled for any microcontroller type without the need for modification 126 Chapter 5 Case Study Deployment in a Prototype System Software 5 3 Energy electronic data sheets 5 3 1 Overview As discussed in Section 4 3 3 the EEDS hardware selected for this system is the DS2502 1kB add only EPROM The device must be programmed before deployment in the sys tem as it requires programming voltages of around 12 0V however the facility is provided for the 1 Wire memory to be programmed with the 1 Wire device soldered in situ into the energy module A cable was fabricated to interface between the EEDS pin of the RJ45 socket of energy modules through to the RJ11 jack required by the programming device The memories were programmed using a DS9097E COM port adapter from Maxim Dallas via their iButton Viewer PC application the programming device has an ex
52. should be noted that in this scheme the actual har vesting device and its power conditioning circuit are considered as one module FIGURE 4 10 Circuit board for the photovoltaic module The circuit interfaces with indoor amorphous silicon PV cells with an operating voltage of between 2 5V and 5 0V In the demonstrator system it is connected to a Schott OEM 1116929 This cell is capable of producing gt 1mW of power under bright indoor lighting The circuit works to hold the cell at its optimal voltage regardless of the unregulated voltage on the multiplexer module This improves the efficiency of the system and delivers faster charging times when cold starting the system It should be noted that the operating voltage of the PV cell is set using a variable resistor on the PCB Overvoltage protection is delivered using a XC61C voltage detector and FET combination which disconnects the photovoltaic cell when the maximum system voltage is reached The same FET combination feeding into an op amp buffer arrange ment is used to permit the microcontroller to analyse the open circuit voltage of the cell and hence estimate the nominal power being harvested Pin Type Function 2 Meas Control Disconnect cell to perform measurement 3 Measurement Analogue open circuit voltage of cell 4 Control None 5 Control None TABLE 4 3 Interface pins from photovoltaic module Chapter 4 Case Study Deployment in a Prototype Syst
53. software and algorithms The proposed scheme has been evaluated with a case study system Chapter 4 described the development of the system hardware including a range of energy modules support ing batteries supercapacitors mains power and a number of energy harvesting sources and an energy multiplexer module The design of each module was explained and its energy management features and the impacts of the energy awareness and management circuitry were described The software for the prototype system was described and evaluated in Chapter 5 This included the contents and structure of the implemented electronic data sheets the actual energy stack developed in the project and the interface between the energy stack and the application layer 6 2 Interpretation of the results The scheme described in this thesis delivers a flexible architecture for wireless sensor nodes through the implementation of a reconfigurable energy subsystem The system incorporates an energy stack with a simplified interface to the application running on the microcontroller The developed scheme extends the state of the art of energy aware sensor nodes by addressing the challenge of allowing a variety of energy hardware to be connected to a sensor node in a plug and play manner and the software interface allows the microcontroller to interface with this hardware in order that it can be aware of its energy resources and act to manage them where appropriate
54. the 1 Wire interface The embedded software structure was also described Detail was given on the implementation of the functions in the energy stack including detailed consideration of the implications of certain mathematical functions High level considerations for energy aware operation were also discussed and the overall evaluation of the system was also considered The embedded software structure that has been implemented for this device provides a robust and flexible system for monitoring and managing the energy hardware on the sensor node While some of the mathematical functions carried out are demanding for a resource constrained microcontroller the com plete energy assessment process on the MSP430F 2274 platform which lacks a hardware multiplier has been demonstrated to complete within a few tens of milliseconds Given that updates of the energy status on the node in the prototype application happen at most once per second falling to once per minute when energy is low and the update frequency can easily be manipulated in software this represents a relatively small power cost in view of the substantial additional features it brings Chapter 6 Conclusions and Future Work 6 1 Summary of work Wireless sensor nodes offer the facility to remotely monitor parameters such as tem perature pressure and vibration in machines buildings or the environment Wireless sensor nodes must be capable of operating autonomously and for ma
55. the system will be individually evaluated and the overall performance of the system is summarised in Chapter 6 3 10 Summary This chapter defined the interfaces and design for the reconfigurable system architecture defining many of the key contributions of this work The overall structure of the system was described along with a discussion of how the individual aspects fit together to deliver a complete system architecture The new aspects of the system architecture comprise energy electronic data sheets and a common hardware interface between the energy modules which were both defined A generalised specification for the hardware of each energy module including the multiplexer module and energy modules was provided and methods and algorithms for determination of the energy status of the node were also presented The novel software structure which has been utilised in this project was also discussed The general operation of the system was covered and the requirements of a prototype to verify this approach were also discussed The prototype system or case study which has been used to verify this approach and delivers a reconfigurable system is now described in chapters 4 and 5 Chapter 4 Case Study Deployment in a Prototype System Hardware 4 1 Introduction The architecture proposed in Chapter 3 is evaluated by way of a case study which is outlined in this chapter along with the hardware design of the reconfig
56. with a low duty cycle to conserve energy consuming tiny levels of power while asleep and utilise low power communications which minimises their active energy consumption This means that their average power requirement is very low and allows them to deliver operational lifetimes that are typically measured in months or years A Glacsweb base sta B Active volcano moni tion toring FIGURE 1 2 Two examples of environmental sensor network deployments a a Glac sweb base station is used to communicate with nodes buried deep in a Norwegian glacier reproduced from 2 and b a node performing seismic and infrasonic measurements on the side of an active volcano in Ecuador reproduced from 3 Examples of real deployments of wireless sensor systems range from the relatively mun dane to the extreme Machinery condition monitoring is a popular application for this technology For example BP have utilised wireless sensor nodes to monitor the condi tion of machinery in one of their oil tankers 4 More challenging applications include embedding wireless sensor nodes deep inside glaciers to monitor their dynamics 2 or on Chapter 1 Introduction 3 volcanoes to monitor seismic events as shown in Figure 1 2 Essentially these are examples of applications where conventional sensing would be expensive or impractical and where the lower complexity and deployment cost of wireless autonomous sensor sys tems enables parameters
57. 0 4 x GaAs REF 10 4 0 light intensity W m2 1000 FIGURE 2 3 Efficiencies of a range of PV cell technologies at different light intensities relative to standard test conditions STC reproduced from Reich et al 25 8 increase e gt Ze eo Y xe Xx 6 4 D a 32 e s 27 gt 7 o o e E o 2 o o Et Z v A o o HZ o 2 E ee a s wini oc o x D 04 T T 0 2 4 Intensity W m2 S i o gt vo gt o bag P eg iv Ki T T 0 2 4 Intensity W m2 Relative efficiency decrease 2 4 Intensity W m2 aSi Tess Filtered AM1 5 3 aSi Sino Filtered AM1 5 s aSi Tess Fluorescent aSi Sino Fluorescent ck PC ICP2 Filtered AM1 5 e PC GCSA Filtered AM1 5 PC ICP2 Fluorescent 4 PC GCSA Fluorescent a xSi EdSi Filtered AM1 5 xSi BP Filtered AM1 5 x xSi EdSi Fluorescent x xSi BP Fluorescent A Amorphous Si B Dye Cell c Crystalline Si FIGURE 2 4 Comparative efficiency of PV cells under filtered AM1 5 and fluorescent spectrum light for two samples of three PV cell technologies reproduced from Randall et al 26 20 Chapter 2 Background Energy Sensing and Wireless Communication open circuit voltage will remain fairly constant Figure 2 5 shows a photovoltaic module from Schott Solar along with its performance
58. 17 pose gis eee os geo Ee ae Ses ee ee Gee Ba a e 117 baat BAe fo Bebe Bs Ho ce we oh A A Se Soe a 118 gc ee ee Gea we BME et wee awe Be Se 119 A SUMMARY as ea EE a Ro ee EP BO RO Er ds Be Swe BO A 120 121 Ss eae ga anus Meh Ga ee EEN 121 EE 121 gh Be Se ide Gee EE Gs 121 d ea RE d er 122 Spleen oe ss 123 ENEE 124 eee Se doses Goes dea age wt ae eee Gear a 126 D OVERVICW poe el A ea ee Se oe toe Meee be e 126 5 3 2 Functions for 1 Wire communications 0 126 5 3 3 Performance costs of 1 Wire communications 127 5 3 4 Data sheet formal lt lt lt lt 129 E are eee 131 5 4 Embedded software structure 133 5 4 1 Overall structure 133 5 4 2 Communication stack so sasare ra 133 5 4 3 Sensing stach 135 5 4 4 Energy stach 136 by bts thoes the ari ee is e 136 5 5 neren stack ls 6 base bog Wa ded Skee SB a eee ade de ae d 136 5 5 1 Physical Energy PYE layer 136 5 5 2 Energy Analysis EAN layer 139 5 5 3 Energy Control ECO layer o o 143 a A Pr dena vi a e 145 5 6 1 Implementing energy adaptive behaviour 145 ENEE 146 5 6 3 Embedded software features 2 2 a a a a a ee 146 Hee ieee os 146 O a e de Gem a Bede eh Go 147 9 89 SUMMA A 149 6 Conclusions and Future Work 151 6 1 Summary of work 151 6 2 Interpretation of the results e 152 CONTENTS ix 6 3 Recommendations for future w
59. 4 define MEAS CUBIC 0x05 define MEAS CURVE 0x06 define MEAS COMPLEX 0x07 LISTING 5 2 Measurement types for devices TModule struct which follows the format defined in Table 3 1 In the system implemented under this project a number of device types were defined in order to enable an energy aware node to be delivered in as efficient a manner as possible The parameters of the devices are stored in the Parameters field the structs that are defined under the TParams field of the TModule struct are shown in Listing 5 5 The TDevicelD field is simply the six byte serial number of the 1 Wire memory and the DeviceType field refers to the device type as defined in Table The Measurement field refers to the measurement scheme for the device i e how the microcontroller can translate the sensed value into an estimate of power or energy which is defined in Listing 5 2 The Multiplier field indicates how to scale the measured value before entering this conversion scaled up by 10 to be stored as an unsigned char and the MaxOutput field defines the maximum output voltage also multiplied by 10 The multiplexer module EEDS stores information specific to the role of the multiplexer module The TMuxData struct that defines the format of this data is shown in List ing 5 3 The TDevicelD field stores the six byte serial number of the 1 Wire memory and the Multiplier field indicates how to scale the measured value of the raw volt age mult
60. A Delin and S P Jackson The sensor web a new instrument concept Pro ceedings of the SPIE The International Society for Optical Engineering 4284 1 9 2001 C M Cianci V Trifa and A Martinoli Threshold based algorithms for power aware load balancing in sensor networks Swarm Intelligence Symposium 2005 SIS 2005 Proceedings 2005 IEEE pages 349 356 June 2005 V Raghunathan C Schurgers S Park and M B Srivastava Energy aware wireless microsensor networks EEE Signal Processing Magazine 19 2 40 50 2002 G V Merrett N R Harris B M Al Hashimi and N M White Energy man aged reporting for wireless sensor networks Sensors and Actuators A Physical 142 1 379 389 March 2008 Energizer Battery Manufacturing Inc Alkaline Manganese Dioxide Hand book and Application Manual http data energizer com PDFs alkaline_ appman pdf 2006 Last accessed May 2010 T Umemura Y Mizutani T Okamoto T Taguchi K Nakajima and K Tanaka Life expectancy and degradation behavior of electric double layer capacitor Part I volume 3 pages 944 948 Nagoya Japan 2003 P Forstner SLAA334A MSP430 Flash Memory Characteristics http focus ti com cn cn general docs lit getliterature tsp literatureNumber slaa334akfileType pdf April 2008 Last accessed May 2010 Texas Instruments Inc MSP430 16 bit Ultra Low Power MCUs ti com msp430 2010 Last accessed May 2010 Microchip Technology Inc eXtre
61. CB Printed Circuit Board PMBus Power Management Bus PP Packet Priority PRIMP Priority Based Multi Path Routing Protocol PV Photovoltaic PYE Physical Energy Layer PYS Physical Sensing Layer QoS Quality of Service ABBREVIATIONS xxvii RAM Random Access Memory RF Radio Frequency RF ID Radio Frequency Identification RMR Rule Managed Reporting ROM Read Only Memory SBD Smart Battery Data SEV Sensor Evaluation Layer SMBus System Management Bus SoC System on Chip SPR Sensor Processing Layer STC Standard Test Conditions TEDS Transducer Electronic Data Sheet TI Texas Instruments UHF Ultra High Frequency WISP Wireless Identification and Sensing Platform WLAN Wireless LAN WPAN Wireless Personal Area Network Chapter 1 Introduction 1 1 Wireless autonomous sensing The powering of remote and wireless sensors is widely cited as a critical barrier limiting the uptake of this technology Energy Harvesting Technologies to Enable Remote and Wireless Sensing Sensors and Instrumentation KTN Technical Report I Remote and wireless sensors referred to hereafter as wireless autonomous sensors or wireless sensor nodes offer the facility to remotely monitor parameters such as tem perature pressure and vibration in machines buildings or the environment These nodes will typically feature at least one type of sensor a microcontroller to manage and process data a wireless communication interface wh
62. Energy Source s Energy Store s FIGURE 2 21 The IDEALS RMR system diagram introduced by Merrett et al re produced from 87 showing how packet priorities and energy priorities are generated In this scheme messages containing more information are given a lower Packet Priority PP number i e the most important messages are allocated PP 1 and for the lowest priority messages PP 5 The energy status of the nodes is classified with a high Energy Priority EP number for higher energy availability For the case where the energy store is empty i e the node cannot sustain even a single transmission the node is allocated EP 0 Merrett et al also introduce the concept of priority balancing in which message prior ities and power priorities are balanced For example a node with plenty of energy will transmit messages of all information levels Conversely a node with constrained energy reserves will only transmit messages of very high importance i e low PP values Thus in order to maintain the operation of the node and in turn the network sensor nodes only transmit messages when this is sustainable For example when a node has EP 4 it will only transmit messages with PP lt 4 The process of priority balancing is Chapter 2 Background Energy Sensing and Wireless Communication Al Power Priority Message Priority PP MP E 5 Low Information Residual Energy
63. IGURE 2 18 Conventional and plug and play smart sensors Device a has config uration data on an attached label that must be manually entered into the interface system while b stores this data on an EPROM memory integrated into the plug which can be automatically read Images reproduced from Watlow data sheets a Chapter 2 Background Energy Sensing and Wireless Communication 33 Description Bit Length Allowable Range Manufacturer ID 14 17 16381 Model Number 15 0 32767 Version Letter 5 A Z data type Chr5 Version Number 6 0 63 Serial Number 24 0 16777215 TABLE 2 6 Basic TEDS content as defined in IEEE 1451 4 2004 reproduced from 2 4 2 Transducer electronic data sheets Transducer electronic data sheet format TEDS information is stored in electronic memories on transducers A range of data about the transducer can be stored and depends on the type of transducer the template it corresponds to and the decision of the manufacturer to provide any further information Basic TEDS information must be stored which identifies the manufacturer by means of a number that is allocated by the IEEE Registration Authority and stores the model and version numbers of the device along with its serial number This information is sufficient to uniquely identify the transducer The basic TEDS format is shown in Table Further to this basic information the TEDS contains a wider variety of detailed data about the
64. Identifiers for modules in the energy subsystem The multiplexer module has addresses 0 and 7 energy modules are accessed through addresses 1 to 6 By first querying address 7 the multiplexer module the EEDS of this module can be read in order to determine its operating parameters These parameters include most 68 Chapter 3 Development Towards a Reconfigurable Energy Subsystem importantly the number of sockets it has up to six its minimum cut off voltage max imum supply rail voltage and regulated voltage In this way through this common interface the microcontroller can self determine the energy subsystem s parameters by sequentially querying the EEDS on all devices attached through the multiplexer mod ule It will also understand the function of each control and measurement line for the individual modules It is the task of the multiplexer module to ensure that the supply voltage and ADC inputs that the microcontroller receives are not damaging to the device nominally by means of diode clamping to the supply output rails The modules are also required to ensure that the measurement outputs they give are within the valid range of OV to 1V which is compatible with the ADCs on low voltage microcontrollers used in wireless sensor nodes as introduced in Section 2 7 1 It is not necessary to mandate specific sockets on the multiplexer module for specific energy module types as all switching hardw
65. Increased temperatures affect other parts of the system such as the energy store and harvesting device Designers must consider the effects of the operating environment on all aspects of the system in order to ensure that the required operating lifetime is achievable 2 6 5 Discussion Energy aware operation can extend the effective lifetime of sensor nodes and allow them to optimise their operation to adapt to their energy status This section has introduced the existing schemes for energy aware routing energy adaptive behaviour and achieving energy awareness along with a consideration of the designed lifetime of components of a sensor system The means of delivering energy awareness and energy adaptive behaviour is non trivial and at present is not a well defined process Existing systems must be highly tailored to the energy hardware deployed and there is no mechanism for the automatic configuration of energy aware systems or of representing the method of monitoring the energy status of the device It is the view of the author of this thesis that due to the increasing complexity of the energy hardware of wireless sensor nodes incorporating a range of energy devices such a scheme is required 2 7 Wireless sensor node technologies 2 7 1 Microcontrollers transceivers and system on chip A number of low power microcontrollers have been developed for use in resource con strained wireless sensing applications Examples include the TI M
66. J C S Reis Sun wind and water flow as energy supply for small stationary data acquisition platforms Computers and Electronics in Agriculture 64 2 120 132 2008 BIBLIOGRAPHY 195 46 C Park and P H Chou Ambimax autonomous energy harvesting platform 47 48 49 50 51 52 53 54 58 for multi supply wireless sensor nodes 2006 3rd Annual IEEE Communications Society Conference on Sensor and Ad Hoc Communications and Networks pages 168 77 2006 J Kymissis C Kendall J Paradiso and N Gershenfeld Parasitic power har vesting in shoes Digest of Papers Second International Symposium on Wearable Computers pages 132 9 1998 T Starner and J A Paradiso Low Power Electronics Design chapter 35 pages 1 30 CRC Press 2004 B Jiang J R Smith M Philipose S Roy K Sundara Rajan and A V Mami shev Energy scavenging for inductively coupled passive RFID systems In IMTC 2005 Instrumentation and Measurement Technology Conference Proceedings of Ottawa Canada May 2005 J A Paradiso and T Starner Energy scavenging for mobile and wireless elec tronics IEEE Pervasive Computing 4 1 18 27 2005 Powercast Corporation Powerharvester Receivers http www powercastco com products powerharvester receivers 2010 Last accessed May 2010 A Sample and J R Smith Experimental results with two wireless power transfer systems White paper Intel Research
67. Mbit s with a range of up to 140 metres Wireless LAN has a much higher power requirement and higher data rate than wireless personal area network WPAN radio protocols meaning it can place costly demands on battery powered devices Effectively WLAN protocols are capable of implementing star or tree topologies The current version of the Bluetooth protocol version 2 1 supports data rates up to 3Mbit s at a range of up to 10 metres Bluetooth Low Energy technology Chapter 2 Background Energy Sensing and Wireless Communication 37 Application Framework ZigBee Device Object ZDO Endpoint 240 i Endpoint 0 APSDE SAP APSDE SAP Interfaces Ze en Application Support Sublayer APS da E APS Security APS Message Reflector rae a Broker Management lt i Security Management T g S Service Snes D 3 NLDE SAP D gt Provider Sey Network NWK Layer SSS ee S E gt 5 Security Message Routing Network fo Management Broker Management Management 3 E eer MLME SAP Medium Access Control MAC Layer ri PD SAP PLME SAP S SES Physical PHY Layer 2 4 GHz Radio 868 915 MHz Radio FIGURE 2 20 Outline of the ZigBee Stack Architecture reproduced from 66 is set to be included in future Bluetooth standards and has the aim of allowing ultra low power e
68. Mechanical Engineering The University of California Berkeley 2003 S Roundy P K Wright and J M Rabaey Energy Scavenging for Wireless Sensor Networks with Special Focus on Vibrations Kluwer Academic Boston 2004 S P Beeby M J Tudor and N M White Energy harvesting vibration sources for microsystems applications Measurement Science and Technology 17 12 R175 R195 2006 S Roundy and Y Zhang Toward self tuning adaptive vibration based microgen erators volume 5649 pages 373 384 SPIE 2005 R Amirtharajah and A P Chandrakasan Self powered signal processing us ing vibration based power generation JEEE Journal of Solid State Circuits 33 5 687 95 1998 194 BIBLIOGRAPHY 34 Perpetuum Ltd PMG17 Vibration Energy Harvesters http www perpetuum 35 36 37 38 39 40 41 42 43 44 45 com pmg17 asp 2010 Last accessed May 2010 Ferro Solutions Inc VEH 460 Electromechanical Vibration Energy Harvester http www ferrosi com files VEH460_May09 pdf May 2009 Last accessed May 2010 R N Torah P Glynne Jones M J Tudor and S P Beeby Energy aware wireless microsystem powered by vibration energy harvesting Power MEMS 2007 Submitted to the 7th Int Workshop on Micro and Nanotechnology for Power Generation and Energy Conversion Applications 2007 S Roundy and P K Wright A piezoelectric vibration based generator for wireless electron
69. Prototype System Hardware 4 4 Multiplexer module 4 4 1 Functional overview The purpose of the multiplexer module is to deliver a reconfigurable energy subsystem by allowing the microcontroller to interface with the energy modules and facilitating the flow of electrical energy between the modules The basic hardware requirements of this module were listed in Section 3 5 1 The prototype multiplexer module is shown in Figure 4 7 and its schematic can be found in Appendix A Figure A 1 Push Buttons SMA a ele e y Cd YO Et Detection Li e z Buffer rer socket 6 o Reco Reg Campe Y Cap H WW et H EPROM glade D SE 74 TOR Gee AAA A OD a OPinddeader Sat L auf A FIGURE 4 7 Prototype energy multiplexer module The connections between the multiplexer module and the other energy modules are through RJ45 sockets compliant with the scheme defined in Table The interface between the multiplexer module and the microcontroller is through a 2 row 10 way header compliant with the scheme outlined in Table The multiplexer module ap pears as two separate addresses address 0 is the low power passive monitoring circuit without a 1 wire EPROM and address 7 is the device that allows the raw voltage on the multiplexer module to be measured and the device to be interrogated The effective connections from each of these addresses are shown in tables and respectively A simplified schematic which shows the dat
70. R3 J1 E ge Ne a A Mi CIL 2R5 L1 SEN F Ic2 ES R2 R4 7 D1 A Vetri Vineas Microcontroller Transceiver ICH FIGURE 4 11 The developed PV energy harvesting circuit which can be interrogated by a microcontroller posed by Dondi et al 120 are unsuitable for use with small cells in low light conditions due to their reliance on the J parameter which must be extremely small and hence falls outside the range of parameters that can be simulated by SPICE The actual model used in this investigation was adapted from that proposed by Hageman 125 and is shown in Figure AMA Va Rem e Rym bo Voc o V FIGURE 4 12 SPICE model for small PV cells Power estimation technique The data sheet parameters for this PV cell have been plotted and a logarithmic trend line has been added as shown in Figure In this instance the open circuit voltage is approximated to the light level by Equation 4 5 Voc 0 2571 In Ey 2 9128 4 5 Where Ey is the illuminance level in lux and can be generalised as Equation 4 6 where A and B are parameters to be found for the individual PV module Voc Aln Ey P 4 6 Chapter 4 Case Study Deployment in a Prototype System Hardware 109 4 8 4 7 4 6 4 5 y 0 25711n x 2 9128 4 4 4 3 4 2 Datasheet Voc Open circuit Volta
71. SP430 91 a 16 bit RISC mixed signal processor which is used in Crossbow motes described later in this chapter and has achieved dominance for use in wireless sensor network applications Generally microcontrollers used in wireless sensor networks will feature very low sleep currents of around 1A fast wake up times and the ability to interface with a range of peripherals including radio transceivers and analogue sensors Other options for low power microcontrollers include the 8 bit RISC Microchip XLP PIC range 92 and Atmel AVR microcontrollers 93 Some higher power node platforms are based on larger pro cessors for example the Imote2 platform 94 is based on the Marvell PXA271 XScale processor The sleep current for these devices is typically around 100 times that of the low power microcontrollers although their active mode power consumption is compara ble which makes these devices more suited to less energy constrained applications and where higher processing power is required Chapter 2 Background Energy Sensing and Wireless Communication 45 In wireless sensor nodes an RF transceiver generally provides the interface between the microcontroller and the physical communications channel The most widely adopted protocol for wireless sensor networks is IEEE 802 15 4 2003 and a typical transceiver is the TI CC2420 as evidenced by its use in Telos motes 5 Many of the requirements of the protocol are implemented in har
72. This thesis will consider only low power sensor devices which typically consume on average less than one milliwatt of power in general operation which means that powering these devices from relatively low levels of harvested energy can be feasible 1 2 Outline of the AEASN project The research presented in this thesis has in part been undertaken under the Data Infor mation Fusion Defence Technology Centre DIF DTC Phase II Adaptive Energy Aware 4 Chapter 1 Introduction Sensor Networks AEASN project The AEASN project was funded jointly by the UK Ministry of Defence and General Dynamics UK The project had two main aims 1 To allow wireless sensor nodes to efficiently manage their own energy resources using energy harvested from their environment where appropriate This should permit nodes to be energy aware and facilitate sustained operation 2 To permit groups of nodes to negotiate and co ordinate their sensing activities taking account of their energy status to fulfil the requirements of the network The work presented in this thesis focuses on the first of these aims delivering energy aware nodes that are able to manage their energy resources efficiently and accept energy harvested from the environment The research carried out by the author has fed into this project and three of his conference publications have also been presented as deliverables of the AEASN project 1 3 Justification for this researc
73. a connections on the multiplexer module is shown in Figure 4 8 and a simplified schematic showing the power connections between the modules is shown in Figure 4 9 Chapter 4 Case Study Deployment in a Prototype System Hardware 103 Raw V m Meas Mux Meas Control Mux Measurement Voltage Mux EEDS Detector So PX So So PX Si e S ad S HF S Energy Sob Energy St Energy Sr Modules Sa Modules St Modules S4 1to6 S 1 to 6 S4 1 to 6 Ber Pin 2 S5 H Pin 3 Ss F Pin 1 S m Se Fp Se e S7 S S7 uC Pin5 D uC Pin6 4 D uC Pin4 D Ta wire Ao Ar Ae EN Ay Ay Ap EN Ao A Ac EN EPROM fa D Mux Control 1 Mux Control 2 Lu uC Pin 1 So PX So X A S L S L HD T yC Pin 2 So Energy So Energy M uC Pin 3 S H Modules S3 Modules Sa 1 to 6 Sa 1 to 6 Ss p Pin4 Ss E Pin5 Se H Se F Voltage j Sr S7 Y Detector uC Pin7 D uC Pin8 D Ao Ai Az EN Ao Ai A2 EN FIGURE 4 8 Simplified schematic of multiplexer module showing data connections Pin numbers refer to Common Hardware Interface pin numbers L VE y Energy reg Modules Energy 1 to 6 Voltage im Vol
74. aces 86 30 EE cia Shae dk eee A Hl wee RR oe doe a a eg ee 86 fie Berets ke E en Wl Ome te A e 86 3 8 3 Monitoring and active management 87 3 8 4 _Network level interactions o a ooa a a 0000 ae 87 3 9 Towards a prototype to verify the approach 88 3 9 1 Overview EI 88 Wa Sup boo a Gee a Ba ea ee dee dh dee ee eG 88 3 10 Dummar A Grala Goak AORO Ek a E o A G a 89 91 pag a a VW A ee a A ee ate ieee e S 91 4 2 Overview of the case study 2 02 ee ee 91 4 2 1 Scenario e 6 eee ee ee ee ya 91 BREET 92 ee eat oa ee a a e E 94 EE 94 4 3 1 Low power system components 94 4 3 2 Conmectors gt pakn s dee aaa dada 96 o 96 4 3 4 State retention for device controll 97 APT 98 ke ie Bh Gl e Gt we tt Vi bo 99 eh GR a Ska ka eb eo bod e eS ea ek ae 102 4 4 1 Functional overview o 102 4 4 2 Energy awareness circuitry ee ee 104 4 4 3 Additional features 104 gh ae a eae Ge ak Aes e 105 4 4 5 Data multiplexing 2 so ccs see ee Pe ee 105 4 4 6 Overall efficiency 0 0 0 00002 eee 105 4 5 Energy modules 106 viii CONTENTS 4 5 1 Photovoltaic module a a a eee eee 106 St Gere nace Ek aes 110 Sys eae E ag ee eae eae 112 4 5 4 Mams module 113 ae Gye ce a oer eS a eee eee Gees 114 eae dls eee a FO ee ee 115 A ae Le Goes buna be 117 be ah fiat ee E aes Gene eae Gee Sees 1
75. achieved dominance 1 In short term deployments or where sensor nodes are accessible and batteries can be changed without causing major disruption alkaline typically alkaline man ganese dioxide or similar cells are most frequently used This chemistry has a low self discharge rate and hence a shelf life of several years This chemistry is very popular for use in consumer electronics devices which means that its cost is relatively low 2 For long term deployments where nodes may be inaccessible and battery replace ment impractical or expensive long life lithium primary cells are generally used typically lithium thionyl chloride LiSOCl2 These have a very low self discharge rate and a higher specific energy and energy density than alkaline chemistries These cells can have an operational life in excess of ten years but are relatively expensive Detailed parameters for these battery types along with information on other popular chemistries are shown in Table 2 1 The nature of wireless sensor network deployments means that long lifetimes are desirable otherwise the costs of regularly changing bat teries could outweigh the savings made through not having to install cabling to sensors The bursty characteristic of current draw by wireless sensor nodes when sensing and Chapter 2 Background Energy Sensing and Wireless Communication 13 communicating must be supported by the battery without resulting in excessive volt a
76. ad of the multiplexer module and all other module data sheets the operations will take up to 104ms Assuming that the 1 Wire bus is being pulled low by either device for 50 of this time the energy disippated through the pull up resistor can be calculated as 94uJ over this time As an example the waveform of the system reading the electronic data sheet on the supercapacitor module is shown in Figure 5 4 STOP 500 Gus une Ca 2 74V y 8 Wid 166us Wid 18Gus BEE Gout 1gl FIGURE 5 4 Oscilloscope trace of the microcontroller reading the 1 Wire EEDS on the supercapacitor module 5 3 4 Data sheet format The content of each module data sheet is dependent on the module s functionality Structs that define the data stored for each type of device are defined in the etypes h header file which is accessible by all layers of the energy stack Parameters for energy modules other than the multiplexer module are stored as shown in Listing 5 1 in the 130 Chapter 5 Case Study Deployment in a Prototype System Software typedef struct _TModule TDeviceID DevicelD unsigned char DeviceType unsigned char Measurement unsigned char Multiplier unsigned char MaxOutput TParams Parameters TModule LISTING 5 1 Struct for energy module EEDS parameters define MEAS UNMEASURABLE 0x00 define MEAS ONOFFMAXIMUM 0x01 define MEAS ENDOFLIFE 0x02 define MEAS_LINEAR 0x03 define MEAS SQUARE 0x0
77. adient at higher voltages is small meaning that the efficiency of both circuits is similar in this situation Therefore it may be inferred that there is no significant performance penalty at higher voltages for utilising this MPP switching converter circuit The system has been tested and evaluated in an office environment with a mix of natural and fluorescent lighting From cold starting the system steadily charged the supercapacitor energy store C2 until the system voltage reached approximately 2 1V at which point the microcontroller became active and started transmitting with a 30s sleep period At its absolute peak level of harvesting placed in close proximity to a fluorescent lamp the system reported a nominal power level of 3 5mW but a more typical level was around 1mW 4 5 2 Vibration energy harvesting module Functional description The vibration module PCB from the schematic in Appendix A Figure A 3 and shown in Figure is designed to interface with a Perpetuum PMG17 vibration energy harvester The manufacturers have found in real deployments that this device can generate 5004 W on 80 of machines and 1 0mW on 60 of machines 14 It provides a half wave rectified output limited to 8V The circuitry on this PCB uses a high efficiency step down converter to limit this voltage to 4 5V To increase the efficiency of the system the device waits by means of a voltage detector IC until the output Chapter 4 Case Study Deployment
78. aic module is shown in Table 5 2 The device type identifies the module as a photovoltaic module The measurement scheme is defined as complex and the method for interpretation of the values is shown in Section 5 5 2 Again the multiplier indicates that the measured value must be scaled by 4 9 to arrive at the actual voltage across the photovoltaic module The maximum output of the module is stated as 6 0V which is the open circuit voltage of the cell under peak illumination conditions Four additional parameters are given A B C and D three of which are floats The parameters are related to those shown earlier in Tablet A Chapter 5 Case Study Deployment in a Prototype System Software 133 Data sheet parameters A and B are simply the values from Table 4 4 scaled up by 100 parameter C is the product of k x m and parameter D is the a 1 scaled up by 10 as given in Section 4 5 1 The parameters were adjusted in this way to maintain the precision and efficiency of the calculation while minimising the storage size of the electronic data sheet parameters for this module Photovoltaic Module Device Type 10 0010 00 0x46 Measurement Complex 0x07 Multiplier 4 9 49 0x31 Max Output 6 0 60 0x36 A 25 71 0x41 CD AB 14 B 291 28 0x43 91 A3 D7 C 0 003813 0x3B 79 E3 86 D 7 8 78 0x4E TABLE 5 2 Electronic Data Sheet for photovoltaic module A selection of data sheets for oth
79. al S 3 0 76 0mm 3 0 7 O Net Output Wattage 0 20 E 2 0 6 S 505 0 15 3 50 12 5mm 2 iS Q Los 0 76 0mm gt 3 Se 0 10 H z 03 05 2 02 _ a 6 0 152 0m1 A 0 05 0 1 0 0 3 5 7 9 11 13 J Output Voltage B Dimen C Performance sions FIGURE 2 13 The Tellurex PG1 Power Generation Kit which generates electrical energy from an open flame and its electrical performance reproduced from J 2 3 5 Small scale fluid flow The generation of power from wind or water flow is popular and successful on a large scale This experience has been transferred to the generation of energy from small scale fluid flow to provide energy to sensor nodes Many sensors are located close to moving fluids for example liquids in pipes or air in ducts The flow of fluid such as air or water past a sensor has the potential to provide a source of power although it would be reasonable to assume that efficiency levels would be significantly lower for smaller devices In their review of potential energy sources for wireless sensor nodes Roundy et al state that power densities from air flow are promising and warrant further research TS The potential power P from a moving fluid is given by Equation 2 6 Here p is the density of the fluid A is the cross sectional area and y is the velocity of the fluid 1 P PAY 2 6 A device harvesting energy from the flow of water in a pipe shown in Figure has been used by Morais et al 45 Th
80. alculated is just a representation of how charged a node is while being adjusted for the method of discharge of the energy store be it resistive power or current A more thorough expla nation of the categorisation of discharge methods is given in Section 3 6 3 and methods for calculating the state of charge and remaining lifetime fraction for supercapacitors and batteries are given in sections and Chapter 3 Development Towards a Reconfigurable Energy Subsystem 75 Voltage V 1 7 T T T T 0 5 10 15 20 25 30 Time s VR VI en VP FIGURE 3 5 Simulated discharge of a 1F supercapacitor through a 4509 resistor 70mA current load and 200mW power load Shows divergent behaviour with time The concept of energy priorities was introduced in Section 2 6 2 The system of energy priority levels used in this project is shown in Table Through this method the activity of the system is controlled by its energy status The energy status information is updated periodically and is classified as one of the energy priority values The provision of eight different classifications is largely arbitrary The relative threshold values are adjustable in software This is a flexible system that allows the operation of the system to be adjusted based on its energy status Priority Max Description EP_Mains Operating from mains power EP_5 EP 4 80 EP 3 60 Intermediate energy leve
81. and K McCabe IEEE begins wireless long wavelength standard for healthcare retail and livestock visibility networks http standards ieee org announcements pr_p19021Rubee html June 2006 Last accessed May 2010 J N Al Karaki and A E Kamal Routing techniques in wireless sensor networks a survey IEEE Wireless Communications 11 6 6 28 2004 C Ma Y Yang and Z Zhang Constructing battery aware virtual backbones in sensor networks Proceedings 2005 International Conference on Parallel Process ing pages 203 10 2005 R C Shah and J M Rabaey Energy aware routing for low energy ad hoc sen sor networks 2002 IEEE Wireless Communications and Networking Conference Record 1 350 5 2002 Y Liu and W K G Seah A priority based multi path routing protocol for sensor networks Personal Indoor and Mobile Radio Communications 2004 PIMRC 2004 15th IEEE International Symposium on 1 216 220 Sept 2004 T Voigt H Ritter and J Schiller Utilizing solar power in wireless sensor net works Proceedings 28th Annual IEEE International Conference on Local Com puter Networks pages 416 22 2003 A Kansal and M B Srivastava An environmental energy harvesting framework for sensor networks Proceedings of the 2003 International Symposium on Low Power Electronics and Design pages 481 6 2003 198 BIBLIOGRAPHY 84 85 86 87 88 89 90 91 92 93 94 95 96 K
82. are interface electronic data sheet format and a hardware interconnect scheme that is intended to maximise the flexibility of the system while maintaining efficiency and a relatively low component count The system has a modular design with each energy module having its own power conditioning and management interface circuitry along with an electronic data sheet It is intended to allow the energy hardware of sensor nodes to be connected together at the time of deployment and an embedded software architecture has been devised which interfaces with the energy hardware As the major parts of the system have now been defined it is important that the validity of the scheme is verified by way of a prototype The proposed scheme is implemented and validated through the use of a case study prototype system The hardware for this is described in Chapter 4 and the embedded software is documented in Chapter 5 The case study represents a realistic and demanding application of the proposed scheme It incorporates a range of energy resources including energy harvesters and other energy related devices such as batteries and even a mains electricity supply 3 9 2 Evaluation criteria The main contributions of this work were stated in Section 1 4 The proposed architec ture will be evaluated by way of a case study reported in the following chapters which will assess the effectiveness of those contributions In short the evaluation of the work wi
83. are is located on the modules themselves rather than on the multiplexer module However it should be noted that each module is responsible for self regulating its output voltage to the maximum supported by the multiplexer module in the case of the prototype modules produced under this project it was 4 5V In the case of harvesting modules this means that a voltage regulator or voltage detector and isolation transistor arrangement may be necessary for the purposes of overvoltage protection It is also essential that digital communication lines are able to interpret voltages in the regulated supply range so that they can be managed by the microcontroller In summary the multiplexer module simply acts as a go between allowing the node s microcontroller to interface with the energy modules There are two types of connection microcontroller to multiplexer and multiplexer to energy module These are described in the following subsections Microcontroller to Multiplexer For the interface between the microcontroller and the energy subsystem the aim is to provide a good level of functionality while minimising the number of microcontroller I O pins required The lines listed in Table 3 5 are provided between the microcontroller and the multiplexer module Two lines are for the power supply the other eight connect to the equivalent of one 8 bit port on the microcontroller the required interface is comprised of one ADC input four digital ou
84. ary to include a hardware abstraction layer which is unique for each microcontroller It provides the ultimate interface between the embedded software stack and the phys ical hardware on the device It is unfortunately the case that the actual method of interfacing with peripherals including ADCs and general purpose I O pins is unique to each microcontroller family and even devices within the same family depending on their feature set A number of functions and macros have been implemented under this project including e halSampleADC which accepts the port and pin numbers as inputs configures the ADC and returns the result of the ADC conversion as a scaled int e g with a value of 1500 corresponding to a voltage of 1 5V e halSampleTemperature which directs the on chip temperature sensor output through the ADC and returns the result as a scaled unsigned int in degrees Celsius so 200 corresponds to 20 C e halSampleSupplyVoltage which uses the on chip facility to monitor the supply voltage being provided to the microcontroller utilising the ADC and returning the result as a scaled 8 bit result nominally an unsigned char with 230 corre sponding to 2 30V e halDirPortPin which sets the direction i e output or high impedance input of individual pins on the microcontroller The port pin and direction are all passed to the function as unsigned char variables Chapter 5 Case Study Deployment in a Prototype System So
85. as is the additional processing overhead due to carry out these calculations The net benefit to the system is in delivering energy awareness through an integrated system The process of developing and of deploying a system which is compliant with this scheme is explored and the benefits and drawbacks are expressed in terms of programming effort and ease of deployment 8 Chapter 1 Introduction 1 5 What is reconfigurability In the context of this thesis reconfigurability refers to the ability 1 To select energy resources as appropriate to the deployment environment and connect them together at the time the system is installed 2 To enable the energy resources to be exchanged while the system is operational and for the system to recognise and adapt to these changes 3 To realise these changes in the field without access to specialist equipment i e without needing to change the microcontroller s embedded software These capabilities mean that it is possible to set up the energy hardware on the device as appropriate to the environmental conditions and to be able to reconfigure the energy hardware on the device without disrupting its operation The reconfigurable system described in this thesis enables the system to remain energy aware and manage its operation regardless of the hardware used to power the device 1 6 Publications The following journal and conference papers have been produced as a res
86. asy to produce a working prototype with satisfactory voltage and current characteristics as shown by Roundy and Wright s prototype 37 which is based on the principle shown in Figure 2 9 In this example the generator is simply formed of a piezoelectric beam and large mass When the beam and mass arrangement are subjected to vibrations at their resonant frequency deformation of the piezoelectric beam produces large voltages FIGURE 2 9 Two layer bender mounted as a cantilever reproduced from 37 An advantage of piezoelectric devices is that they do not need an additional voltage source to begin operation differing from electrostatic generators in this respect and are better suited to mass production than electromagnetic generators It may be sum marised that piezoelectric generators share many of the advantages of electromagnetic and electrostatic generators Piezoelectric generators however are not as suitable as electrostatic generators for production using standard MEMS processes Another draw back is that they are high impedance sources that produce high voltages but low current which can be difficult to convert efficiently to DC Piezoelectric based vibration energy harvesting technology has been commercialised by AdaptivEnergy 88 as shown in Figure 2 10 and Mid with their Volture generator range 39 24 Chapter 2 Background Energy Sensing and Wireless Communication Disassembled RLP voltage Low los
87. at each consume 1 8u4A Additionally when recharging of the cell is enabled the voltage detector IC consumes an additional 0 7A and the voltage regulator IC consumes 0 8uA of quiescent current Futhermore a diode is required in the recharge path to prevent the reverse flow of energy through the voltage regulator IC again this will result in an approximate 9 reduction in recharging efficiency when recharging at 1mA The fact that a linear regulator is used means that efficiency will be limited when the raw voltage on the multiplexer module is significantly higher than 3V Chapter 4 Case Study Deployment in a Prototype System Hardware 117 4 5 7 Supercapacitor module The supercapacitor module from the schematic in Appendix A Figure A 5 and shown in Figure is designed to accommodate both CAP XX thin flat supercapacitors such as the GS206F 0 55F model fitted in this prototype and Panasonic Gold HW series supercapacitors The DS2502 EPROM stores information on the size and type of supercapacitor fitted Additional circuitry can be added to allow the microcontroller to query the stored voltage on the supercapacitor but this has not been fitted as in the configuration used in the demonstration the voltage on the supercapacitor is the same as the unregulated voltage on the multiplexer module Provision has also been made for balancing resistors to be fitted to the module This is arguably the simplest module developed for the case stu
88. at the current state of the art in sensor nodes and existing systems operating from harvested energy Later chapters present the fundamental work carried out in line with the aims of the project along with descrip tion of practical developments including a demonstration system Chapter 3 describes the design of a wireless sensor node as a reconfigurable system The chapter covers the overall design of the system for reconfigurability including the development of an energy electronic data sheet common hardware interface and algorithms for energy manage ment General structures for the system s energy subsystem and its embedded software are also introduced The development of prototype system hardware is described in detail in Chapter 4 with a discussion of the overall system design along with detailed consideration of each module developed The embedded software implementation for the system is documented and evaluated in Chapter 5 This thesis concludes in Chapter 6 10 Chapter 1 Introduction with a review of the work carried out and considers opportunities for standardisation and future development Chapter 1 Introduction Key Contributions Chapter 6 Conclusions amp Future Work FIGURE 1 6 Thesis chapter structure Chapter 2 Background Energy Sensing and Wireless Communication 2 1 Introduction This chapter gives an overview of the technologies related to this project and the current state
89. ata in locations indicated by pointers Initialisation and change detection Essential to the energy aware operation of the node these functions initialise the pins on the device to be initialised and enable changes to the devices connected to the multiplexer module to be detected by the microcontroller e pyelnit initialises the inputs and outputs from the microcontroller along with the pointers to the data sheet locations In the case of the MSP430 the EEDS data for energy modules are stored in flash pointers direct to the appropriate flash location or other memory location for alternative microcontrollers Address pins are set as outputs and initialised to zero and the EEDS interface pin is configured as an input pin but with a zero output so that when its direction is changed it will pull the line low The measurement control line is set as an output and initialised to zero and the measurement line is set as an input The device control pins are both set as inputs but again with zero outputs The halInitialiseEEDSPointers function is called to initialise the EEDS pointers appropriately e pyeRescan sequentially queries each port on the multiplexer module and issues a 1 Wire reset checking for a presence pulse in reply If a presence pulse is detected and there is no record in memory of a device being connected to that address the function returns a 1 Similarly if there is no device detected when the system has a
90. ate of charge of the battery This is either by analysing the open circuit voltage or the voltage after pulsed discharge through a small resistive load set by resistors soldered onto the board This board has two bistable multivibrators which maintain the discharge enable and charge enable status of the module This means that these parameters can be asserted or negated by the microcontroller and that state will be maintained by the module The two push buttons on the bottom of the board are manual on and off switches which permit the output to be manually enabled to allow the system to be cold started To 116 Chapter 4 Case Study Deployment in a Prototype System Hardware deliver a near instant start up to the system the on button may be pressed which will cause the system to receive power from the battery The second bistable multivibrator which controls recharging of the cells is solely controlled by the microcontroller there are no push buttons for this function The recharging mechanism is designed only for NiMH cells NiMH cells have a nominal voltage of 1 2V but this rises on charging For example Sanyo Eneloop cells peak at over 1 6V when charging Commonly constant current charging with charge termina tion techniques such as AT or dT dt are used for these cell types However in this application particularly where energy is obtained directly from harvesting sources a constant c
91. ated corresponds to its measured voltage and how to interpret measurements from the device This enables autonomous sensors to monitor and manage the status of the individual modules in their energy subsystem thus delivering an architecture which achieves energy aware operation The design of the enabling hardware is described in Section and supports a number of energy devices while permitting each EEDS to be individually read by the microcontroller 64 Chapter 3 Development Towards a Reconfigurable Energy Subsystem 3 3 2 Data sheet format The basic EEDS structure is shown in Table 3 1 This permits the microcontroller to obtain sufficient information to identify interpret and manage each energy module It is anticipated that for later revisions of the data sheet format should the work be taken forward to standardisation descriptors such as manufacturer ID and model number would also be included as is presently the case with TEDS to provide traceability for modules This could be managed by an organisation such as the IEEE Field Size Description Device ID 64 bits Serial number of EPROM memory Module Type 8 bits Identifies class of module and its capabilities Parameter 1 8 bits Parameter 2 8 bits 8 bits Dependent on module class parameters enable energy calculations TABLE 3 1 Outline format for the Energy Electronic Data Sheet The parameters stored in the EEDS are sufficient for the
92. aults on the node is non trivial The ability to monitor the performance of each harvesting device would allow the identification of otherwise undetectable problems such as the poor performance of a photovoltaic module for example due to dust accumulation or periodicities for example a node being powered by vibration from a piece of machinery that runs at set times during the day I make the case here that a plug and play capability which would facilitate the config uration of energy hardware at any time leading up to the deployment and first switch on of the system is highly desirable This hardware would by default act to provide power to the microcontroller and the software running on the microcontroller would be able to recognise and manage its energy hardware being able to measure both the amount of energy stored and the level of power being generated Furthermore a system capable of plug and play operation would recognise changes to the hardware of the system and allow these to be taken into account thereby achieving the reconfigurability described in Section 1 5 Thus the system requires 1 A modular design with each energy module incorporating its own power condi tioning and management interface circuitry 2 The ability to be plugged together at the time of deployment and modified after wards with energy devices being attached exchanged or removed as required 3 Embedded software which can rec
93. bat OF Southampton University of Southampton Research Repository ePrints Soton Copyright and Moral Rights for this thesis are retained by the author and or other copyright owners A copy can be downloaded for personal non commercial research or study without prior permission or charge This thesis cannot be reproduced or quoted extensively from without first obtaining permission in writing from the copyright holder s The content must not be changed in any way or sold commercially in any format or medium without the formal permission of the copyright holders When referring to this work full bibliographic details including the author title awarding institution and date of the thesis must be given e g AUTHOR year of submission Full thesis title University of Southampton name of the University School or Department PhD Thesis pagination http eprints soton ac uk UNIVERSITY OF SOUTHAMPTON FACULTY OF ENGINEERING SCIENCE AND MATHEMATICS School of Electronics and Computer Science A Comprehensive Scheme for Reconfigurable Energy Aware Wireless Sensor Nodes by Alexander Stewart Weddell Thesis for the degree of Doctor of Philosophy May 2010 UNIVERSITY OF SOUTHAMPTON ABSTRACT SCHOOL OF ELECTRONICS AND COMPUTER SCIENCE Doctor of Philosophy A COMPREHENSIVE SCHEME FOR RECONFIGURABLE ENERGY AWARE WIRELESS SENSOR NODES by Alexander Stewart Weddell Wireless sensor nodes are devices that perform
94. been designed to interface with a range of energy hardware via a multiplexer module as specified in Section 3 5 making use of the data stored on the EEDS of each energy module The energy stack enables the energy aware operation of the system by allowing the incoming power and stored energy to be measured and the energy priority of the system to be computed It also allows the system to make decisions about whether to enable charging or discharging of energy stores The three distinct layers allow the physical interfacing with the components to be effectively separated from the higher level computation thus meaning that higher levels can be written in a device agnostic way There are certain limitations inherent in the necessity of using fixed point variable types which are described later however the system has been developed to be as flexible as possible and where relevant these limitations are defined and tailored for use in sub milliwatt sensor nodes 5 4 5 Application layer and control scheme The shared application layer interfaces with the three individual stacks In the imple mentation developed under this project the application layer takes information from the energy stack in order to control its duty cycle effectively sleeping for longer in the case that the energy status is lower Data is taken from the sensing stack and used to format messages to be sent through the communication stack The system developed acts as an autonomou
95. cSigType Transducer electrical signal type ID MinPhysVal Minimum temperature CAL 11 C MaxPhysVal Maximum temperature CAL 11 C MinElecVal Minimum resistance output CAL 18 Ohms MaxElecVal Maximum resistance output CAL 18 Ohms MapMeth Mapping method ID o RTDCoef_RO Resistance of thermistor at 0 C ID 20 Ohms SteinhartA Steinhart Hart Coefficient A ID 32 1 C SteinhartB Steinhart Hart Coefficient B ID 32 1 C SteinhartC Steinhart Hart Coefficient C ID 32 1 C RespTime Sensor response time ID 6 seconds ExciteAmplNom Nominal current excitation CAL 8 Amps ExciteAmplMax Maximum current excitation ID 8 Amps SelfHeating Self heating constant ID 5 Ww C CalDate Calibration date CAL 16 CalIntitials Calibration initials CAL 15 CalPeriod Calibration period CAL 12 days MeasID Measurement location ID USR 11 TABLE 2 7 Thermistor Template ID 38 Summary reproduced from 57 2 4 3 Extensions to the electronic data sheet concept The concept of electronic data sheets has been taken forward by Bandari et al 60 who have developed the component electronic data sheet CEDS which stores information about electrical and mechanical components in a system This forms part of a process for integrated systems health management ISHM in which the overall reliability of systems is optimised by increasing the level of information stored about each individual component s operating parameters in t
96. cal damping of the system 32 Electromagnetic Electromagnetic generators normally use a resonant beam and coil arrangement with movement of a magnet relative to a coil inducing an electrical current A number of re search groups have developed electromagnetic converters with varying levels of success This technology is particularly applicable to harvesting lower frequency vibrations for example Amirtharajah and Chandrakasan s generator was designed to work with fre quencies of approximately 2Hz the frequency of footfalls for human walking with a claimed power output from their generator of the order of 4004W 33 Commercial generators are now able to harvest energy from vibrating machinery PMG Perpetuum have developed two versions of their PMG17 generator shown in Figure 2 7 tuned to 100Hz and 120Hz and FerroSolutions manufacture a similar device the VEH 360 which is tuned to 60Hz 100 0 NS 10 0 HA O o PMG17 Power Output mW Di 10 Hz 5 Hz CENTRE 5 Hz 10 Hz Vibration Frequency A Generator B Performance FIGURE 2 7 PMG Perpetuum s PMG17 vibration harvesting generator and its data sheet frequency response reproduced from 34 Efforts are continuing to miniaturise the electromagnetic generator the PMG Per petuum model pictured has a height of 55mm and a diameter of 55mm The device shown in Figure 2 8 developed by Torah et al has a total volume of approximately
97. ce between the capacitor plates o is the permittivity of free space and and w are the length and width of the capacitor plates ae 2 2 colw Hence the capacitance C is given by Equation 2 3 C 2 3 Electrostatic generators are perhaps the best suited to fabrication using standard MEMS processes but they have two main disadvantages Firstly they require an initial voltage Chapter 2 Background Energy Sensing and Wireless Communication 25 to be applied across the capacitor plates in order to begin the conversion process which means they are unsuitable for applications where a cold starting capability is required Secondly the requirement for the plates to remain separated to avoid short circuit necessitates the use of mechanical stops to constrain the motion of the plates This brings its own reliability issues A number of attempts have been made to fabricate MEMS electrostatic generators but at the time of writing this report prototypes were experiencing reliability issues connected to the endurance of the devices It appears likely that advances in processing techniques and design will yield positive results in the In plane overlap varying Ki In plane gap closing near future c Out of plane gap closing FIGURE 2 11 Variants of electrostatic energy harvesters reproduced from 29 2 3 4 Thermoelectric energy generation Thermoelectric generators exploit the Seebeck effect
98. cean Alliance there is currently some confusion over which standard will dominate In the coming years I expect to see one standard emerge as dominant or for each to find their own niche which will in turn increase confidence and encourage investment in wireless sensing systems for industrial applications Appendix A Module schematics This chapter shows schematics for the 1 Multiplexer Module Figure A 1 2 Photovoltaic Module Figure A 2 3 Vibration Module Figure A 3 4 Mains Module Figure A 4 5 Supercapacitor Module Figure A 5 6 Battery Module Figure A 6 These modules were described in full in Chapter 4 157 Appendix A Module schematics 158 0O ETIOFEANO 0 ETIOFEANO rola Ou zH IN 21H Obed ICM rd z E Spsllva UMANO ONO E r 9378 LaNg FIGURE A 1 Multiplexer module schematic 159 Appendix A Module schematics Spry zy NM ONS MWe IMO SON C e Lames rre nD y OF TOVERNO Bo ss ig m a ZHW 1H amp 8 Lu O SOIOSSLZZOINT g k S AR m E noo za a hee y iy CS a a 3 E zg ay w o D 16 WEIT SON
99. ces To the best of the author s knowledge there are no systems that allow the selection of appropriate energy hardware at the time of deployment or indeed allow it to be changed after deployment It may be argued that this is a major limiting factor for the applicability of wireless sensor nodes able to operate from environmental energy 6 Chapter 1 Introduction The justification for the work presented in this thesis is the development of reconfig urable efficient energy aware wireless sensor nodes which are able to overcome these limitations There is a need for devices to have a flexible energy subsystem that can ac commodate a range of devices including energy harvesters buffers and non rechargeable batteries Information on the energy status of the system can be provided to the soft ware application running on the node in order to control the node s activity level As outlined in the following section the scheme is verified on a sensor node in a challenging indoor environment allowing energy components to be exchanged and for the node to recognise and handle these hardware changes and remain aware of its energy status 1 4 Contributions of this research The scheme proposed in this thesis permits the energy resources of a sensor node to be configured in situ by system installers at the time of deployment and for the node to au tomatically recognise and manage its energy subsystem For example when the system is installe
100. cilitate digital communications with the microcontroller without the need for level shifting otherwise the microcontroller would likely be exposed to out of range and damaging digital voltages Pin Connector Description 1 1 wire EEDS line Digital 1 wire interface 2 Measurement Control Initiates measurement 3 Measurement Analogue measurement output 4 Device Control Control outputs to module these are bi directional 5 Device Control control state can be determined 6 Vreg Regulated voltage used as supply to uC 7 Vout 8 GND Raw voltage interface TABLE 3 6 Common Hardware Interface connections between multiplexer module and energy modules 3 4 4 Integration of multiple energy sources There are two main options for combining separate energy harvesting devices on a sin gle node Figure shows a system where two energy sources act through diodes to charge a single supercapacitor adjacent to the microcontroller and Figure 3 4 shows a scheme where each energy source has a separate supercapacitor which feeds energy to the microcontroller via a diode Option 1 can potentially offer better long term perfor mance where both sources act to maintain a high voltage on the supercapacitor Option 70 Chapter 3 Development Towards a Reconfigurable Energy Subsystem vn Ss Vint a Vino FIGURE 3 3 Option 1 for multiple energy source combination Features a singl
101. connection of energy modules and their connection to the multiplexer module 4 3 3 EPROMSs for energy electronic data sheets The memory device selected to store the EEDS for this system is the DS2502 1kB add only EPROM Each device has a unique serial number and 1024 bits of programmable memory partitioned into four pages Its maximum communication speed is 16 3kbps and communications are enabled by a 5k 2 pull up resistor on the 1 wire bus The de vice is rated for operation between 2 8V and 6 0V from 40 C to 85 C but its operation in this system has been successfully tested down to 2 0V at room temperature The de vice must be programmed before deployment in the system as it requires programming Chapter 4 Case Study Deployment in a Prototype System Hardware 97 voltages of around 12 0V which are not common on sensor nodes a large amount of additional hardware would be required to achieve this voltage through level shifting and voltage boost circuitry 4 3 4 State retention for device control Normally power generated by energy harvesters will be exploited whenever it is avail able being used to power the sensor node or to recharge energy stores such as su percapacitors or rechargeable batteries Conversely in the case of energy stores or non rechargeable sources such as primary batteries it is commonly desirable to be able to turn these devices on or off to enable or disable discharging to c
102. ction to limit the supply voltage to prevent damage Two types of circuit can be used to maintain the correct operation of the system Firstly the undervoltage protection circuit shown in Figure 4 2 acts to disconnect the power supply to the microcontroller if the raw voltage is too low The device features a micropower CMOS voltage detector which has built in hysteresis meaning that with a 2 0V variant of the voltage detector the circuit will turn on at around 2 1V and off at approximately 2 0V Secondly overvoltage protection can be achieved through the circuits in Figure The circuit in Figure again exploits a voltage detector and is used to disconnect an energy harvester from the rest of the circuit thus preventing further charging The circuit shown in Figure 4 3b simply exploits a voltage regulator IC which will typically be a linear device Vino pe Vout FIGURE 4 2 Undervoltage protection circuit Alternatively the undervoltage and overvoltage protection circuits can be combined if a linear regulator with an enable input is used As shown in Figure 4 4 the output from the voltage detector is fed directly into the enable input of the voltage regulator The circuit therefore cuts off its output when its input voltage is below the sensed voltage of Chapter 4 Case Study Deployment in a Prototype System Hardware 99 Vino oVout Voltage Detector Voltage Regulator A Circuit 1 B Circuit 2 FIGURE 4 3
103. ctive management The microcontroller keeps a table of which sockets on the multiplexer module are oc cupied The microcontroller will periodically re scan the sockets on the multiplexer module by issuing a reset pulse on the 1 Wire bus for each socket then listening for a presence pulse from the 1 Wire EPROM in response By comparing the presence pulses received against its table of known connections newly connected or disconnected modules can be detected Newly connected modules can be interrogated for their full data sheet while disconnected modules can be deleted from memory and removed from future calculations An important note however is that hardware changes could be missed if devices are quickly swapped on the same socket For example if a battery module on socket 2 is swapped for another device and this happens within the period between scans the microcontroller will detect a device still present in that socket and not realise that the module has been exchanged This drawback can be countered by scanning the serial number of each module after the presence pulse is detected clearly this is process will consume more energy but will result in increased confidence in the energy estimates 3 8 4 Network level interactions As discussed in Section a number of schemes exist to facilitate energy aware in teraction between sensor nodes in a wireless sensor network This proposed scheme and in particular the embedded software archite
104. ctors are widely available in 2 0V 2 7V 3 0V and 4 5V versions They consume around 0 7uA quiescent current and have CMOS active high outputs e Linear regulator Torex XC6215 This series of linear voltage regulators has an enable input which makes it suitable for being controlled by an XC61CC voltage detector It has a quiescent current draw of 0 8uA and 100nA in standby e Switching step down regulator Maxim Integrated Products MAX639 This step down DC DC converter has a 104A quiescent current draw 4 3 2 Connectors As defined in tables and 3 6 the interface between the multiplexer module and the energy modules require 8 lines and the interface between the microcontroller and the multiplexer module has 10 lines The cost of connectors is important as is their ro bustness It is expected that the energy modules will be connected and disconnected more frequently than connectors between the microcontroller and the multiplexer mod ule It is for these reasons that RJ45 connectors have been selected for the interface between multiplexer and energy modules in this project The connection between the microcontroller and the multiplexer module is via a 2 x 5 way 0 1 pin header In future systems however it is anticipated that a stacking system would be used facilitated by the reduced size of circuit boards on energy modules to reduce the overall footprint of the system A stacking 36 way connector would be suitable for both the inter
105. cture acts to structure the node level processing in order to simplify these interactions The energy stack acts to present the application layer with a simple energy priority value In future iterations of the sys tem the sensing stack will provide data with a packet priority value and messages needing to be routed will be received through the communications stack with their own packet priority values This will implement the proposed IDEALS RMR scheme which 88 Chapter 3 Development Towards a Reconfigurable Energy Subsystem was proposed by Merrett et al and was discussed earlier The scheme facilitates priority balancing in which unimportant messages are discarded when energy is scarce thus ensuring that important messages are still able to traverse the network Similarly when energy is plentiful node activity is increased to make best use of the available energy which would otherwise be wasted It may be envisaged that nodes will be able to share their energy status and negotiate to share sensing tasks this is particularly of interest in applications where sensing is the dominant consumer of energy as opposed to communications such as in gas sensing or image processing 3 9 Towards a prototype to verify the approach 3 9 1 Overview The scheme described in this chapter has been devised to deliver a plug and play ca pability for reconfigurable energy aware wireless sensor nodes It comprises a common hardw
106. d Products Inc Application Note 126 1 Wire Communication Through Software http www maxim ic com app notes index mvp id 126 August 2009 Last accessed May 2010 R Freeland ISA100 Power Sources Working Group TBC Encouraging a Single Power Source Standard www hansonwade com events energy harvesting presentations Roy Freeland Encouraging A Single Power Source Standard pdf April 2010 Last accessed May 2010
107. d background to the family of motes is given by Polastre et al 5 Used by over 100 research organisations the motes are normally loaded with TinyOS an embedded operating system for wireless sensors and are available in a number of incarnations with a range of sensor boards available The latest sensor platforms produced by Crossbow are the TelosB and Imote2 TelosB is the low power MSP430 based platform and the Imote2 is a high bandwidth sensing platform featuring an enhanced processor and larger memory along with the flexibility to communicate via a range of protocols including IEEE 802 11 and Bluetooth 96 A major competitor to the Imote2 is the Sun SPOT from Sun Microsystems 97 which has 32 bit ARM CPU and is Java based which may appeal to many software developers 46 Chapter 2 Background Energy Sensing and Wireless Communication A number of bespoke platforms have also been developed including those nodes pro duced as part of the Glacsweb and ScatterWeb projects Some users find the commercially available motes too restrictive expensive or large for the intended appli cation and some require specialist capabilities that cannot be accommodated by the mote platform Benefits may be gained from custom designing sensor nodes but this is no simple task and is beyond the capabilities of most end users of the sensor node hardware which has in part led to the dominance of Crossbow motes 2 7 3 Discussion A range of se
108. d conservatively at all times to compensate for the poor resolution of energy awareness in order to ensure that desired system lifetimes can be achieved 4 0 1 5 cut off voltage 60 70 80 90 Depth of discharge FIGURE 2 1 End of life indication for lithium primary batteries The grey curve shows seasonal temperature cycles the solid curve shows discharge on continuous load at 25 C and the dashed curve shows the trend of test pulses reproduced from Tadiran technical brochure T0 2 2 3 Rechargeable secondary batteries Some of the most frequently used rechargeable also known as secondary battery chemistries in wireless sensor network applications are lithium ion lithium polymer and nickel metal hydride Figure shows a comparison of the energy densities of rechargeable battery types the battery capacity values quoted in data sheets however only give part of the story The amount of energy that can be provided by a battery is highly dependent on the methods of charge and discharge and the conditions in which it is kept 11 For example lithium ion cells are highly sensitive to overcharge and un dercharge and thus protection circuitry must be built in Different battery chemistries require other charging methods lithium ion batteries require a constant charging cur rent until they have reached 90 capacity followed by a constant voltage NiMH and NiCd batteries prefer a constant current throughout the
109. d in a capacitor before being fed through a step up switching converter The step up converter used in this circuit consumes a relatively high 504A with a 3uA shut down current Similar to the vibration energy harvesting circuit the power generated is estimated by driving the rectified output from the generator through a known load and monitoring the output The circuit has an overvoltage protection mechanism as shown in Figure 4 3a except that the JFET transistor in that circuit is exchanged for a MOSFET and additional Schottky barrier diode Once again due to the use of a shutdown pin on the converter IC there is no need to interrupt the power path for energy awareness circuitry so the only efficiency cost of the energy awareness circuit is that when measurements in progress The pin connections for the wind module are shown in Table Pin Type Function 2 Meas Control Connect rectified generator output to fixed load 3 Measurement Analogue voltage across fixed load 4 Control None 5 Control None TABLE 4 6 Interface pins from wind module The thermoelectric generator gives a DC output so does not require rectification The voltage produced by the generator is stepped up using a dedicated IC the quiescent current consumption of the step up converter is 164A with a shutdown input The power produced by the generator is measured by connecting it through a fixed load Chapter 4 Case Study Deployment in a Prototy
110. d the appropriate types and sizes of energy harvesting and storage devices can be chosen and connected to the system On first power on or indeed afterwards when a change is detected the node interrogates a memory on each device to determine its operating parameters and other information The introduced scheme permits multiple energy devices to be connected to a node with the node able to address and manage each device individually This represents a major step forward for the application of energy harvesting for sensor nodes as nodes can now be efficient and energy aware without being limited to a specific energy subsystem The scheme includes the implementation of a new embedded software structure a novel common hardware interface to allow the sensor node to communicate with its reconfig urable energy subsystem and a new energy electronic data sheet format which allows data on parts of the energy subsystem to be stored on the modules themselves Together this represents a comprehensive scheme for reconfigurable energy aware sensor nodes This scheme is validated through the application of the methodology to a challenging application using a range of energy devices A prototype system has been deployed in an indoor environment harvesting energy of the order of one milliwatt It was developed under this scheme to sustain its operation and has a plug and play energy subsystem which allows its energy components to be swapped at the time
111. dded removed or exchanged at any time No reconfiguration of the embedded software is required as the system can auto detect any changes made and continue to manage the energy subsystem It is anticipated that system designers will call upon a library of energy modules in a similar way to how batteries are selected today There will obviously be cost con siderations regarding the selection of modules and the dynamics of the environmental energy resource must also be taken into account For example if considering the use of a photovoltaic module one will typically need to model the dynamics of the system over a 24 hour period to cover both day and night and ensure that the energy subsystem is capable of supplying energy to maintain the sensing schedule Similarly if the system is powered from waste energy from the machinery it is mounted on for example on a pump powered from mains electricity one must consider the effects to the energy subsystem of a malfunction or failure of that machine and whether this should affect its sensing abilities 3 2 4 Major challenges and strategy As discussed in Chapter 2 wireless sensor nodes are highly resource constrained They are based on microcontrollers with relatively small amounts of memory low clock speeds and a restricted number of inputs and outputs The energy resources for these devices are also very limited conventionally primary batteries have been used which are norma
112. disconnected from socket 3 and connected to socket 6 which resulted in two transmissions firstly to shown that socket 3 was now empty secondly to show that a device had now been connected to socket 6 The third test presented here involved the monitoring of the photovoltaic energy module The microcontroller was instructed to query the device and to report the measured value From this the open circuit voltage and hence nominal power can be found from the equations given in Section 4 5 1 With a light intensity measured at 900 Lux the measurement output from the photovoltaic module was 1 09V The unregulated voltage on the multiplexer module was 3 89V which led to an estimated open circuit voltage estimate of 4 54V The expected open circuit voltage at this light intensity was 4 65V Therefore the error in the estimated open circuit voltage is approximately 2 Given that standard tolerance resistors were used in this module it is expected that the accuracy of this estimate can be improved Initial start up 2 852V 26 0C 1 00000576c739 852V 26 0C 2 000005770b4b 852V 26 0C 3 000005770407 852V 26 0C 4 00000576f5b6 852V 26 0C 5 000000000000 2 892V 26 0C 6 000000000000 Connected vibration module to socket 3 2 844V 26 0C 5 00000576d4c3 Disconnected photovoltaic module from socket 3 2 846V 25 9C 3 000000000000 Reconnected photovoltaic module to socket 6 2 846V 25 3C 6 000005770407 2 2 2 2 FIGURE 5 9 Annotated Hyperte
113. dware such as data handling and buffering burst transmissions encryption and authentication clear channel assessment quality indication and timing information It is a requirement of 802 15 4 transceivers that they must have an adjustable output power Settings for the transmit power along with various other parameters are accessed via memory registers on the transceiver Further developments have led to single chip solutions being developed for example the Texas Instruments TI CC2430 integrates an enhanced 8051 microcontroller and 802 15 4 compliant transceiver onto a single chip A CC2430 evaluation module is shown in Figure 2 23 alongside an eZ430 RF2500 module the eZ430 has a separate MSP430 microcontroller and transceiver whereas the CC2430 has a single chip TI have recently developed the CC430 which integrates an MSP430 and sub GHz radio transceiver in a 9 1mm x 9 11 system on chip package 95 SEI Cc Chipcan AS FIGURE 2 23 The Texas Instruments CC2430 evaluation module and eZ430 RF2500 next to a five pence coin 2 7 2 Commercially available sensor nodes A number of wireless sensor platforms have been developed each having different ca pabilities and features such as on board sensors or processors Motes developed at the University of California Berkeley have been commercialised through Crossbow Technol ogy Inc and have achieved dominance through their flexible interfacing small size and reasonable cost A goo
114. dy with no significant additional components The management connections to the module are shown in Table 4 11 o a Ute E FIGURE 4 18 Circuit board for the supercapacitor module Pin Type Function 2 Meas Control Measure supercapacitor voltage 3 Measurement Analogue supercapacitor voltage measurement 4 Control None 5 Control None TABLE 4 11 Interface pins from supercapacitor module 4 6 System integration 4 6 1 Complete system A demonstration system including a multiplexer module and four energy modules su percapacitor battery photovoltaic and mains is shown in Figure 4 19 A further two energy module sockets on the multiplexer module are left unconnected Energy modules can be connected to any RJ45 socket on the multiplexer module and are connected here by standard 300mm RJ45 patch leads The energy subsystem shown is connected to Port 0 of a TI CC2430 evaluation module EM via a 10 way IDC cable The interface with the CC2430 EM is via its breakout board without batteries attached 118 Chapter 4 Case Study Deployment in a Prototype System Hardware CC2430EM 2 Photov ltaic Module 3 J FIGURE 4 19 Complete system connection with battery mains supercap and photo voltaic modules Connected to TI CC2430EM via its breakout board 4 6 2 Default operation The default behaviour of the system on first installation is to allow the energy harvesting device s to
115. e TABLE 4 8 Interface pins from mains module 114 Chapter 4 Case Study Deployment in a Prototype System Hardware 4 5 5 Primary battery module Design overview As stated in Section this module supports a range of battery types including a lithium thionyl chloride 1 2AA cell a CR2430 coin cell and a pair of alkaline AAA cells The module from the schematic in Appendix A Figure A 6 and shown in Figure 4 17 uses the same PCB as the secondary battery module except that a number of the components are not fitted in this case The module has a bistable multivibrator circuit as described in Section 4 3 4 which controls the discharge state of the battery It may be observed that the PCB is also fitted with two push buttons These interact with the discharge state retention circuit and allow the discharge state of the module to be controlled manually This is useful when installing the system for the first time as it allows the system installer to ensure that the system can start up instantly from the battery without having to wait for other energy buffers to be filled from harvested energy As the microcontroller outputs are tri stated when not actively setting the state of the energy module there is rarely a conflict between the push buttons and the microcontroller state In the rare case of a conflict the microcontroller state will take precedence as the push buttons are connected through 10kQ resistors to the state
116. e area Sensors have also been used to monitor the habitat of petrel on Great Duck Island with up to 150 Crossbow based motes deployed weather motes monitored surface conditions and burrow motes monitored the condition and occupancy of nesting burrows 105 ZebraNet used nodes equipped with GPS to monitor the movement of zebra in the Sweetwaters game reserve in central Kenya with mixed results a detailed analysis of the hardware issues is presented in a paper by Zhang et al 106 50 Chapter 2 Background Energy Sensing and Wireless Communication comms vo ma i 433 MHz 6 off I e a l l al 1 Tilt sensors 7 I SEH wl pepe s227 2 yReoRc V 1 Micro I controller AN1 AN2 Strain gauge I l 3 L vottage reg besoin PIC16LF876A 5 digital 3V ma a Pressure BS i sensor lt 4 ER l 35 ZE l SW y y FLASHROM Temperature I 64kB sensor I Voltage reg l sch analogue 3V E E E E E E E E E E e pe A B Architecture C Base station Probe FIGURE 2 27 The Glacsweb Mk II architecture reproduced from 98 Some large scale sensor network deployments have been reported The largest of which ExScal has deployed around 1 200 nodes over a 1 3km by 300m open space Preci sion agriculture appears to be a large and growing application area for wireless se
117. e constrained sensor nodes this means that it can easily be transferred to higher power more capable sensor nodes and deliver equivalent functionality Given the nature of technological advances with a convergence between the power out put of energy harvesting devices and the power consumption of wireless sensor nodes systems will need to operate close to the limits of energy availability Energy aware schemes such as that described in this thesis will become more important This work is well placed for future developments in energy harvesting and wireless sensor node technologies to enable energy awareness for increasingly resource constrained systems 154 Chapter 6 Conclusions and Future Work 6 3 Recommendations for future work 6 3 1 Overview This thesis has described the development of a scheme for reconfigurable energy aware wireless sensor nodes It has delivered a hardware interface and electronic data sheet format that enables plug and play configuration of the energy hardware of wireless sensor nodes The system has been demonstrated by way of a case study implemented on the MSP430 and CC2430 microcontroller platforms The work so far has concentrated on the development of the energy stack and the energy hardware The concept of reconfigurable sensor nodes which are able to be connected and deployed in a plug and play manner remains compelling The future work anticipated includes reducing the form factor of the system and
118. e flash memory cannot be written to below a certain voltage this means that data cannot be written to the MSP430 s memory space allocated for EEDS data Effectively this means that the system must start up into the unknown state and delay reading the module electronic data sheets until the microcontroller supply voltage has risen above this threshold It also means that should the voltage again drop below this threshold the electronic data sheet values cannot be overwritten or manipulated This fact conflicts with the conventional need to keep the supply voltage to the microcontroller as low as possible in order to minimise its power consumption 5 5 2 Energy Analysis EAN layer Energy and power estimation functions These are key functions for this layer of the stack that are used to calculate the overall energy status of the node Three functions are provided eanGetPower allows the system to estimate the amount of incoming energy at a given instant eanGetEnergy estimates the energy stored on the system and eanGetVoltEnergy estimates the amount of energy that would be stored by the system given a certain system voltage 140 Chapter 5 Case Study Deployment in a Prototype System Software Legend 16 MHz 4 Supply voltage range N during flash memory 3 programming T 12MHz 4 3 H H E H Supply voltage range i during program execution uw 7 5 MHz y 4 E H 2 d V i gt i 4 15 MH
119. e generally unable to directly power nodes in their active mode so it is normally necessary to buffer energy to accommodate the periodic large bursts of current draw caused by duty cycled operation i e when the node becomes active after a period of sleep Electrical energy may also be stored for longer periods for example overnight to compensate for periods where energy cannot be generated The buffering of energy is normally achieved by using rechargeable batteries supercapacitors or a combination of the two This section reviews a selection of popular battery and supercapacitor technologies their characteristics and considerations for their use with wireless sensors Some alternative methods of energy storage are also considered and the section concludes with a com parison between commercially available energy storage devices considering factors such as energy density leakage power and cost 2 2 2 Non rechargeable primary batteries Non rechargeable otherwise known as primary batteries are by far the most prevalent technology for powering wireless sensor nodes as they are relatively cheap and offer a high energy density compared to rechargeable batteries or supercapacitors The batteries are either replaced when depleted or the node is deployed with a battery which will last for the system s entire operational lifetime In the field of low power wireless sensor networks two primary battery chemistries have
120. e hydrogenerator was manufactured by Vulcano and is more commonly used in domestic water heating providing electricity to ignite a gas boiler when water flows with one of the main benefits being that the boiler does not need a mains electricity connection The same project has also developed a vertical wind powered generator shown in Figure and Similarly the AmbiMax project has used a small off the shelf wind turbine to power a wireless sensor node as shown in Figure 28 Chapter 2 Background Energy Sensing and Wireless Communication 20 20 18 18 nnn J e Ier 7 9 z 14 Bur 1 S e er Bist g D 12 lt 12 F 4 g 10 E1oL 5 BE B E Ss 420 3 3 2 6 a E 0 Output power Pour Ru 1000 Output voltage Vou Ri 1002 T o Output current fou Ri 1002 t Output voltage Voute Ri 00 05 Lrpiliililliil_ dd lo 0 0 01 02 03 04 05 06 07 08 09 10 g A Water flow m s A Device B Performance FIGURE 2 14 Bosch Hydropower generator used with the MPWiNodeX reproduced from 45 900 r r r r r m y 800 4240 E S 700 A 210 amp E o 600 180 8 gt S 8 500 150 E ES E g wr e 4 120 2 Z mr 90 5 5 200 60S E ei L ER on 10 20 3 0 4 0 5 0 6 0 7 0 8 0 9 0 10 0 Wind speed ms a B c FIGURE 2 15 Wind powered generators used by wireless sensor nodes a is a pro totype vertical unit used by Morais et al and b i
121. e included here 1 Weddell A S Grabham N J Harris N R and White N M 2009 Modular Plug and Play Power Resources for Energy Aware Wireless Sensor Nodes Sixth Annual IEEE Communications Society Conference on Sensor Mesh and Ad Hoc Communications and Networks SECON 2009 22 26 June 2009 Rome Italy 2 Weddell A S Merrett G V Harris N R and Al Hashimi B M 2008 Energy Harvesting and Management for Wireless Autonomous Sensors Measurement Control 41 4 3 Weddell A S Harris N R and White N M 2008 Alternative Energy Sources for Sensor Nodes Rationalized Design for Long Term Deployment International Instrumentation and Measurement Technology Conference May 12 15 2008 Vic toria British Columbia Canada 165 Appendix C Selected Publications 167 Weddell A S Grabham N J Harris N R and White N M 2009 Modular Plug and Play Power Resources for Energy Aware Wireless Sensor Nodes In Sixth Annual IEEE Communications Society Conference on Sensor Mesh and Ad Hoc Communica tions and Networks SECON 2009 22 26 June 2009 Rome Italy This publication is not available in the online version of this thesis but may be down loaded from http eprints ecs soton ac uk 17325 168 Appendix C Selected Publications Weddell A S Grabham N J Harris N R and White N M 2009 Modular Plug and Play Power Resources for Energy Aware Wireless Sensor Nodes In Sixth A
122. e required low impedance input to the ADC of the microcontroller Resistors R3 and R4 act to ensure that the circuit operates normally on start up when supercapacitor C2 has built up insufficient voltage for the microcontroller to become active and the switching converter is required to operate normally The Vo of the module can be calculated by means of Equation 4 4 where a is the ratio of R1 to R2 and Vc is the voltage at which the microcontroller is operating i e the voltage across supercapacitor C2 Voc bas x a 1 e Vue 4 4 Here R6 and D1 provide a reference voltage input to the micropower comparator IC2 R7 and R8 provide hysteresis to the system In normal operation when Ve is low Cl acts as a small buffer capacitor When the voltage across C1 exceeds the threshold set by R5 comparator 1C2 gives a low output which switches MOSFET M2 on and permits current to flow through D2 and L1 After a very short period the voltage across C1 will have dropped sufficiently to cause IC2 to turn M2 off and for energy to be transferred from L1 through D3 to be stored in supercapacitor C2 note the polarity of this component For SPICE simulation of the power conditioning circuit a model was developed for the PV module used in this investigation Typical models of solar cells such as those pro 108 Chapter 4 Case Study Deployment in a Prototype System Hardware R6 M2 y A
123. e su percapacitor adjacent to the microcontroller A K Vi Fae uC Vie FIGURE 3 4 Option 2 for multiple energy source combination Features separate supercapacitor for each energy source 2 while featuring a higher component count may offer faster start up times as energy sources could charge up smaller capacitors only one of which is required to reach the microcontroller s turn on voltage for the system to function The system developed un der this project uses a variant of the Option 1 scheme all power conditioning electronics are located on the energy modules 3 5 Generalised system hardware specification 3 5 1 Energy multiplexer The purpose of the multiplexer module is to permit the interconnection of energy mod ules to supply a stable power supply to the microcontroller and to allow it to monitor and manage the energy modules through their data lines The multiplexer module as a minimum features the following hardware 1 Five 8 to 1 multiplexers one 1 wire line one analogue measurement line one digital control and two bidirectional digital I O to facilitate the connection of up to six energy modules smaller multiplexers can be used to facilitate connection of fewer modules Chapter 3 Development Towards a Reconfigurable Energy Subsystem 71 2 Undervoltage protection circuit to inhibit supply to microcontroller 3 Voltage regulation nominally to 3 0V f
124. ecomes more limited therefore x 8 is the highest power that can be computed with reasonable precision with approximately 5 error to 13 terms In realising the calculations to estimate the power being generated by the photovoltaic module as shown in equations and 5 3 which relate to the pa rameters given in Table 5 2 care must be taken to ensure that in normal operation e is not raised to a power beyond 8 in some cases it may be necessary to optimise the parameters held in the EEDS for photovoltaic modules in order to ensure this oe r r rt e r E E Ee 5 1 n 0 d Voc Vmeas x 10 Vraw 5 2 Voc b Pa Ha xecne e 5 3 3500 3000 2500 as 5 2000 K KI So CG 1500 1000 500 0 0 10 15 Nn Number of Terms FIGURE 5 8 Calculation of ef using Taylor expansion showing how the number of terms affects the precision of the result Another detail is that sentinel values are used by some of the functions to indicate that the system is connected to the mains the eanGetPower function returns the maximum value for the unsigned int variable 65 535 in order to signal that it is a device supplying a large amount of power This is interpreted by the other functions as a signal that the device is under mains power and allows the EP value to be set accordingly 5 5 3 Energy Control ECO layer The ECO layer makes a number of functions available which enable the microcontroller to co
125. ed Hardware Software Architecture for Embedded Sensor Nodes In 17th International Conference on Computer Communications and Networks 3 7 August 2008 St Thomas Virgin Islands USA 7 Merrett G V Weddell A S Berti L Harris N R White N M and Al Hashimi B M 2008 A Wireless Sensor Network for Cleanroom Monitoring In Eurosensors 2008 7 11 September 2008 Dresden Germany 8 Weddell A S Grabham N J Harris N R and White N M 2009 Modular Plug and Play Power Resources for Energy Aware Wireless Sensor Nodes In Sixth Annual IEEE Communications Society Conference on Sensor Mesh and Ad Hoc Communications and Networks SECON 2009 22 26 June 2009 Rome Italy A selection of these papers may be found in Appendix The author of this thesis has also co authored a chapter entitled Wireless Devices and Sensor Networks for the book Energy Harvesting for Autonomous Systems edited by Beeby S P and White N M which is to be published by Artech House London in July 2010 1 7 Document structure The overall structure of this thesis is shown in Figure Chapter 2 introduces the basis for the operation of sensor nodes energy sensing and communications It covers the available technologies for energy storage and generation and how to monitor these resources Intelligent sensing standards are also discussed along with wireless transmis sion protocols relevant to sensing It then looks
126. elsing S J Thompson and Y B Cheng A power management architecture for sensor nodes IEEE Wireless Communications and Networking Conference WCNC pages 3010 3015 2007 J Taneja J Jeong and D Culler Design modeling and capacity planning for micro solar power sensor networks In IPSN 08 Proceedings of the 7th interna tional conference on Information processing in sensor networks pages 407 418 Washington DC USA 2008 IEEE Computer Society GE Energy Essential Insight mesh Wireless Condition Monitoring http www gepower com prod_serv products oc en bently_nevada essential_insight htm 2010 Last accessed May 2010 M H Schneider J W Evans P K Wright and D Ziegler Designing a thermo electrically powered wireless sensor network for monitoring aluminium smelters Proceedings of the Institution of Mechanical Engineers Part E Journal of Process Mechanical Engineering 220 3 181 90 2006 EnOcean GmbH Thermal Energy Harvester ECT 100 http www tdc co uk index php key ect100 August 2007 Last accessed May 2010 A Hande T Polk W Walker and D Bhatia Indoor solar energy harvesting for sensor network router nodes Microprocessors and Microsystems 31 6 420 432 2007 BIBLIOGRAPHY 201 119 120 121 122 123 124 125 126 127 128 129 E Leder A Sutor and R Lerch Solar powered low power sensor module with a radio communication and a user in
127. em Hardware 107 Maximum power point circuit A circuit for indoor PV energy harvesting has been developed Figure 4 11 which con sists of a modified buck boost converter circuit with additional isolation and metrology circuitry During normal operation this circuit acts to maintain a constant voltage across its input terminals in order to keep the PV cell at a voltage set by R4 The PV cell is kept at this fixed voltage in the range of maximum power point voltages of the cell at expected light levels in order to keep its operation near its indoor MPP voltage This simplification is acceptable as the normal range of indoor light levels is 100 1 000 lux and in this range the power loss due to voltage clamping at a fixed voltage has been determined to be less than 3 The combination of JFET J1 and MOSFET M1 act to disconnect the load from the photovoltaic module when a high signal is raised on the Very line This permits the open circuit voltage Voc of the cell to be measured However a complication of this circuit is that as it is based on a buck boost converter the input and output stages do not share a common ground effectively the ground of the input is the Vec of the output For this reason a high impedance resistor divider comprised R1 and R2 is used to bring the Vo signal into a range which can be interpreted by the microcontroller The output from this divider is fed through a unity gain buffer around IC1 to provide th
128. entage values representing the amount of useable energy on the node compared with the capacity of the node However in alternative implementations the node may use absolute energy values this would be especially useful in heterogeneous networks where nodes have widely varying energy subsystems and would be enabled by simply exchanging the modules in the ECO layer of the stack This is a key benefit of the stacked node architecture Chapter 5 Case Study Deployment in a Prototype System Software 145 while 1 ecoUpdateEnergy if ecoPowerStatus EPEMPTY EP_UNKNOWN d appGetData msg SMPL_Ioctl IOCTLOBJ_RADIO IOCTL ACT RADIO AWAKE 0 SMPL _Send SMPL_LINKID_USER_UUD msg sizeof msg SMPL_Ioctl IOCTL_OBJ_RADIO IOCTL_ACT_RADIO_SLEEP 0 switch ecoEnergyStatus case EP_5 SleepDuration 1 break case EP_4 SleepDuration 2 break case EP_3 SleepDuration 4 break case EP_2 SleepDuration 8 break case EP_1 SleepDuration 16 break case EP MAINS SleepDuration 1 break default SleepDuration 60 break halSleep SleepDuration LISTING 5 9 Main task loop in shared application layer 5 6 Evaluation 5 6 1 Implementing energy adaptive behaviour The process of delivering energy adaptive behaviour is straightforward as shown by the code shown in Listing 5 9 which shows the operational loop that has been implemented on the eZ430 RF2500
129. ents of a prototype to verify this approach are discussed in Section 3 9 The prototype system or case study is described in chapters and 5 3 2 Design for reconfigurability 3 2 1 Plug and play energy subsystem As outlined in Chapter 2 wireless sensor nodes may be powered by a range of energy hardware including energy harvesters and energy storage devices as well as convention ally by primary batteries At present systems are designed for specific energy hardware and are difficult to adapt for different environmental conditions there is also a tendency to over engineer systems to guarantee a certain operational lifetime although this may 59 56 Chapter 3 Development Towards a Reconfigurable Energy Subsystem not actually be realised in practice It is clearly desirable to be able to mix and match the energy hardware on the sensor node dependent on the expected activity of the node and its available environmental energy for example a node may have photovoltaic and thermoelectric energy harvesters a supercapacitor energy store and a primary battery It is also advantageous for the sensor node to be able to adapt its activity to the avail able energy to best exploit resources and achieve continuous operation at the simplest level it may adjust its duty cycle based on the amount of stored energy more complex systems may learn the dynamics of the energy source and adapt to this and in order to do this there must be a
130. er modules produced by the author can be found in Appendix 5 4 Embedded software structure 5 4 1 Overall structure The overall structure of the software on the sensor node was introduced in Section 3 7jand is shown in Figure The focus of the work undertaken and described in this thesis is concerned with the energy subsystem the communication and sensing stacks have only received cursory attention It is left for future investigations to decide the optimal way of interfacing with and processing the data from transducers in the intelligent sensing stack Only a basic implementation of the sensing stack was produced and a pre existing communication stack was used under this project This section describes the overall features of the system and how the system is co ordinated Detailed information on the content and operation of the energy stack is given in Section 5 5 5 4 2 Communication stack The communication stacks used in the prototype systems are inherited from example code provided by Texas Instruments The CC2430EM uses the Simple Packet Protocol which is no longer supported and the eZ430 RF2500 uses the SimpliciTI protocol which is being actively developed for a range of platforms Both protocols are cut down efficient radio stacks which interface with the 2 45GHz radio transceivers on these platforms 134 Chapter 5 Case Study Deployment in a Prototype System Software Shared Application Sensor Evaluation P
131. ergy is from footfalls In the late 1990 s Massachusetts Institute of Technology developed the MIT Shoe 47 aiming to generate power from heel strike energy or the flexing of the shoe Starner and Paradiso present a thorough review of human generated power including heat respiration and blood pressure 48 2 3 7 Inductive and RF energy transfer Transmission of energy via RF or induction is already used by a number of devices including contactless smart cards and RF ID tags 49 The operating range of such devices is typically small of the order of centimetres to metres and the means of powering the device is normally directional Equation gives the power received by a wireless node P ignoring the effects of reflections or interference 30 Here Po is the initial transmitted power A is the wavelength of the transmission and r is the radius of the transmission RF is a common method of distributing power to embedded electronics However Paradiso and Starner state that the potential of RF harvesting technology is limited and that systems powered from an ambient RF source require a large collection area or have to be located close to the radiation source Nevertheless Powercast has recently released products that are designed to harvest energy from transmitted radio waves 51 PX Arr 2 7 Intel Research have demonstrated two systems operating from RF based wireless power transfer The first system known as WISP
132. es Yes Soon Air Flow 380 Yes Yes Soon Vibrations 200 Yes Yes Yes Denotes sources whose fundamental metric is power per square centimetre gt Demonstrated from a 5 C temperature differential Theoretical value assumes air velocity of 5m s and 5 conversion efficiency TABLE 2 5 Comparison of energy harvesting sources for wireless sensor networks adapted from Table shows realised power density for each harvester type In summary the maturity of photovoltaic cell technologies means that they are the natural choice in environments with significant levels of light where fixed frequency vi bration is present vibration energy harvesting is a feasible solution and where significant temperature differences are present these can be exploited by emergent thermoelectric energy generation There are also technologies available which can exploit wind en ergy or the fluid movement The harnessing of human heat and movement is not a well developed area but wireless energy transfer is an interesting area with significant limitations with regard to its effective range and delivered power 32 Chapter 2 Background Energy Sensing and Wireless Communication 2 4 Technologies for intelligent sensing 2 4 1 The IEEE 1451 standards family The IEEE 1451 family of standards define a common communication interface for trans ducers and processors The standards deliver plug and play capabilities to industrial sensors by
133. ets are routed in line with the probabilities sharing the workload between a number of paths A similar scheme is the priority based multi path routing protocol PRIMP developed by Liu et al BI which marks nodes as high priority or low priority dependent on their energy status or on the accumulated number of hops required to route data to the sink This is done at the time of interest propagation where requests for data are sent out and the route for returned data is set up Voigt et al propose a varitation of diffusion which takes the stored energy along with the energy harvested from a photovoltaic module into account when routing packets Perhaps the most intricate energy aware scheme is described by Kansal and Srivastava 83 who look at using information on energy harvesting patterns to predict future energy input and hence make intelligent decisions for routing packets 2 6 2 Energy adaptive behaviour Energy adaptive behaviour or energy aware algorithms are useful as they permit the sensor node to use information on its energy status to adjust its participation in the network generally increasing its activity when energy is plentiful or decreasing it when energy is restricted Communications tasks are normally taken to be the most energy intensive operation for sensor nodes but energy aware algorithms may also control sens ing or processing tasks which also consume large amounts of power Perhaps the simplest form of
134. excessive voltage as driving the device outside its absolute maximum ratings is likely to cause permanent damage It is also essential to ensure that the microcontroller does not attempt to switch on before the output voltage from the circuit has passed the min imum operating voltage for the microcontroller Attempting to start a microcontroller when the voltage is too low will cause the microcontroller to behave unpredictably and experiments carried out under this project indicate it is likely to cause excessive power to be drawn thus meaning that the sensor node is unlikely to start up reliably and will draw excessive amounts of current meaning that the node s energy stores are likely to deplete rapidly 3 6 Energy status determination and algorithms 3 6 1 Overview In order for sensor nodes to react to changes in their stored energy the node must be able to effectively compute its energy status In the scheme proposed in this thesis the node must support a range of energy hardware and use appropriate methods to calculate the level of stored energy It must also present information on the energy status in a standardised way so that the application running on the sensor node can effectively interpret this information This section explores methods for calculating and presenting the energy status of the sensor node including methods for calculating the stored energy in batteries and supercapacitors and uses a discretised energy pr
135. f the embedded software which interfaces with this hardware is documented in Chapter Chapter 5 Case Study Deployment in a Prototype System Software 5 1 Introduction This chapter carries forward the case study outlined in Chapter 4 by describing and evaluating the software interface for the system which is co located with the commu nication stack on the microcontroller This chapter also covers the EEDS content and mechanism for programming the content of data sheets The microcontroller platforms used in this investigation along with their important features of relevance to this work are described in Section 5 2 The implementation of the EEDS for the deployed modules is outlined in Section including examples for the data stored in the implemented modules and detail on the implementation and energy costs of the 1 Wire interface The embedded software structure is described in Section 5 4 and detail on the energy stack is given in Section High level considerations for energy aware operation and the overall evaluation of the system are considered in Section 5 6 5 2 Microcontroller platform 5 2 1 Platform capabilities The software developed under this project has been implemented on two separate micro controller platforms the eZ430 RF2500 and CC2430EM both from Texas Instruments The devices were shown in Section 2 7 1 and the capabilities of each device of interest to this project are described below The CC2430EM
136. f components and their degradation over time may also act to reduce the accuracy of estimates To be used by algorithms such as IDEALS RMR the energy status of the node must be classified into discrete levels as shown in Figure 2 22 Clearly the act of energy monitoring on sensor nodes is potentially a complex task but the aim is to deliver a good estimate of the energy status sufficient for use by energy aware algorithms without using too much energy in performing the associated measurements and calculations 2 6 4 Overall lifetime prediction and extension A major motivating factor behind the development of wireless sensor node technology is that nodes can be deployed for long periods without the need for regular maintenance Clearly the overall lifetime of a sensor node is dictated by the lifetime of its individual components including the power supply Sections and 2 3 have already covered the capacities and other properties of energy stores and the potential power output from energy harvesting devices along with the expected lifetime of primary batteries Many Chapter 2 Background Energy Sensing and Wireless Communication 43 components have a guaranteed lifetime which may be relatively short for example capacitors often have a 1 000 hour lifetime at the extremes of their rated conditions The actual expected lifetime is dependent on a number of factors including temperature and voltage but the expected lifet
137. fetime of 1 000 hours 16 Contin ued developments in supercapacitor technology are improving their characteristics For example the CAP XX range of supercapacitors have thin prismatic packages low self discharge currents and are able to sustain high levels of peak current For example the HS208 model has a typical leakage current of 1 04A after being charged for 72 hours and can tolerate a comparatively high peak current of 20A 2 2 5 Other energy storage mechanisms Some alternative energy storage technologies exist For example fuel cells have been forecast to be a high capacity replacement for batteries however current research indi cates that due to constraints on size and the requirement for precision machining micro fuel cells are unlikely to offer substantially higher energy densities than primary batteries Even where electrodes can be microfabricated the task of microfabricating the fuel Chapter 2 Background Energy Sensing and Wireless Communication 17 reservoir and plumbing is substantial Fuel cells however offer higher power densities and therefore may be of use in sensor nodes with a high current requirement Some research micro fuel cells have been demonstrated with success including a fuel system providing 25mA from a thin film cell with an area of 2cm 19 Alternatively advances in micromachining have opened up the possibility of creating micro heat engines to pro duce electrical power from hydrocarbon f
138. ffectively use a go between circuit board between the energy harvesting storage device and the multiplexer module to deal with power con ditioning enable the management and monitoring functionality and host an electronic memory that can be interrogated by the microcontroller The energy harvesting storage device and this circuit board are together known as an energy module The circuitry on each energy module typically includes low resistance transistor based switches to imple ment the additional functions For example when the microcontroller wishes to measure the nominal power from the photovoltaic module it will send the measurement control line high which will cause the photovoltaic cell s positive terminal to be disconnected from the load the control line will also power an op amp which will buffer the resultant open circuit voltage and allow the nominal power to be estimated through parameters stored in the electronic data sheet on the module Chapter 3 Development Towards a Reconfigurable Energy Subsystem 63 Ultimately the energy aware system developed in this project is reliant on three novel elements which are described in turn in the following sections of this chapter 1 Energy Electronic Data Sheets these store relevant operational information for energy devices in EPROM memory on the energy modules themselves allowing the microcontroller to execute and interpret measurements and manage the mod ule The e
139. fferences are introduced here along with detailed discussions about the realistic levels of power which can be derived and the methods of interfacing with the devices Other less common energy harvesting methods are also considered in this section along with wireless power transfer technologies A number of reviews of energy harvesting techniques have been published with Roundy et al providing one of the most comprehensive surveys Mateu and Moll have also provided a rounded review including some consideration of conversion circuitry design Perhaps one of the most recent and comprehensive reviews of energy harvesting from the motion of humans or machines has been reported by Mitcheson et al 24 2 3 2 Photovoltaics Photovoltaic PV cells produce electricity from photons by means of a semiconductor p n junction This technology is at a relatively advanced stage of development and many deployment situations for autonomous sensors are lit by natural or artificial lighting or a mix of both Table shows typical lighting levels in a range of situations Reich et al have analysed the efficiencies of a range of photovoltaic cells at low illumination levels The results of this investigation are of some interest as they give an idea of the levels of power to be expected from solar cells located indoors in artificially lit environments Figure shows the results of one such survey and indicates that the choice of cell technology is of great
140. ftware 125 e halSetPortPinOutput which sets the actual output i e high or low of the general purpose I O pin specified The port pin and output state are sent to the function as unsigned char variables e halReadPortPin which reads the digital input to the general purpose I O pin specified the port and pin are specified as unsigned char variables and the result is also returned as an unsigned char e halEraseFlash which is used to erase the blocks of flash memory that are used to store the microcontroller s copy of the energy module EEDS e halInitialiseEEDSPointers which is used to initialise pointers to the memory locations for the energy electronic data sheet of each energy module ordered by multiplexer address number The following macros are also provided which configure the sleep timer of the micro controller and allow it to be put into a low power mode e halInitSleepTimer which sets up the sleep timer facility setting up the clock divider and oscillator input appropriately e halShortSleep which sleeps the microcontroller for a short time generally used to allow the ADC reading to settle e halLongSleep which sleeps the microcontroller for a longer period generally of the order of a few seconds e halSleep which takes an unsigned char as an input this is used for duty cycling the node and controls the duration of the sleep The specified value controls the interval of the sleep timer in
141. g General principles Vibration energy harvesting is a technology which allows electrical energy to be parasit ically generated from generally unwanted vibrations normally of machinery to power electronic devices such as wireless sensors Vibration energy harvesters have the advan tage that unlike photovoltaics they are not subject to the effects of dust accumulation Generators are typically one of three types which are discussed later in this subsection electromagnetic piezoelectric and electrostatic Table 2 4 shows the results of a survey carried out in the USA which shows the peak acceleration and frequency of vibration of a range of objects it may be noted that as the mains frequency in the USA is 60Hz a number of electrical devices in this table exhibit twice mains frequency vibrations Chapter 2 Background Energy Sensing and Wireless Communication 21 Vibration Source Peak Acc Frequency of Peak Hz Base of 5 HP 3 axis machine tool with 36 bed Kitchen blender casing Clothes dryer Door frame just after door closes Small microwave oven HVAC vents in office building Wooden deck with people walking Breadmaker External windows size 2 ft X 3 ft next to a busy street Notebook computer while CD is being read Washing Machine Second story floor of a wood frame office building Refrigerator TABLE 2 4 List of vibration sources with their maximum accelera
142. g devices depending on the amount of energy being harvested in order to reduce battery usage AmbiMax 46 a further development by the same team is a notable example which combines energy harvesting from wind and light and stores it in supercapacitors and lithium rechargeable batteries An advantage of the AmbiMax power module is that it is entirely analogue and autonomous however the system design must be adapted to accommodate changes of energy resource Furthermore the sensor node powered by the module has no means of finding out the levels of production or availability of energy as the output voltage of the module is fixed at 4 1V The AmbiMax system is shown in Figure Reservoir Capacitor Array Subsystem Energy Harvesting F ES Control and Charger Subsystem Subsystem III gt 1 Energy Harvesting I Window Subsystem Il 4 1 Comparator ue ja a s el Super SS Reservoir d A R ti gt hd Capacitors 1 Energy Harvestillg 1 Current r D a 1 Subsystem 1 l Limit on i H Switch i 1 Boost l Super H gt ME Li Polymer Control charger 1 ui Regulator I Capacitor Battery igh La 1i E Window Se A ti T Compartor y Battery Er ul Ambient H Power IG e e 1 i Source d Hysteresis i Comparator 1 p Sensor H A Architecture B Circuit FIGURE 2 30 The Ambima
143. g raw readings values can be made available to the EAN layer 3 Controlling outputs For controlling the performance of the energy subsystem For example setting pins high or low to control switching of transistors to isolate energy sources for performance measurement Chapter 3 Development Towards a Reconfigurable Energy Subsystem 85 The layer acts to hide the complexities of interfacing with the hardware of the energy subsystem from the EAN layer This layer isolates the rest of the system from the intricacies of interfacing with the actual energy hardware it is deployed on Again the aim is to present a common interface to the EAN layer regardless of the deployed hardware 3 7 3 The Sensing Stack The design of the intelligent sensing stack shown in Section 3 11 is similar to that of the energy management stack The software tasks for sensor interfacing are modularised and arranged into the sensing stack comprised of the layers as described here e Sensor Evaluation SEV takes a high level view of the sensing devices on the system responding to queries for sensor readings and presenting important information about sensors or the sensed data This layer may also perform trend detection and implement the packet priority features of IDEALS RMR e Sensor Processing SPR uses device models to provide an adjusted sensor reading taking account of linearisation and offset adjustments and providing error
144. g seen as over restrictive and complex 103 Despite its limitations no other operating systems have achieved the prominence of TinyOS 48 Chapter 2 Background Energy Sensing and Wireless Communication EMA r 1 1 Enforcement 4 Mechanisms 1 gt 2 D a E oO bd gt Specification Application FIGURE 2 25 Energy management architecture reproduced from Jiang et al TinyOS 2 0 based on a new hardware abstraction architecture outlined by Hanziski et al and shown in Figure is now being used by the community The archi tecture effectively divorces the application running on the sensor node from the detail of interfacing with its hardware thus permitting applications to be used on a range of platforms with the appropriate interfaces being provided by Hardware Interface Layer HIL Hardware Adaptation Layer HAL and Hardware Presentation Layer HPL for each piece of hardware implemented in this scheme Cross platform applications Platform specific Platform specific applications applications Platform independent hardware interface Ve BON fa 7 e SC 7 D Wie D Ce S m C m2 Q HIL3 D HIL4 O lt y po O HAL 2 fa gt A N HAL 1 C HAL 3 Ve J a L see oui HPL 2 ua IL ti C gz HW SW 7 J N J N boundary eege R E E EE HW Pla
145. ge V 4 1 Log Datasheet Voc 4 4 T T T T 1 0 200 400 600 800 1000 Tluminance lux FIGURE 4 13 Logarithmic fit of V against illuminance for Schott Solar OEM 1116929 Indoor PV module This equation can be rearranged as shown in Equation to determine the level of illuminance from Voc Voc B Ey e 4 7 A useful property of silicon PV cells is that their MPP voltage Vmpp is related to their Voc by parameter k This parameter is typically between 0 70 and 0 80 Equation 4 8 and can be determined for individual modules 126 This property can be used by simpler maximum power point tracking circuits Furthermore the current obtained from the PV module at its MPP Impp is related to its illuminance by parameter m as shown in Equation 4 9 Vmpp KVoc 4 8 Impp mEy 4 9 Parameter Description Value k Ratio between Voc and Vmpp 0 71 m Ratio between Ey and Impp 5 4 x 1077 A Natural log fit of Voe Ey 0 2571 B Natural log fit of Voe Ey 2 9128 TABLE 4 4 Parameters found for Schott Solar OEM 1116929 Indoor PV Module There are certain limitations to this model in that the parameter k only holds if the module is under uniform illumination i e no shadowing Secondly parameter m holds only for relatively low levels of current While it varies by less than 1 in the range 100 1 000 lux it can be expected to vary more widely in brighter deployment envir
146. ge depression These requirements rule out some common battery types such as zinc carbon and smaller lithium coin cells from use in this application Zinc Alkaline Lithium Lithium Carbon Manganese Thionyl Dioxide Chloride Specific energy Wh kg 85 145 380 230 Energy density Wh L 165 400 715 535 Nominal voltage V 1 5 1 5 3 6 3 0 Open circuit voltage V 1 6 1 5 1 6 3 65 3 3 Midpoint voltage NI 1 25 1 1 1 25 1 15 3 6 3 3 3 0 2 7 End voltage V 0 9 0 9 3 0 2 0 Self discharge year 7 4 1 2 1 2 Operating T C 10 to 50 20 to 55 60 to 85 20 to 55 Discharge profile Sloping Mod slope Flat Flat Status High produc Most popular Mainly for Increasing tion decreas primary bat special appli consumer ing popularity tery cations production Advantages Low cost High capac High E den High E den ity good low sity long shelf sity good low T low rate life T high rate performance Limitations High gassing Moderate cost Voltage delay Limited sizes rate poor but best at after storage and shipping performance high rates a At 20 C P For cylindrical cell For favourable discharge conditions d Normally decreases with time of storage TABLE 2 1 Non rechargeable battery types and characteristics adapted from 8 One of the longest documented deployments of a wireless sensor an Automatic Meter Reading AMR system for a
147. good interface between the embedded software and the energy hardware While a limited number of deployments have already employed multiple energy har vesting devices or hybrid energy storage and management systems these were not truly flexible as they were tailored for specific energy hardware and could not be configured in situ Additionally the energy awareness capability of these existing systems was highly restricted in many cases with the system only being able to monitor the voltage across its energy storage device s in order to assess the power status of the node Because of this restriction systems of this type are unable to detect the poor performance or possible malfunction of energy harvesting devices or observe the dynamics of energy generation Indeed the only energy metric they typically use for deciding on their ac tivity level is the amount of stored energy These limitations are potentially serious for the end users who would be deploying and using these devices for the following reasons 1 The ability of present systems to accommodate only a limited set of energy harvesters or storage devices limits their deployments to only those areas that feature suitable environmental conditions for that set of devices To achieve energy awareness with changeable hardware and even support the use of primary batteries in certain conditions the hardware must be much more flexibly designed than is the case at present Ultimately
148. gy being lost or of components being damaged The power consumption of taking measurements is also important the operational am plifiers used in this project draw a quiescent current of around 1 24A however it is likely that the effective energy cost of measurement operations will be many times this figure For example for the test of primary cells proposed earlier closed circuit current measurements with a small impedance actually discharge the cell at each test In a similar way for many energy harvesting devices they must be disconnected from their load in order for the measurement to be taken for a device generating 1mW the system will miss out on 1mW of power for the duration of the test Furthermore when the microcontroller is engaged in taking measurements it is likely to be in an active mode with its ADC enabled once again this is a substantial consumer of energy and must be optimised in order to avoid excessive usage of energy The following chapter explores ways for the system to self manage its energy resources while using as little energy as possible in achieving this objective The decision to use linear regulators was made in order to keep the complexity of the energy hardware low it was also to keep the quiescent current draw of the circuit to a minimum While this has indeed been the case it limits the efficiency of the system 120 Chapter 4 Case Study Deployment in a Prototype System Hardware when the ra
149. h Wireless sensor nodes must be capable of operating autonomously and for many this means they must operate without the constraint of a wired power supply Convention ally such devices have been powered by non rechargeable batteries which are replaced when depleted However energy harvesting also known as energy scavenging tech nology now offers the potential to sustain the operation of sensor nodes indefinitely In this process environmental energy is converted into electrical energy which is then used to directly power the sensor node or is buffered in rechargeable batteries or su percapacitors for use later Recent progress in the development of energy harvesting technologies harvesting electrical energy from such sources as light vibration or tem perature difference now permits sensors to be free from the limitations of operation from non rechargeable batteries Indeed the lifetime of the node will ultimately be limited by its other components such as sensors or flash memory To deliver a long operational lifetime nodes must ensure that they use energy carefully Non rechargeable batteries store a finite amount of energy which cannot be replaced so nodes must operate efficiently in order to achieve their designed lifetime Similarly where nodes are powered from harvested energy they must on average use less power than is generated in order to deliver sustained operation Energy is arguably the most critical resource for wire
150. h Annual IEEE Communications Society Conference on Sensor Mesh and Ad Hoc Communica tions and Networks SECON 2009 22 26 June 2009 Rome Italy This publication is not available in the online version of this thesis but may be down loaded from http eprints ecs soton ac uk 17325 172 Appendix C Selected Publications Weddell A S Grabham N J Harris N R and White N M 2009 Modular Plug and Play Power Resources for Energy Aware Wireless Sensor Nodes In Sixth Annual IEEE Communications Society Conference on Sensor Mesh and Ad Hoc Communica tions and Networks SECON 2009 22 26 June 2009 Rome Italy This publication is not available in the online version of this thesis but may be down loaded from http eprints ecs soton ac uk 17325 Appendix C Selected Publications 173 Weddell A S Grabham N J Harris N R and White N M 2009 Modular Plug and Play Power Resources for Energy Aware Wireless Sensor Nodes In Sixth Annual IEEE Communications Society Conference on Sensor Mesh and Ad Hoc Communica tions and Networks SECON 2009 22 26 June 2009 Rome Italy This publication is not available in the online version of this thesis but may be down loaded from http eprints ecs soton ac uk 17325 174 Appendix C Selected Publications Weddell A S Grabham N J Harris N R and White N M 2009 Modular Plug and Play Power Resources for Energy Aware Wireless Sensor Nodes In S
151. he case of their example for a rocket testing facility Schmalzel et al extend the concept to the health electronic data sheet HEDS and further to xEDS meaning that the electronic data sheet concept may be applied to other devices and applications 2 4 4 System management schemes Two major standards enabling intelligent system management have been developed The first entitled System Management Bus or SMBus was initially defined by Intel in 1995 with the purpose of enabling system management in PCs and servers Devices such as temperature sensors fans and voltage sensors could be monitored and managed over a standard interface which was based on the DC protocol The actual commands used over the interface were to be defined by hardware manufacturers and are not specified in the standards The standard proved particularly popular for batteries especially laptop batteries for which an add on standard was developed by Duracell and Intel defining Smart Battery Data SBD 63 The second major standard to be developed is entitled Power Management Bus PMBus 64 This standard defines the format of Chapter 2 Background Energy Sensing and Wireless Communication 35 data exchanges between power related devices for example the READ_VCAP command returns the voltage on the energy storage capacitor These standards are best suited to higher power complex computing systems and are perhaps less suited to res
152. he connection of 1 wire devices through a multiplexer meaning that EEDS can be addressed on a module by module basis rather than using the ID of the 1 wire device on a shared bus The two interface schemes 66 Chapter 3 Development Towards a Reconfigurable Energy Subsystem are shown in Figure Further detail on the hardware interface implemented in this system is given in Section 3 4 1 wire 1 wire 1 wire 1 wire 1 wire 1 wire 1 wire Device Device Device Device Device Device Device Vcc O A Multi drop bus configuration Vcc O SE Multiplexer S ES 1 wire 1 wire 1 wire 1 wire 1 wire 1 wire 1 wire Device Device Device Device Device Device Device B Multiplexed bus configuration FIGURE 3 2 Conventional and multiplexed 1 wire bus configurations 3 4 Common Hardware Interface CHI 3 4 1 Overview and justification Microcontrollers have a limited number of I O pins available especially those that are connected to internal ADCs and it is necessary to conserve these as far as possible Furthermore within the scheme of EEDS equipped modules it is important that the microcontroller is able to access each module independently firstly in order to determine its data sheet parameters and secondly to perform measurement and control operations In the scenario outlined earlier it would be unrealistic to expect that each module would have multiple direct connections to I O pins of the microcontroller For t
153. he state of the art in self powered wireless 54 Chapter 2 Background Energy Sensing and Wireless Communication Ee i Ee ime Eo Hydrogenerator O EC O H External low power cH4 general purpose Power lo me ef Wind generator O conditioning 2 08 Solar Panel d subsystem ka Wireless ee O os a 103 104 O CH MPWiNodeX Client o Energy 3 3 5 ON OFF Stor E SC EE Sensor Power EES Battery D MAX710 Maxim MPWiNodeX platform A System Architecture B Deploy ment FIGURE 2 31 The MPWiNodeX architecture and deployment reproduced from 45 sensor nodes does not allow the energy subsystem to be reconfigured and has limited energy awareness or energy management functionality The scheme proposed in Chap ter Blintroduces methods to overcome these limitations 2 10 Summary This chapter provided an overview of energy technologies and other background related to sensing and wireless communications Energy harvesting technologies such as photo voltaics thermoelectrics and small scale fluid flow have been used in this project and are discussed in later chapters Energy storage devices including supercapacitors and primary and rechargeable batteries are also used The plug and play architecture for the energy hardware of sensor nodes developed under this project draws inspiration from the TEDS standard The prototype system delivers energy awareness to enable adaptive operation wh
154. hin film technology combined with micro system tech nology This has been commercialised by Micropelt which has developed a prototype thermoelectric generator integrated into an M24 bolt as shown in Figure 2 12 It may be observed that the power obtained from the device is highly dependent on the airflow it is exposed to A separate development by Tellurex has produced a device which al lows burning fuels to be converted into electrical energy by means of the their PG1 kit shown in Figure 2 13 This device has a heatsink with fan the power provided by the thermoelectric generator is sufficient to run the fan which forces airflow through the heatsink thus improving the overall efficiency of the generator and increasing the level of generated electrical power AT chuckT air speed Yio pS SS E voltage matched load matched power gt E v oD L S z o gt a 0 5 10 15 20 25 30 35 40 time min A Bolt B Performance FIGURE 2 12 The Micropelt Generic Power Bolt based on an M24 machine bolt and its performance in natural and forced convection with air speeds of 3ms 5ms and l0ms demonstrating the effect of movement on efficiency of the harvester repro duced from 43 Chapter 2 Background Energy Sensing and Wireless Communication 27 a 1 0 0 30 Gr He 25 83 4mm 0 9 L a mg el eae 0 8 CH SE A E S
155. his project show that when exposed to varying volt ages a CC2430EM module under a continuous sense transmit cycle behaves more closely as a current dominant load than resistive or power dominant A graph of the current and power consumption of the device is shown in Figure 3 6 The voltage fraction gives an indication of the state of charge of the capacitor assuming it is discharged by a cur rent which is independent of the store voltage or energy and hence I CdV dT used in Equation 78 Chapter 3 Development Towards a Reconfigurable Energy Subsystem V min V max V min V oe Pt PV Where V is the store voltage at a given time min V and max V are the minimum and maximum store voltages Finally the logarithmic discharge fraction exploits the fact that the voltage across a capacitor obeys the relationship V Voe C for a constant resistance and capacitance where Vo is the initial voltage t is time elapsed and C and R are capacitance and resistance respectively This can be rearranged to find t as shown in Equation 3 3 V In 3 3 t CR 2 3 3 Thus to find the logarithmic discharge fraction the result Equation D A is independent of t C and R V n max V min V In max V pi b s 3 4 This quantity is the most computationally expensive to calculate with results for the actual time taken to carry out a complete calculation given in Section 5 5 2 It is likely to give a usef
156. his reason it is necessary to multiplex parts of the system to make best use of resources and allow queries to be effectively executed A multiplexer module is employed in this scheme that enables five signals one 1 wire EEDS line and four other general I O including one analogue signal line to be routed to each module from the microcontroller Up to six individual energy modules are supported Chapter 3 Development Towards a Reconfigurable Energy Subsystem 67 3 4 2 EEDS interface A drawback of 1 wire devices is that they draw current when connected to the supply even when not actively communicating with the host In order to isolate all 1 wire de vices the first multiplexed channel address 0 is used for ancillary communication be tween the microcontroller and the multiplexer module without a 1 wire device installed on this channel The microcontroller leaves the address lines in this state when not actively communicating with the energy modules thus minimising the power consumed by the 1 wire bus and this permits the multiplexer module to use the measurement line to trigger an interrupt on the microcontroller e g it can be used to indicate that a volt age threshold has been passed The maximum number of energy modules supported by this scheme is six three address lines enable eight channels two of which are used for communication with the multiplexer module The allocation of two addresses to the multiplexer module
157. http eprints ecs soton ac uk 12955 September 2006 Last accessed May 2010 X Jiang J Taneja J Ortiz A Tavakoli P Dutta J Jeong D Culler P Levis and S Shenker An architecture for energy management in wireless sensor net works ACM SIGBED Review 4 3 31 36 2007 D Gay M Welsh P Levis E Brewer R v Behren and D Culler The nesC language A holistic approach to networked embedded systems Proceedings of the ACM SIGPLAN Conference on Programming Language Design and Implementa tion PLDI pages 1 11 2003 D K Arvind M Adams A Burdett T Dillon P Garner J Gilby G Matich and G Peggs Wireless sensor networks a mission to the USA Report of a DTI Global Watch Mission DTI Global Watch November 2005 V Handziski J Polastre J H Hauer C Sharpt A Wolisz and D Culler Flex ible hardware abstraction for wireless sensor networks Proceedings of the Second European Workshop on Wireless Sensor Networks EWSN 2005 2005 145 157 2005 R Szewczyk A Mainwaring J Polastre J Anderson and D Culler An anal ysis of a large scale habitat monitoring application Proceedings of the Second International Conference on Embedded Networked Sensor Systems pages 214 26 2004 Pei Zhang Christopher M Sadler Stephen A Lyon and Margaret Martonosi Hardware design experiences in ZebraNet SenSys 04 Proceedings of the Second International Conference on Embedded Networked Sensor Systems page
158. ich is almost always radio frequency RF based and a power supply conventionally a battery but potentially harnessing energy harvested from the environment A typical architecture for a sensor node is shown in Figure 1 1 The applications of this technology are extremely varied but key drivers for the popu larity of wireless sensors are the high cost or the impracticality of installing wiring for Comms Processor Interface MCU Sensor Memory Interface 0 8 ADC Management Sensing Hardware Energy Resources FIGURE 1 1 The major parts of a wireless sensor node 1 2 Chapter 1 Introduction conventional sensors Wireless technology allows parameters to be monitored or exper iments to be carried out which would previously have been prohibitively expensive or time consuming Furthermore in contrast to conventional data logging equipment wire less sensors on remote long term deployments permit parameters to be monitored with a high temporal resolution and provide a real time or near real time representation of the situation Wireless sensing technology enables devices to be deployed quickly and easily by way of the fact that they are able to self configure without the need for a fixed layout or infrastructure State of the art communication protocols allow sensor nodes to form mesh networks and work together to route information to a central point Normally nodes operate
159. ics Smart Materials and Structures 13 5 1131 1142 2004 AdaptivEnergy Joule Thief Modules http www adaptivenergy com application 20chart index html 2009 Last accessed May 2010 Mid Technology Corporation Piezo energy harvester catalog http www mide com products volture volture_catalog php 2010 Last accessed May 2010 S Meninger J O Mur Miranda R Amirtharajah A Chandrakasan and J H Lang Vibration to electric energy conversion IEEE Transactions on Very Large Scale Integration VLSI Systems 9 1 64 76 2001 J P Fleurial G J Snyder J A Herman M Smart P Shakkottai P H Giauque and M A Nicolet Miniaturized thermoelectric power sources In 34th Intersociety Energy Conversion Engineering Conference Vancouver BC Canada 1999 H Bottner J Nurnus A Gavrikov G Kuhner M Jagle C Kunzel D Eber hard G Plescher A Schubert and K H Schlereth New thermoelectric compo nents using microsystem technologies Journal of Microelectromechanical Systems 13 3 414 20 2004 Micropelt GmbH MPG D602 D751 thin film thermogenerator sens ing devices http www micropelt com down datasheet_mpg_d602_d751 paf March 2008 Last accessed May 2010 Tellurex Corporation PG 1 Product Details http www tellurex com pdf PG1_spec_sheet pdf January 2009 Last accessed May 2010 R Morais S G Matos M A Fernandes A L G Valente S F S P Soares P J S G Ferreira and M
160. ile remaining hardware agnostic The prototype system trans mits data wirelessly and has been developed for the CC2430 and MSP430 platforms but is written in C so could be used in a range of microcontrollers The limitations of existing sensor nodes which incorporate energy harvesting or even a range of en ergy devices have also been discussed and these shortcomings are addressed by the plug and play system the architecture of which is discussed in detail in Chapter 3 Chapter 3 Development Towards a Reconfigurable Energy Subsystem 3 1 Introduction This chapter defines the interfaces and design for the reconfigurable system architecture outlining many of the key contributions of this work The overall structure of the system is described in Section with a discussion of how the individual aspects fit together to deliver a complete system architecture The new aspects proposed include energy electronic data sheets EEDS which are discussed in Section and a common hardware interface CHI between the energy modules which is defined in Section 3 4 A generalised specification for the hardware of each energy module can be found in Section 3 5 and methods and algorithms for determination of the energy status of the node may be found in Section 3 6 The novel software structure which has been utilised in this project is defined in Section 3 7 The general operation of the system is covered in Section 3 8 and the requirem
161. ime may be many times the guaranteed lifetime Ideally for an energy aware sensor node the gradual degradation of the energy hardware would be compensated for and the node should adapt its operation accordingly Taking an example a Panasonic Gold 1F HW series supercapacitor 16 which has a 1 000 hour guaranteed lifetime at its maximum voltage and temperature being 2 3V and 70 C at the end of the guaranteed lifetime their capacitance can be expected to have dropped by no more than 30 and their internal resistance to have increased by no more than four times The expected lifetime is not a guaranteed value but can be estimated by Equation 2 8 provided by Panasonic Where te is the expected lifetime ty is the guaranteed lifetime a is the temperature factor and ay is the voltage factor Considering the use of the device for a sensor node operating at between 2 0 and 3 6 volts 3 6 volt maximum at a temperature of 30 C and with a maximum current draw of 25mA the expected lifetime of the supercapacitor can be calculated Given that the node is operating at up to 3 6 volts and each device has a rated maximum voltage of 2 3V it will be necessary to connect two supercapacitors in series This results in an effective capacitance of 0 5F and a maximum voltage across each device of 1 8 volts te ty X Ar X Oy 2 8 Panasonic state that the temperature factor a is expressed by Equation which expresses the fact that the life doubles f
162. introducing a modular stacking architecture Other proposed work includes the further development of the embedded software to better define the sensing stack and improve the interconnection between the three separate stacks It may also be interesting to look at the potential for taking this technology towards standardisation most probably through an industry organisation 6 3 2 Hardware development As mentioned in Section 4 3 2 it is anticipated that future iterations of the device design will feature a stacking architecture It is envisaged that a stacking 36 way or greater connector could be used and that each stacking board will act as a pass through for data signals to the board below In order for this to be realised the physical size particularly the height of each module s PCB must be reduced The reduced form factor and improved integration would be essential for deployment in an end product The physical packaging of the device would also have to be explored At the moment each energy device requires its own power conditioning and interface circuitry the in terface circuitry acts as a go between from the energy device to the rest of the system This may be viewed as a strength of the developed architecture in that the system de signer is free to select an appropriate power conditioning device to interface between the energy device and the rest of the system however in the developed system particularly for the energy ha
163. ion of energy consumption across all nodes on the network Many protocols aim to extend the useful life of the network without compromising on the delivery of data Often there is a trade off between latency and power consumption Al Karaki and Kamal have produced an excellent survey of routing techniques for wireless sensor networks The authors of the survey categorise the routing techniques as either flat hierarchical or location based The protocols are also classified as multipath based query based negotiation based and QoS based They discuss the main challenges and design issues including node deployment schemes data reporting methods time driven event driven query driven or a hybrid and quality of service and scalability issues Routing is a notoriously complex area of research and favourable results obtained through simulation are rarely realised in the field It is not possible to go into detail here about the different schemes but they form the basis of energy adaptive algorithms described in Section 2 6 2 5 5 Discussion The field of wireless communications for sensor networks is wide and growing with an overwhelming number of publications in the field many of which are of dubious real value Wireless communication technology is not a major focus of this thesis but is instead an enabler to demonstrate the concept of reconfigurable energy aware wireless sensor nodes A small number of technologies IEEE 802 15 4 ZigBee Bl
164. iority scheme for representing the levels of stored energy 74 Chapter 3 Development Towards a Reconfigurable Energy Subsystem 3 6 2 Energy monitoring It is important for energy aware systems to have an indication of the state of charge of energy stores in order that systems can adjust their behaviour dependent on their energy status There are a number of different ways of calculating the energy status value and up to now most methods have represented the stored energy or voltage as a percentage Rather than simply an indication of how full the energy store is though it may be more helpful to get an idea of how long a node can operate for compared to its neighbours To represent this the author proposes the introduction of ve the remaining lifetime fraction compared to when the energy store is full A value of 0 means that the node has insufficient energy to remain operational and a value of 100 means that the energy store is full Values between 0 and 100 represent normal operation of the node with 50 meaning that the node s energy store is half way through from a time perspective being discharged Figure shows an example of the discharge profile of an ideal 1F supercapacitor through a 4500 resistor 70mA current load or 200mW power load These values were chosen as they draw similar levels of current at 3V The simulation was started at 3 0V as this is a typical maximum voltage used to power microcontrol
165. iplied by 10 to be stored as an unsigned char The NumPorts field states how many modules the multiplexer module can support up to six which enables the scheme to be used for smaller multiplexer modules which support fewer energy mod ules The MaxVoltage and MinVoltage fields state the maximum and minimum raw voltage that the multiplexer module can support while remaining functional and the RegulatedVoltage field represents that the multiplexer will regulate to These voltage values are scaled up by a factor of ten as with the other energy modules so that the values can be stored as an unsigned char This minimises the amount of memory re Chapter 5 Case Study Deployment in a Prototype System Software 131 typedef struct _TMuxData TDeviceID DevicelD unsigned char Multiplier unsigned char NumPorts unsigned char MaxVoltage unsigned char MinVoltage unsigned char RegulatedVoltage TMuxData LISTING 5 3 Struct for multiplexer EEDS parameters typedef union _TParams TParamCurve ParamCurve TParamComplex ParamComplex TUInt IntMultiplier TPrimaryParams PrimaryParams TParams LISTING 5 4 TParams union for storing module parameters quired to stored these parameters and the amount of energy required to read this data from the 1 Wire EPROM memory The parameters stored for the energy module are represented in the Parameters field of the TModule struct The parameter fields are shown i
166. is aware of the true impact of energy usage Clearly the discharge characteristics of different battery chemistries are also variable and must be taken into account when measuring energy for batteries subjecting cells to a large known load and comparing this to the discharge curve is a common way to assess their state of charge Sai In a similar way to estimating the stored energy the rate of power generation from energy sources must also be determined For a mains adapter the rate of generation is typically rapid and hence need not be computed the system may merely look at whether such resources are on or off however for devices such as vibration energy harvesters and photovoltaic modules the rate of power generation is variable and must be more carefully computed In general the nominal power of energy harvesting devices may be estimated by analysing the open circuit voltage of the harvesting device or by putting its output through a fixed load and monitoring the operating voltage The measured data are then combined with device models to determine the rate of energy generation Delivering energy awareness for resource constrained embedded systems is a non trivial task and is highly dependent on the use of device models Clearly such models must be highly simplified to reduce their memory requirement and computational complexity but must be of a sufficient quality to ensure their accuracy Other effects such as the tolerance o
167. ish Columbia Canada This publication is not available in the online version of this thesis but may be down loaded from http eprints ecs soton ac uk 15361 Bibliography 1 C Kompis and S Aliwell editors Energy Harvesting Technologies to En able Wireless and Remote Sensing Sensors amp Instrumentation KTN Action Group Report June 2008 http server quid5 net koumpis pubs pdf energyharvesting08 pdf Last accessed May 2010 2 K Martinez P Padhy A Elsaify G Zou A Riddoch J K Hart and H L R Ong Deploying a sensor network in an extreme environment Proceedings of Sensor Networks Ubiquitous and Trustworthy Computing pages 186 93 2006 3 G Werner Allen K Lorincz M Welsh O Marcillo J Johnson M Ruiz and J Lees Deploying a wireless sensor network on an active volcano EEE Internet Computing 10 2 18 25 2006 4 F M Discenzo D Chung and K A Loparo Pump condition monitoring using self powered wireless sensors Sound and Vibration pages 12 15 May 2006 5 J Polastre R Szewczyk and D Culler Telos enabling ultra low power wireless research 2005 Fourth International Symposium on Information Processing in Sensor Networks pages 364 9 2005 6 S George Development of a vibration powered wireless temperature sensor and accelerometer for health monitoring Aerospace Conference 2006 IEEE 2006 8 pp 2006 7 P Dutta J Hui J Jeong S Kim C Sharp J Taneja
168. itors to permit nodes to draw large bursts of power during radio transmis sions and sensing operations Recently GE Energy have incorporated vibration energy harvesters in their Insight mesh system which monitors oil refinery machinery 115 52 Chapter 2 Background Energy Sensing and Wireless Communication Very few sensor node deployments based on thermoelectric energy harvesting have been reported in the literature Typically systems have depended on a very large temperature gradient being present in order to deliver sufficient levels of energy for example an investigation successfully deployed a wireless sensor network on aluminium smelters in a factory 116 Further developments in thermoelectric energy harvesting devices have substantially increased their efficiencies EnOcean have demonstrated the ECT100 thermal energy harvester kit which incorporates a thermoelectric energy harvester and radio transmitter module 117 Most of the literature on solar energy harvesting focusses on outdoor deployments where light is plentiful and energy need not be carefully managed A small number of papers have been published describing the development of systems to harvest energy from indoor low intensity lighting Perhaps the most tenuous use of indoor solar energy harvesting is presented by Hande et al in which energy is harvested from a ceiling mounted fluorescent lighting unit in order to power a wireless routing node As shown in
169. ittle additional circuitry required above that already used on sensor nodes it is just distributed to the energy devices rather than being integrated onto the main circuit board of the sensor node An example of a typical configuration for the developed system is shown in Figure Here the energy subsystem is comprised of three separate energy harvesters vibration photovoltaic and thermoelectric with energy being buffered in a supercapacitor and optionally in a NiMH secondary battery A lithium primary battery provides an emer gency backup facility The lines between modules denote physical wires and in the prototype developed in the case study in Chapter 4 and Chapter 5 form cables that physically plug into sockets on the multiplexer module The actual socket that each energy module is connected to is unimportant as this is a flexible plug and play system the power conditioning and switching circuitry is on the modules themselves therefore little additional hardware is needed on the multiplexer module The microcontroller interfaces with the energy subsystem as shown and has a separate interface with its sensing hardware and the antenna as is conventional for wireless sensor nodes It should be noted that the default behaviour of modules in the energy subsystem is to harvest energy to charge up the rechargeable energy stores and to provide power to the microcontroller The system supports non renewable energy resources s
170. ixth Annual IEEE Communications Society Conference on Sensor Mesh and Ad Hoc Communica tions and Networks SECON 2009 22 26 June 2009 Rome Italy This publication is not available in the online version of this thesis but may be down loaded from http eprints ecs soton ac uk 17325 Appendix C Selected Publications 175 Weddell A S Grabham N J Harris N R and White N M 2009 Modular Plug and Play Power Resources for Energy Aware Wireless Sensor Nodes In Sixth Annual IEEE Communications Society Conference on Sensor Mesh and Ad Hoc Communica tions and Networks SECON 2009 22 26 June 2009 Rome Italy This publication is not available in the online version of this thesis but may be down loaded from http eprints ecs soton ac uk 17325 Appendix C Selected Publications 177 Weddell A S Merrett G V Harris N R and Al Hashimi B M 2008 Energy Har vesting and Management for Wireless Autonomous Sensors Measurement Control 41 A This publication is not available in the online version of this thesis but may be down loaded from http eprints ecs soton ac uk 15342 178 Appendix C Selected Publications Weddell A S Merrett G V Harris N R and Al Hashimi B M 2008 Energy Har vesting and Management for Wireless Autonomous Sensors Measurement Control 41 4 This publication is not available in the online version of this thesis but may be down loaded from ht
171. k 15342 Appendix C Selected Publications 183 Weddell A S Merrett G V Harris N R and Al Hashimi B M 2008 Energy Har vesting and Management for Wireless Autonomous Sensors Measurement Control 41 A This publication is not available in the online version of this thesis but may be down loaded from http eprints ecs soton ac uk 15342 Appendix C Selected Publications 185 Weddell A S Harris N R and White N M 2008 Alternative Energy Sources for Sensor Nodes Rationalized Design for Long Term Deployment International Instru mentation and Measurement Technology Conference May 12 15 2008 Victoria British Columbia Canada This publication is not available in the online version of this thesis but may be down loaded from http eprints ecs soton ac uk 15361 186 Appendix C Selected Publications Weddell A S Harris N R and White N M 2008 Alternative Energy Sources for Sensor Nodes Rationalized Design for Long Term Deployment International Instru mentation and Measurement Technology Conference May 12 15 2008 Victoria British Columbia Canada This publication is not available in the online version of this thesis but may be down loaded from http eprints ecs soton ac uk 15361 Appendix C Selected Publications 187 Weddell A S Harris N R and White N M 2008 Alternative Energy Sources for Sensor Nodes Rationalized Design for Long Term Deployment I
172. ld generate milliwatts of power from sub stantial air flows of several metres per second The device provided an AC output which had to be rectified to DC for use by the sensor node These modules were then adapted to interface with the plug and play system developed in the course of this work The functionality and connections for these modules are described in Section 4 5 3 Mains electricity In situations where mains or any inexhaustible supply of electricity is available this can be exploited by the mains electricity module Mains power must first be regulated to 5 11 5V DC by a mains power adapter before being fed into the module which will then regulate the power supply for the sensor node The use of an off the shelf power adapter was considered essential for safety reasons The mains electricity module is described in Section 4 5 4 94 Chapter 4 Case Study Deployment in a Prototype System Hardware 4 2 3 Utilised energy stores Primary batteries Typically conventional nodes are run from primary batteries which are generally either alkaline or lithium chemistries For the purposes of this investigation a module was developed to accommodate alkaline with two cells connected in series or lithium thionyl chloride standalone cells The AAA alkaline cells had a nominal 1 5V operating voltage and 1200mAh capacity and 1 2AA lithium cells had a nominal 3V operating voltage and 950mAh capacity The module will also accommodate
173. lectronic data sheets have a set format which is defined in Section 3 3 2 Common Hardware Interface this is the interface between the energy mod ules the multiplexer module and the microcontroller It allows the microcontroller to obtain information about its energy hardware and act to monitor and manage these resources The format of the common hardware interface is defined in Sec tion 3 Embedded Software Structure provides a stack structure which is similar to the communications stack that enables the application on the microcontroller to interface with the energy hardware The format of the software structure is described in Section 3 3 Energy Electronic Data Sheet EEDS 3 3 1 Overview and justification The concept of using electronic data sheets for a range of components was introduced in Section 2 4 3 It is extended here by the proposed use of electronic data sheets to be known as the Energy Electronic Data Sheet EEDS to store parameters of energy modules The aim of the EEDS is to promote the reconfigurability of the energy resources of sensor nodes by storing parameters related to the operation of each energy module on the modules themselves rather than hard coding this information into the embedded software of the sensor node as is the case with present systems The data sheet identifies the type of module for example vibration energy harvester and its operating parameters including how the power gener
174. lers on sensor nodes The graph shows that the load type has a major effect on the discharge characteristics with the curves diverging substantially as the capacitor discharges In this example the constant power discharge curve reaches 2 0V approximately 32 faster than the constant resistance curve 2 0V is a typical minimum voltage at which microcontrollers will function This investigation was carried out in simulation as it would not be possible to ensure that real supercapacitors had exactly the same level of charge at the start of each test due to their complex characteristics and long time constants Clearly the dynamics of discharge have a major impact on the endurance of the power supply Assuming a resistive discharge profile when the load is actually power dominated can result in overly optimistic remaining lifetime calculation results and for these esti mates to decrease much more rapidly than expected towards the end of its lifetime an undesirable situation as it would lead to an accelerated decline in resources The value of y represents the node s proportion along the theoretical time axis from being fully charged to empty in the case of the CC2430EM this is from 3 6V to 2 0V The reader may question the value of using the time axis when nodes are energy harvesting and their stored energy may increase as well as decrease This method is simply a snapshot of the present energy status of the node and the value c
175. less sensitive to repeated charge discharge cycling than batteries and the system could be expected to last longer than simpler systems such as Heliomote 111 which buffer energy only in batteries Several larger networks incorporating solar energy harvesting have been deployed Under Chapter 2 Background Energy Sensing and Wireless Communication 51 WIRELESS SENSOR NODE TELOS ENERGY LEVEL CONTROL 1 0 POWER SENSING ADC hr a l d Li ry E 1 SWITCH i 1 1 1 A ENVIRON i i MENTAL LA ENERGY SOLAR PRIMARY ae PANEL BUFFER LA CHARGING super M CONTROL LITHIUM EAPACHOR RECHARGEABLE A Architecture FIGURE 2 28 The Prometheus architecture and prototype reproduced from 15 which incorporates a solar cell and stores energy in a battery and supercapacitors the Trio project a test bed of 557 solar powered motes were deployed in 2005 over an area of 50 000m 7 The Trio motes are based on the Prometheus energy harvesting system The actual deployment exposed a range of problems including oversights in the energy harvesting device solar cells were obscured by dirt and bird droppings issues with transfer of energy between the primary and secondary stores and the high power requirement of the TinyOS operating system Over the air programming was severely disrupted by nodes power cycling and losing the content
176. less sensors and careful control of node activity to conserve energy is essential Energy harvesting is often by its nature an inconsistent operation and the management of energy is a non trivial task Where nodes harvest energy from their environment their energy status may change rapidly and adaptive behaviour is Chapter 1 Introduction 5 desirable In addition seemingly straightforward operations such as assessing the state of charge of certain energy storage devices can be difficult The real time assessment of the energy status of a sensor node can therefore be a highly complex task FIGURE 1 4 A S NAP vibration powered sensor node mounted on an air compressor during a trial reproduced from 6 Supercapacitor Solar Cell Microphone User Reset Buzzer USB Port PIR 4 FIGURE 1 5 Trio wireless sensor node and its key components reproduced from 7 At present systems incorporating energy harvesting such as the S NAP vibration powered node shown in Figure and the Trio outdoor solar powered node shown in Figure 1 5 are designed for specific harvesting devices and are relatively inflexible The re configuration of the device to interface with an alternative energy harvesting device would require a re design of the node s hardware and embedded software Furthermore there are very few nodes which allow the combination of multiple energy harvesting devi
177. ll follow this strategy 1 Energy harvesting and management The performance of each energy mod ule will be explored This will include analysis of overall efficiency the impact of Chapter 3 Development Towards a Reconfigurable Energy Subsystem 89 the added energy management or control circuitry and where relevant the quies cent power draw will also be assessed The associated overheads of the prototype multiplexer module will also be evaluated 2 Embedded software development The energy related functionality of the node will be evaluated in terms of the programming effort in delivering energy aware operation and the applicability of the embedded software to cross platform operation The energy impact of processing operations will be explored as will its applicability to different hardware e g the impact of mathematical operations on microcontrollers equipped with hardware multipliers or low power modes The overall effectiveness of thee energy aware functionality will be assessed through the case study 3 Electronic data sheets and hardware interfaces The overall operation of the prototype will be reported with particular reference to the start up and plug and play features of the system In particular the energy demands of the EEDS and CHI will be explored The applicability of these features to a range of energy hardware will be assessed and any limitations of the prototype system will be stated Each aspect of
178. ll from its load and the open circuit voltage of the cell can be measured through the analogue measurement line In other situations the lines are used to connect the harvesting source to a known load to estimate the power being generated Furthermore some modules will be managed by the microcontroller For example it may be desirable to only switch a primary battery into the system if there is a high priority message to be transmitted by the device and its energy status is low In this situation one of the bi directional digital lines can be used to enable the discharge of Chapter 3 Development Towards a Reconfigurable Energy Subsystem 87 the battery to the load in normal operation other lines can be used to gauge the state of charge of the battery or its connection status In general the control line must have persistence This is because the bi directional lines are multiplexed and it is desirable for the microcontroller to be able to alter settings on a module and ensure that they will be maintained in their requested state after the multiplexer is switched away from their address Some modules such as rechargeable batteries require different types of control which is why two bidirectional digital control lines have been allocated for these operations This provides a high level of flexibility when used in combination with the additional digital measurement control and analogue measurement lines 3 8 3 Monitoring and a
179. lly expected to last for months or years meaning that average power consumption must be very low and energy harvesting devices are also becoming established most of which produce less than 1mW of power This project is concerned with the development of a flexible energy subsystem that can accommodate a range of energy resources Energy harvesting sources which deliver between 1004W and 10mW during normal operation are considered This power level alone is insufficient to power typical sensor nodes in 62 Chapter 3 Development Towards a Reconfigurable Energy Subsystem their active mode when transmitting or receiving data therefore system operation must be duty cycled and harvested energy must be accumulated during periods when the node is sleeping In order to make best use of its energy resources especially where energy harvesting provides power to the system energy awareness is essential Energy aware operation permits the node to adapt its activities to exploit plentiful or scarce energy resources and potentially to learn and predict any patterns in energy availability When only a small amount of energy is buffered such as where only a supercapacitor used for this purpose sufficient energy for less than a minute of active power consumption may be stored If the node fails to adapt its activity to its energy status the stored energy may quickly become absolutely depleted This is undesirable for two reasons firstly the n
180. ls EP_2 40 EP_1 20 Very limited energy EP_Empty 2 Cannot sustain activity EP_Unknown Error calculating status unknown TABLE 3 7 Energy Priority Levels 3 6 3 Categorisation of load type In the context of state of charge determination or calculation of the remaining lifetime fraction it is helpful to know what the dynamics of the load are Here the discharge types are given three categories These categories do not imply that a constant load exists rather that the dynamics of the load in response to a changing supply voltage are closest to that of the analogous component be it resistance current or power 76 Chapter 3 Development Towards a Reconfigurable Energy Subsystem Resistive dominated discharge In this type of discharge the load s response to changing supply voltage is similar to that of a resistor It is not implied that the load imposes a constant resistance on the energy store For example for a given activity level the node will draw around 50 more current at 3V than it would at 2V which can be modelled by V IR Current dominated discharge Here the load responds to changing voltages in a similar way to a current source For a given activity level the current drawn from the store is constant but the power used changes dependent on the relationship P VI Therefore at 3V the node will consume around 50 more power than at 2V Power dominated discharge This type of discharge implie
181. lux will generate approximately 1 8mW at 3 33V and 0 537mA Low indoor light levels mean that power tracking circuitry designed for outdoor use is often too power hungry to operate indoors Conventionally indoor PV energy harvesting circuits will simply consist of a diode between the PV module and a battery or capacitor which inhibits the reverse flow of energy from the store when the cell is not exposed to sufficient light This arrangement is particularly inefficient when charging a supercapacitor from empty as in this situation the PV module is forced to operate far from its maximum power point MPP voltage For the purposes of this investigation a maximum power point circuit suitable for indoor use has been devel oped and is described in detail in Section 4 5 1 It has been designed to interface with photovoltaic modules that operate with a nominal voltage of around 3 5V and with a nominal power output of around 1mW under typical lighting conditions Vibration electromagnetic vibration energy harvester As part of the AEASN project described in Section a colleague Neil Grabham developed a circuit to interface with a vibration energy harvester This circuit was then adapted by the author of this thesis to interface with the plug and play architecture Chapter 4 Case Study Deployment in a Prototype System Hardware 93 developed under this project The vibration energy harvester is a prototype Perpetuum device and generates energ
182. manner The scheme has been evaluated by way of a prototype which accommodates a range of energy devices The main contributions of this research are threefold firstly the system is enabled by a new hardware interface between the energy devices and sensor node secondly an embedded software structure is implemented to interface with the energy hardware and thirdly efficient energy aware modules compliant with the scheme have been produced The combined result is a novel energy subsystem for wireless sensor nodes that supports a range of energy devices and can deliver energy aware operation for a range of microcontroller platforms while imposing a minimal additional resource requirement to deliver this functionality Contents xvii xix xxiii XXV 1 1 1 Wireless autonomous sensing e e 1 1 2 Outline of the AEASN project o o e o 3 1 3 Justification for this research e e e 4 1 4 Contributions of this research 2 e e e 6 15 What is reconfigurability In 8 L6 Publicacions a ie a e e a a ea ee 8 1 7 Document otructurg 9 11 A A E E 11 E E A 11 2 2 1 Motivations and OVervieW e 11 2 2 2 Non rechargeable primary batteries 12 2 2 3 Rechargeable secondary batteries 14 er thee oe Ge Oh go ahd ide a Tee pee Ge 16 oD GM BG a ee 16 Dh g hh bk ep dee ee bee a ates 17 A 18 2 3 1 Motivati
183. mbination so that the difference between the time of arrival of the two transmissions can be measured to determine the distance 38 Chapter 2 Background Energy Sensing and Wireless Communication between transmitter and receiver 74 In underwater environments communications are most frequently acoustic such as the CORAL miniature communication subsystem 75 There have been some radio technologies developed for underwater communica tions for example the 1510 Underwater Radio Modem from Tritech 76 but these systems generally use low frequencies high transmission powers and large antennae A new technology known as RuBee is being developed under IEEE standard P1902 1 77 RuBee devices have a low frequency radio around 131kHz with a very small an tenna compared to their wavelength which is approximately 2 3km and operate in the near field hence exploiting the effect of magnetic induction RuBee signals can travel through metal water and solid objects far more effectively than ultra high frequency RF transmissions The maximum range achievable by RuBee is approximately 30 metres and its data rate is a comparatively slow 1 2kbit s 2 5 4 Networking and routing The aim of routing protocols is to get data from one point in the network to another point in an efficient manner The efficiency of a routing protocol may be judged on overall energy consumption the number of hops and sometimes the evenness of the distribut
184. me Low Power http www microchip com en_us technology xlp 2010 Last accessed May 2010 Atmel Corporation AVR Solutions http www atmel com products avr 2010 Last accessed May 2010 Crossbow Technology Inc Imote2 Data Sheet http www xbow com Products Product_pdf_files Wireless_pdf Imote2_Datasheet pdf April 2007 Last accessed May 2010 Texas Instruments Inc The all new CC430 combines leading MSP430 MCU and low power RF technology http www ti com cc430 2010 Last accessed May 2010 J Hill M Horton R Kling and L Krishnamurthy The platforms enabling wireless sensor networks Communications of the ACM 47 6 41 6 2004 BIBLIOGRAPHY 199 97 Sun Microsystems Inc Sun SPOT World Program the World 98 99 100 101 102 103 104 105 106 107 2010 Last accessed May 2010 K Martinez R Ong and J Hart Glacsweb a sensor network for hostile environ ments 2004 First Annual IEEE Communications Society Conference on Sensor and Ad Hoc Communications and Networks pages 81 7 2004 J Schiller A Liers H Ritter R Winter and T Voigt Scatterweb low power sensor nodes and energy aware routing Proceedings of the Annual Hawaii Inter national Conference on System Sciences pages 286 94 2005 G V Merrett A S Weddell N R Harris N M White and B M Al Hashimi The unified framework for sensor networks A systems approach
185. metres and a low duty cycle Short range wireless communications can be delivered via a range of media including infra red radio frequency magnetism and acoustics RF communications have achieved dominance for wireless sensor networking mainly due to their low component cost early standardisation and applicability to a wide range of deployment environments Other technologies however have niche applications where radio communications are impractical Dependent on the complexity of the protocol schemes can permit star tree and mesh network topologies as shown in Figure Do 2 5 2 RF based methods The dominant basic standard for short range RF wireless sensor communications is IEEE 802 15 4 65 The standard defines the radio channels and transmission schemes to be 36 Chapter 2 Background Energy Sensing and Wireless Communication Dei Dei Ka h 7 4 ei lt p 7 7 Dei 3 d p7 p Y NV SEN 7 N y EOV SCH e ug dh _ Too P sp 7 TON 7 Pi 4 N 7 d d Star Tree Mesh FIGURE 2 19 Star tree and mesh network topologies used in wireless personal area networks WPANSs and facilitates but does not directly implement schemes which allow data to be routed through a number of sensor nodes The initial release of the standard IEEE 802 15 4 2003 which is widely used defines physical channels in the Industrial Scientific and Medical ISM bands at 868MHz in Europe 915MHz in
186. munications and information exchange between systems local and metropolitan area networks IEEE Std 802 11 2007 Revision of IEEE Std 802 11 1999 June 12 2007 D Vassis G Kormentzas A Rouskas and I Maglogiannis The IEEE 802 11 standard for high data rate WLANs Network IEEE 19 3 21 26 May June 2005 BIBLIOGRAPHY 197 71 72 73 74 75 E 77 78 79 80 81 82 83 Texas Instruments Inc CC2430DK Development Kit User Manual ti com lit pdf swru133 October 2007 Last accessed May 2010 Texas Instruments Inc SimpliciTI Overview http www ti com litv pdf July 2007 Last accessed April 2008 Infrared Data Association IrDA IrDA Data Specifications year 2006 how published http www irda org displaycommon cfm an 1 subarticlenbr note Last accessed May 2008 Crossbow Technology Inc MCS410 Cricket Wireless Location System Data Sheet http www xbow com Products Product_pdf_files Wireless_pdf MCS410_Cricket_Datasheet pdf January 2006 Last accessed May 2010 S Pandya J Engel J Chen Z Fan and C Liu CORAL miniature acoustic communication subsystem architecture for underwater wireless sensor networks 2005 IEEE Sensors page 4 pp 2005 Tritech International Ltd and Wireless Fibre Systems Ltd Underwater Radio Mo dem 1510 http www wirelessfibre co uk index php page downloads October 2006 Last accessed May 2010 J K Stevens
187. n Listing 5 5 The TParamCurve field stores complex parameters used in this case study by the photovoltaic module The PrimaryParams field stores operational data for primary batteries for which state of charge cannot be calculated it simply stores the maximum capacity as an unsigned int and an end of life threshold value that indicates when the store is approaching empty Curves are represented by the TParamCurve struct which defines a set of individual points that are interpolated between in order to determine the state of charge of an energy store The total amount of energy is stored alongside this curve and each point on the interpolated curve is represented by a TParamCurveSingle data point which stores the voltage and the percentage of full energy with which it corresponds In order to minimise the amount of memory used by these fields the voltage values are stored in HalfVolts field which in fact correspond to the measured voltage multiplied by 50 and stored as an unsigned char This enables a voltage range of 0 5 10V to be stored effectively with a 20mV resolution in an 8 bit data field 5 3 5 Example data sheet contents The contents of two data sheets are shown here for illustration purposes The mecha nisms for reading the electronic data sheet values interrogating the energy modules and calculating the overall energy status of the node is documented later Table 5 1 shows the datasheet for the NiMH battery module In thi
188. n devices including supercapacitors and secondary batteries The specific devices selected for this case study represent what is commercially available and realistically able to provide power for a sub milliwatt wireless sensor node The sensor node uses the electrical energy provided by these resources to carry out sens ing processing and transmission of sensed data The actual sensing application is not a focus of this thesis so a simple temperature sensing application is used to demonstrate the capabilities of the system Consideration is given to the power requirements of sens ing operations and the system is designed in such a way as to accommodate more energy or processor intensive tasks such as vibration analysis The system developed for this case study demonstrates and enables the evaluation of the energy adaptive features of the architecture implementing the hardware specification defined in Chapter 4 2 2 Available energy sources Light photovoltaics PV Initial measurements indicated that lighting on the author s desk in the ESD lab varied between 700 and 1 200 lux depending on time of day Amorphous silicon a Si photo voltaic cells have a relatively high efficiency at low light levels compared to other types of cell which makes them particularly suited to use indoors Therefore a number of pho tovoltaic cells were procured from Schott Solar GmbH The cells part number 1116929 have dimensions 90x72mm and at 1 000
189. n that MSP430 microcontrollers have a typical supply voltage range of 1 8 3 6V and the CC2430 requires 2 0 3 6V in order to support a wide range of microcontrollers the multiplexer module in the prototype will regulate the supply voltage to between 2 0V and 3 0V The multiplexer module supports an unregulated raw voltage that is higher than the maximum supply voltage of the microcontroller It is believed that this will normally be 4 5V as this is the maximum voltage supported by many commercially available supercapacitors 3 5 2 Energy modules Each energy module will feature the following hardware 1 Switches and diodes to prevent the unwanted backflow of energy to harvesting devices from the multiplexer module 72 Chapter 3 Development Towards a Reconfigurable Energy Subsystem 2 Optional switching hardware to allow the querying of the module s energy status 3 Optional resistor divider to bring analogue measurements into the range of the microcontroller s ADC buffered by an operational amplifier 4 Optional persistent management channels for example to maintain the charge or discharge enable control status for battery modules 5 1 wire device programmed with module EEDS 6 Overvoltage protection limiting the module s output voltage if necessary 7 A socket to enable connection with the multiplexer module Essentially the purpose of the energy modules is to ensure that they generate
190. nable to complete due to low supply voltage it returns a 0 otherwise if changes were found it returns a 2 Together these functions enable the energy status to be updated periodically and for the addition or removal of energy modules to be detected and appropriate action taken Detail of capabilities The limitations of a resource constrained microcontroller mean that some mathematical functions need to be performed with care in order to maintain the precision of results For example the calculations carried out for estimating the power obtained from the photovoltaic module involves the use of a Taylor expansion to implement Equation mathematically The Taylor expansion of the exponential function e is shown in Equation The calculation involves a substantial amount of floating point arith metic which is challenging for these resource constrained microcontrollers Indeed the limitations of the fixed point numbers also mean that the precision of the Taylor expan sion approximation is limited As an example Figure 5 8 shows the computation of ef using the expansion shown in Equation to a varied amount of terms Due to the limits of the long variable which is used to hold the result of the factorial function the thirteenth term 13 6 227 020 800 is the highest term that can be computed as e Chapter 5 Case Study Deployment in a Prototype System Software 143 is raised to higher powers the precision b
191. nd devices to operate for over a year from a single coin cell None of the Bluetooth technologies are able to route data packets with all being capable only of star network operation i e all end devices must communicate directly with their host Texas Instruments have developed two proprietary protocols Their Simple Packet Pro tocol SPP is included in their demonstration applications and features address recog nition acknowledgement retransmission and error checking 71 It interfaces with their IEEE 802 15 4 compliant transceivers but is not a compliant protocol as such A newer development is their SimpliciTI Network Protocol which supports a similar set of devices and is in active development first released as a stand alone product in 2007 72 2 5 3 Alternative communication methods In very short range less than a metre line of sight applications infra red may be a vi able solution IrDA offers data rates of between 115 2kbit s and 16Mbit s 73 Devices must be aligned accurately as transmission cones can be as little as 15 which means that in practice devices would have to be permanently fixed in position to guarantee a reliable data link This property does have some advantages for security of transmissions but this is rarely of interest in wireless sensor networks Alternatively acoustic based communications are occasionally used in sensor networks Some wireless sensor nodes use sound and radio transmissions in co
192. ne with the basic template stack in Figure Da For completeness the communications stack shown here has three levels but for simplicity it is not anticipated that pre existing communication stacks would be shoehorned into this scheme There is no reason why the number of stacks could not be extended beyond the three implemented in the prototype to include other functions such as locationing but this is beyond the scope of this project Shared Application Energy Control Sensor Evaluation Energy Analysis Sensor Processing n E 2 bel 9 c E E e O Energy Management Intelligent Sensing FIGURE 3 8 A combined stack comprising stacks for communications energy man agement and sensing Reproduced from 124 A three layer stack structure was chosen after much deliberation as it offered the most logical method to separate the device interface into discrete levels while offering the facility for distinct interfaces to be offered between each layer From a high level point of view the interface layer deals with the physical interface with the device the medium The Unified Framework was initially conceived by Merrett and early definition work was carried out jointly by Merrett and Weddell being documented in a technical report 100 and later published 24 Further work particularly in the definition development and deployment of the Energy Stack has been carried out by Weddell 82 Chapter 3 Developmen
193. neration 2 Reporting To report the results of queries by setting the energy priority level and other variables accessible from the ECO to the application layer It may also detect trends and report these 3 Decision making In more developed systems to make decisions to maintain the energy integrity of the node by for example transferring charge between energy stores or activating energy converters Effectively the ECO layer takes the highest level view of the energy subsystem and manages the resources below it The ECO layer presents a generic interface so that the method of interfacing between the application layer and the energy stack are standard ised regardless of the exact nature of the energy subsystem Energy Analysis EAN Layer The energy analysis layer provides the interface between the ECO and the PYE layers While not being concerned with the detailed operation of switching and measurement 84 Chapter 3 Development Towards a Reconfigurable Energy Subsystem hardware it uses energy device models in order to compute energy and power levels from sensed parameters It initiates voltage measurements physical switching and reconfiguration of the energy subsystem all through the PYE layer The aim is to provide a consistent interface to the ECO layer independent of the connected hardware 1 Interpretation Through using energy source and store models this layer has the task of converting raw sensed values
194. nfigure the energy stack and prompt it to update The functions include 144 Chapter 5 Case Study Deployment in a Prototype System Software typedef enum ENERGY PRIORITY PP MAINS EP_5 EP_4 EP_3 EP 2 EP 1 EP EMPTY EP UNKNOWN ENERGY PRIORITY typedef struct ENERGY _THRESHOLDS unsigned char HU A unsigned char ET_3 unsigned char ET_2 unsigned char ET_1 unsigned char ET_0 JENERGY THRESHOLDS LISTING 5 8 Energy priority enumeration and threshold struct from ECO layer e ecoSetThresholds enables the energy priority thresholds to be manipulated they are initialised to the states shown in Table e ecoSetupEnergy is called to initialise the energy stack variables it is called on first device start up e ecoUpdateEnergy is called regularly and updates the energy status values e ecoQuerySupplyVoltage enables the application to query the supply voltage of the microcontroller this is occasionally useful for other node functions Further functions can be implemented to allow the application running on the sensor node to obtain more detailed information on the energy hardware for example this may be useful to allow the node to transmit information on the function of each of its energy modules perhaps for fault finding purposes to identify a malfunctioning device Listing shows the structs for the energy priority and threshold values held by the node In this implementation the thresholds are perc
195. nnual IEEE Communications Society Conference on Sensor Mesh and Ad Hoc Communica tions and Networks SECON 2009 22 26 June 2009 Rome Italy This publication is not available in the online version of this thesis but may be down loaded from http eprints ecs soton ac uk 17325 Appendix C Selected Publications 169 Weddell A S Grabham N J Harris N R and White N M 2009 Modular Plug and Play Power Resources for Energy Aware Wireless Sensor Nodes In Sixth Annual IEEE Communications Society Conference on Sensor Mesh and Ad Hoc Communica tions and Networks SECON 2009 22 26 June 2009 Rome Italy This publication is not available in the online version of this thesis but may be down loaded from http eprints ecs soton ac uk 17325 170 Appendix C Selected Publications Weddell A S Grabham N J Harris N R and White N M 2009 Modular Plug and Play Power Resources for Energy Aware Wireless Sensor Nodes In Sixth Annual IEEE Communications Society Conference on Sensor Mesh and Ad Hoc Communica tions and Networks SECON 2009 22 26 June 2009 Rome Italy This publication is not available in the online version of this thesis but may be down loaded from http eprints ecs soton ac uk 17325 Appendix C Selected Publications 171 Weddell A S Grabham N J Harris N R and White N M 2009 Modular Plug and Play Power Resources for Energy Aware Wireless Sensor Nodes In Sixt
196. nsing technology A number of reported deployments have used nodes to monitor soil param eters such as moisture content over a wide area 109 A range of other test beds and deployments have been described in the literature including some by the Centre for Embedded Networked Sensing CENS 110 Details of systems incorporating energy harvesting are discussed in the following subsections 2 9 2 Systems featuring a single form of energy harvesting A number of projects have used energy harvesting technologies to deliver sustainable power for wireless sensor nodes Photovoltaic modules are by far the most prevalent form of energy harvesting technology in part due to the plentiful supply of light in many deployment settings their simplicity and low cost Nodes conventionally store electrical energy in supercapacitors or batteries to achieve operation in darkness and during bursts of high current draw A notably sophisticated solar energy harvesting platform Prometheus buffers energy both in supercapacitors and a lithium polymer rechargeable battery 15 The system architecture is shown in Figure 2 28 along witha photograph of a prototype The supercapacitors are used for short term energy storage while the battery stores excess energy during the day and is used to top up the superca pacitor when it becomes depleted for example overnight during darker winter months In this way stress on the battery is minimised supercapacitors are far
197. nsor nodes are commercially available with systems based on the Texas Instruments MSP430 now becoming dominant Most systems incorporate 2 45GHz radio transceivers with many being IEEE 802 15 4 2003 compliant Typical wireless sensor nodes draw a sleep current of lt luA and an active current of approximately 25mA A number of system on chip microcontrollers and radio transceivers are now available meaning that systems can potentially be made very small Capable of operating at very low duty cycles typically below 1 these devices are able to operate comfortably from average power levels of lt 1mW These devices normally have a large number of input and output ports being capable of performing analogue to digital conversion on selected pins The increasing capabilities of microcontrollers in this area and their low power consumption means they are now capable of operating from harvested energy and of monitoring their energy hardware using their input output pins 2 8 Software and algorithm development 2 8 1 Software structures Sensor node embedded software is almost always built around the communication stack Figure 2 24 shows the stack structures for a range of wired and wireless communication protocols The reasons behind the dominance of communication stacks in sensor nodes are largely historical the most complex hardware module in sensor nodes was normally the communications subsystem Moves towards integration and SoC have added fu
198. nt D bt DS2502 BATT MEAS OP mes LG 8 IR BATT MEAS DATA IR VREG FIGURE A 6 Battery module schematic Appendix B Energy Electronic Data Sheet Contents Mains Module Device Type 10 0001 00 0x84 Measurement OnOffMaximum 0x01 Multiplier 10 0x0A Max Output 4 5 45 0x25 TABLE B 1 Electronic Data Sheet for mains module Vibration Energy Harvester Module Device Type 10 0011 00 0x8C Measurement Square 0x03 Multiplier 6 6 66 0x42 Max Output 9 0 90 0x5A PMultiplier 15 Ox 00 00 00 OF TABLE B 2 Electronic Data Sheet for vibration energy harvester module 163 164 Appendix B Energy Electronic Data Sheet Contents Primary Battery Module Device Type 10 0001 10 0x46 Measurement EndOfLife 0x02 Multiplier 4 9 49 0x31 Max Output 3 6 36 0x24 PEndOfLife 20 0x14 PMaxCapacity 10260 0x00 00 28 14 TABLE B 3 Electronic Data Sheet for primary battery module Supercapacitor Module Device Type 01 0011 00 0x4C Measurement EndOfLife 0x02 Multiplier 4 9 49 0x31 Max Output 4 5 45 0x2D PMultiplier 0 275 28 0x1C TABLE B 4 Electronic Data Sheet for supercapacitor module Appendix C Selected Publications The following publications ar
199. nt pages 522 524 Piscataway NJ 08855 1331 United States 2008 SBS Implementers Forum System Management Bus SMBus Specification smbus org specs smbus20 pdf August 2000 Last accessed March 2009 H Taylor and L W Hruska Standard smart batteries for consumer applications page 183 New York NY USA 1995 Power Management Bus Implementers Forum PMBus Power Management Pro tocol Specification http www powersig org February 2007 Last accessed March 2009 IEEE Standards Association IEEE standard for information technology telecom munications and information exchange between systems local and metropolitan area networks specific requirements part 15 4 Wireless medium access control MAC and physical layer PHY specifications for low rate wireless personal area networks WPANs IEEE Std 802 15 4 2006 Revision of IEEE Std 802 15 4 2003 2006 ZigBee Standards Organization ZigBee Specification Document 053474r17 wuw zigbee org Products DownloadZigBeeTechnicalDocuments aspx 2007 Last accessed April 2008 Microchip Technology Inc MiWi Wireless Networking Protocol Stack wwi microchip com downloads en AppNotes 01066a pdf 2007 Last accessed April 2008 HART Communication Foundation Wireless HART Technology http www hartcomm org protocol wihart wireless_technology html 2009 Last ac cessed May 2010 IEEE Standards Association IEEE standard for information technology telecom
200. nternational Instru mentation and Measurement Technology Conference May 12 15 2008 Victoria British Columbia Canada This publication is not available in the online version of this thesis but may be down loaded from http eprints ecs soton ac uk 15361 188 Appendix C Selected Publications Weddell A S Harris N R and White N M 2008 Alternative Energy Sources for Sensor Nodes Rationalized Design for Long Term Deployment International Instru mentation and Measurement Technology Conference May 12 15 2008 Victoria British Columbia Canada This publication is not available in the online version of this thesis but may be down loaded from http eprints ecs soton ac uk 15361 Appendix C Selected Publications 189 Weddell A S Harris N R and White N M 2008 Alternative Energy Sources for Sensor Nodes Rationalized Design for Long Term Deployment International Instru mentation and Measurement Technology Conference May 12 15 2008 Victoria British Columbia Canada This publication is not available in the online version of this thesis but may be down loaded from http eprints ecs soton ac uk 15361 190 Appendix C Selected Publications Weddell A S Harris N R and White N M 2008 Alternative Energy Sources for Sensor Nodes Rationalized Design for Long Term Deployment International Instru mentation and Measurement Technology Conference May 12 15 2008 Victoria Brit
201. nvariably is more time consuming The choice of which method to use is left to the system designer and will depend on the likelihood of devices being swapped between re scans of the energy subsystem and the impact of the EEDS on the microcontroller being out of date It may be the case that a hybrid perhaps with one re scan per day method of change detection may be effective whilst remaining energy efficient Device interfacing The device interfacing functions deal with the physical interface to the energy modules They deal with setting address pins and obtaining measurements e pyeSetAddress takes an unsigned char input the address that is desired and sets the address pins appropriately e pyeTestSupplyVoltage simply calls the halSampleSupplyVoltage to sample the supply voltage of the microcontroller and returns the supply voltage as an un signed char multiplied by 10 so that 2 2V would be reported as 22 e pyeGetMeasurement configures the on board ADC to take a measurement The function takes an unsigned char input the address of the device of interest in case the address has not already been set It initiates a measurement and returns it as an unsigned int The module scales the value appropriately dependent on the value of the multiplier field in the device EEDS The value returned is the voltage multiplied and scaled up by 100 e pyeGetChargeStatus reads the charge status of the module if charging is enabled it
202. ny this means they must operate without the constraint of a wired power supply Conventionally these devices have been powered by non rechargeable batteries which are replaced when de pleted Energy harvesting also known as energy scavenging now offers the potential to sustain the operation of sensor nodes indefinitely In this process environmental en ergy is converted into electrical energy which is then used to power the sensor node The sporadic nature of energy harvesting means that energy must be used carefully and buffered in rechargeable batteries or supercapacitors Ideally nodes should be energy aware and adapt their operation to the available energy however this is non trivial in systems that feature complex energy subsystems that may potentially include a number of energy harvesters and energy storage devices The current state of the art in wireless sensor nodes and their energy supplies was discussed in detail in Chapter The work carried out under this project started with a new concept to take the integra tion of wireless sensor nodes a step further by enabling the energy hardware on sensor nodes to be configured at the time of deployment in a plug and play manner The work carried out spans across several areas and has addressed a number of academic chal lenges This thesis detailed the development of a novel and comprehensive scheme for reconfigurable energy aware sensor nodes the described system allows sensor nodes to su
203. o microcontrollers and transceivers which are cheaper faster and more energy efficient This means that there is now an increasing overlap between the amount of energy that can be generated through energy harvesting and the amount of energy required for wireless sensing Photovoltaic technology including fam ilies for indoor use is now mature and vibration and thermoelectric energy harvesters are finding widespread commercial applications although these are typically in the test deployment or evaluation stage at present Low self discharge batteries are now readily available and improvements in supercapacitor technologies have decreased their internal resistance and reduced the problems of leakage There is no reason to believe that the pace of development will slow substantially in the coming years Solid state rechargeable lithium batteries are now being launched onto the market and promise thousands of recharge cycles and 20 year lifetimes It is expected that developments in vibration thermoelectric and wind energy harvesting will continue spurred by the success of deployments in the field Furthermore developments in microcontroller technology will further reduce the power consumption of sensor nodes while increasing their capabilities However perhaps the most fluid area is in wireless communication standards development with many standards families being developed including Bluetooth Low Energy ZigBee Green Power and the EnO
204. ode will be forced to turn off and therefore may miss important events and secondly the node will have to go through a full start up process when it has accumulated sufficient energy to turn on wasting further energy and potentially losing some important data that may have been stored in volatile memory Achieving energy awareness in this application is challenging but strategies have been adopted in this project to reduce the computational load and storage requirements and the overall impact on the energy efficiency of the system This reaches from the algo rithms used to calculate power and energy down to the circuitry added to each device to facilitate the functionality There is also a trade off between precision and energy consumption some mathematical processes such as Taylor expansions gain precision when taken to more terms also many ultra low power components such as operational amplifiers have notably poorer performance than their high power counterparts Ul timately a precise calculation for energy stored or power generated is neither needed nor possible in general a tolerance of at least 10 on estimates of energy status is acceptable as this is within the tolerance of many components and all that is required is an indicative figure on which to base decisions about the node s activity level Delivering a flexible reconfigurable system architecture is also non trivial The approach adopted in this project is to e
205. of a control facility for this module imposes a 1 8uA quiescent current draw to maintain the state of the bistable multivibrator Additionally the interruption of the power path by one MOSFET and a Schottky barrier diode means that there will be an associated voltage drop At 1mA and 3V the voltage drop is around 260mV the presence of the diode causes an approximate 9 efficiency loss but is unavoidable due to the risk of unwanted recharging of the primary cell For simplicity there is currently no voltage step up capability This means that the module cannot supply a higher voltage than that of the cell s This is of little impact to the overall operation of the circuit as it simply means that the batteries will not charge a supercapacitor energy buffer above their own voltage which will typically be above 3V the regulated voltage for the microcontroller 4 5 6 Secondary battery module Control and startup features The secondary battery module is designed to accommodate a pair of AAA rechargeable low self discharge NiMH batteries in holders It uses the same PCB as the primary battery module shown in Figure 4 17 but with a number of additional components fitted to enable the recharging functionality The additional circuitry permits the mi crocontroller to control the charging of rechargeable batteries fitted to the module In common with the primary battery module circuitry is provided to allow the micro controller to assess the st
206. of energy stored by the system at a given time this may be as straightforward as determining the energy stored on a capacitor by measuring its voltage and using the standard capacitor equation In systems with multiple energy sources and stores the situation can be complex and the monitoring of the energy subsystem should report with sufficient detail to enable appropriate decisions to be made For example the new system architecture described later in this thesis supports up to six different energy devices These may include energy stores such as supercapacitors and primary or secondary batteries and energy sources including mains adapters and energy harvesting devices Ultimately the application running on the sensor node will need to assess the energy stored on the node by individually monitoring each energy device It must determine 42 Chapter 2 Background Energy Sensing and Wireless Communication the level of stored energy in supercapacitors and batteries and as a basic requirement must classify this as rechargeable or non rechargeable i e the system must know whether the resource it is using is capable of being recharged or if it is a single use resource More complex systems may take the degradation of the energy store into account for example by attaching a higher cost to the use of a rechargeable battery over that of a supercapacitor due to the lower cycle life of the former and thus ensure that the application
207. of the art in energy harvesting wireless sensor nodes The energy devices covered are appropriate for milliwatt scale sensor nodes so large scale energy generation and storage technologies such as those used to augment or replace a mains electricity sup ply for high power devices are not covered Technologies related to energy storage are introduced in Section and energy harvesting and wireless power transfer tech nologies are discussed in Section Standards and methods for intelligent sensing including plug and play sensors and transducer electronic data sheets are introduced in Section Section introduces wireless transmission protocols used by wireless sensor networks and strategies for energy aware operation are discussed in Section 2 6 Wireless sensor nodes and their underlying components are introduced in Section 2 7 and Section 2 8 discusses software and algorithm development A range of sensor node deployments including energy harvesting systems and a limited number of prototypes that support multiple energy devices are introduced in Section 2 9 2 2 Energy storage 2 2 1 Motivations and overview As discussed in Chapter 1 wireless sensor nodes will normally operate from non rechar geable batteries or less commonly energy harvested from their environment The low rate of power generation from most energy harvesting devices means that they 11 12 Chapter 2 Background Energy Sensing and Wireless Communication ar
208. offer similar features in terms of code editing project management and debugging Code is written in an extended version of C The extensions differ for each microcontroller which meant that it was necessary to implement a hardware abstraction layer documented later to facilitate interactions with the physical aspects of each device The fact that code written in C is supported by both devices meant that the remainder of the stack structure could be written in a generalised manner which was applicable to both devices and is portable to many other microcontroller families Physical programming of devices and debugging was enabled by hardware supplied by the manufacturers as part of their evaluation kits The eZ430 RF2500 was pro grammed via a simple USB dongle which also acted as a debugger and serial port over the same USB interface The eZ430 RF2500 was programmed via a SmartRFO4EB which is pictured in Figure The SmartRF04EB provides a USB interface for pro gramming debugging and a separate serial connection for data via an RS232 port The device shown in the figure is acting as a receiver for test transmissions from a node and has a CC2430EM module mounted directly on the board Clearly debugging devices while they were connected to energy harvesting systems with a fluctuating energy supply was very challenging It was necessary to remove resistor R1 from the eZ430 RF2500 board in order to allow it to function on a different su
209. ognise adapt to and manage the reconfigurable energy subsystem and use knowledge of its energy status to control the node s overall operation These three broad requirements form the basis of the work described in this thesis the development of a reconfigurable energy subsystem for wireless sensor nodes The aim of the reported work is to provide a generalised scheme that permits this energy adaptive behaviour but in a hardware agnostic way i e without being constrained to specific types of energy device The following subsections outline the requirements of the modular design the usage scenarios for the described system the major challenges and the general strategy that has been adopted Clearly the provision of these capa bilities should not adversely impact on the overall performance of the system it would be useless if the additional management hardware rendered the system significantly less efficient or unable to perform its sensing tasks The overall aim is that the tangible benefits of the system in achieving energy awareness for a plug and play reconfigurable 58 Chapter 3 Development Towards a Reconfigurable Energy Subsystem system outweigh any drawbacks in the potential loss of quantifiable efficiency of power conversion circuitry and additional processor cycles taken in managing the system Ob viously this is weighing a subjective measure against one that can easily be determined however this will be set in conte
210. onments 110 Chapter 4 Case Study Deployment in a Prototype System Hardware However the situation of interest to this investigation indoor artificially lit environ ments means that this simplification is acceptable Lastly the equation for nominal power assumes that the cell is operated at its maximum power point and the conversion circuitry is 100 efficient The following section gives an idea of the efficiency of the cir cuit and other complications but the result given for nominal power is sufficient to give an idea of the energy harvesting status of the node for use by energy aware algorithms Performance evaluation SPICE simulations indicate that this circuit runs at over 70 efficiency with the PV module described under normal indoor lighting Although the absolute efficiency of the circuit could not be tested due to the difficulties in ascertaining the maximum power from incident light practical tests comparing the circuit shown in Figure against a conventional diode only system show a 30 improvement in node start up times under normal office lighting The results of a race between a conventional circuit with the supercapacitor connected through a diode and the new switching based circuit similar to that shown in Figure 4 11 is shown in Figure Here it may be observed that the switching circuit delivers significantly better performance than the diode based circuit at lower store voltages The difference in gr
211. ons and OVervieW 0200080 18 2 3 2 Photovoltaics A e ek a A oe 18 2 3 3 Vibration energy harvesting e 20 EE 25 hey whe ooo ee be og ee dae ba ee 27 BR Bon A Ene Sed a St ee ake E 28 2 3 7 Inductive and RF energy transfer 29 WEEN 30 Ob e E ee ae ee A 30 2 4 Technologies for intelligent sensing o 32 2 4 1 The IEEE 1451 standards Tam 32 V vi CONTENTS 2 4 2 Transducer electronic data sheets e 33 2 4 3 Extensions to the electronic data sheet concept 34 pisas a a 34 ale amp an e e E e e Es 35 oe as te ich Has eee eee eg 35 2 5 1 Overview e 35 2 5 2 REF based method 35 2 5 3 Alternative communication methods 37 2 5 4 Networking and routing e o 38 DESEN 38 poe eee shee ee cs ee eee Gee oe GG 39 2 6 1 Energy aware routing 2 00 eee ee ee 39 ada BO E AE Ek a ae 39 oe Gee ee ee ee ee 41 EEN 42 Ab a Oe Be Bo Oa a Se ees o e aE 44 eor ER EE A FH oo aga Sk ewe eS 44 2 7 1 Microcontrollers transceivers and system on chip 44 EES 45 ade a A he he Ee we Be 46 2 8 Software and algorithm development 46 2 8 1 Software structures sis p rebs e e EE o 46 ite a od OS dp aa a Oe 4 47 Ply Py eb guea eRe eh beds a ae da 48 Op ee Fe pe ee hoe Be ce ee ee EE 49 HENNEN 49 2 9 2 Systems featuring a single form of energy harvesting 50 Sege 52 Bowes apa ee Bie de ae A
212. ons stack into their own separate software stacks The energy stack interfaces with the energy hardware including the electronic data sheets through the hard ware interface It allows the application running on the sensor node to adapt its behaviour dependent on the amount of available energy 3 Energy harvesting and management circuits have been developed to manage energy from harvesting devices and batteries Algorithms and circuits for the management of energy resources and determination of power status have also been developed The dynamics of energy harvesting and storage devices have been characterised and implemented in the deployed system with operating information being stored in the electronic data sheets and used by the embedded software Brought together the work described in this thesis represents a comprehensive scheme for delivering reconfigurable wireless sensor nodes The scheme defines the hardware and software interfaces between components and provides a plug and play capability for the energy subsystems of nodes The work has been verified through the deployment of the system in a demonstrator with a range of swappable energy components and sensor node platforms which are able to interface with the energy hardware The overall evaluation of the system is both quantitative and qualitative Some param eters such as the impact of adding circuitry to deliver energy awareness is relatively straightforward to evaluate
213. onserve energy for another time In the case of rechargeable batteries it may also be desirable to be able to disable recharging to ensure that any available power is used directly to power the microcontroller instead of to recharge batteries As these energy modules are normally controlled by the microcontroller via the multiplexer module it is essential that modules are able to remember their state The microcontroller may only connect to the energy module for management purposes once every few seconds or minutes or potentially hours so a capability must be implemented to allow devices to remember their status for extended periods For this application a bistable multivibrator shown in Figure 4 1 is a useful circuit to uC lt AN 220k to rest of module LA LMC7215 m 4 00n FIGURE 4 1 Bistable multivibrator circuit Used for charge discharge control state retention on energy modules The bistable multivibrator allows the state of the circuit to be retained The circuit utilises a comparator and the state of the output is stored on a capacitor This has the effect that the current state of the energy module can be both set and read by the microcontroller hence it must be connected to a bidirectional digital pin As will be discussed later it is possible for some energy modules to be controlled both by the microcontroller and externally by a human operator e g to power up the
214. operation of the node and may also indicate trends in the energy subsystem e Energy Analysis Layer EAN is concerned with device models processing requests from the ECO layer and interpreting data from the PYE It also carries out operations and measurements at the request of the ECO e Physical Energy Layer PYE executes the lowest level operations such as configuring ports obtaining ADC readings and controlling inputs and outputs Chapter 3 Development Towards a Reconfigurable Energy Subsystem 83 Shared Application Energy Control High level view of energy subsystem Energy Analysis Calculation of stored energy using models E o E D D G gt Em e o G u FIGURE 3 10 An Energy Stack reproduced from 124 Energy Control ECO Layer The energy control layer presents an interface to the shared application layer The basic interface includes the energy priority value of the system along with functions to permit the energy priority thresholds to be changed It may also be useful for the application layer to ascertain the supply voltage of the node and possibly the raw values of energy stored and other system level energy parameters The tasks of the ECO layer are broadly as follows 1 Query processing To process general queries from the shared application layer such as to determine the present supply voltage energy priority of the node and parameters related to energy ge
215. or each 10 C temperature decrease which is apparently found from the Arrhenius equation Here T is the rated temperature and Te is the expected temperature Thus at 30 C for a device rated at 70 C the temperature factor is found to be 16 Investigations into large supercapacitor behaviour indicate that the expected lifetime will also double for every 0 1V operating voltage reduction 89 Therefore at a voltage of 1 8V the voltage factor ay is 5 Hence for this device the expected lifetime in hours is found to be 80 000 This corresponds to an expected lifetime of approximately nine years Note that an increase in operating temperature of 10 C will effectively halve the expected lifetime To T Qr 2 10 2 9 Flash data retention is specified at a minimum of 100 years at 25 C for TI microcon trollers such as the MSP430 Flash lifetimes at higher temperatures can also be predicted by using the Arrhenius equation At 50 C memory retention can be expected to fall to 17 years 90 Other limitations include the number of write cycles to flash mem ory There are also some concerns that the progression towards reduced feature sizes 44 Chapter 2 Background Energy Sensing and Wireless Communication in integrated circuits may reduce their endurance due to the propagation of materials within the device there are however some power related reasons why feature sizes for low power microcontrollers remain relatively large
216. or future use it may be used to identify energy consumers should the scheme be extended Code Description 00 Multiplexer 01 Energy Store 10 Energy Source 11 Undefined reserved for future use TABLE 3 2 Module class codes Chapter 3 Development Towards a Reconfigurable Energy Subsystem 65 Initial EEDS identifiers for devices are shown in Table The first two bits of the type identifier correspond to the codes expressed in Table The middle four bits define the type of device for the multiplexer which is a special case this describes how many ports it has while for other devices it simply expresses what type of module it is e g for an energy storage device 0010 is a rechargeable battery The last two bits of the code indicate the function of the control lines from the microcontroller i e for the rechargeable battery the first bit indicates that the module s discharge can be turned on or off by the first control line and the second bit indicates that its charging function can be controlled by the second control line As a complete example 01000110 represents an energy storage device a primary battery whose discharge can be controlled through the first control line but which cannot be recharged This scheme is used by the prototype system developed as part of the case study reported later in this thesis Class Description 00011000 Multiplexer module 6 inputs 01000110
217. or microcontroller and 1 wire devices 4 Diode clamping on bi directional lines 5 Pull up resistor on microcontroller side of 1 wire multiplexer 6 Pull down resistors on address lines of multiplexers 7 Query facility to measure raw voltage scaled to 0 1V range 8 Small low ESR capacitor to act as operating buffer 9 1 wire device programmed with multiplexer EEDS 10 Up to six sockets for connection of energy modules The components selected for this module should draw as little current as possible and should not adversely impact on the overall efficiency of the system The 8 to 1 multi plexers must be capable of operating at voltages down to the minimum supply voltage from the multiplexer module nominally 2 0V The undervoltage protection circuit acts to disconnect the microcontroller if the unregulated voltage of the system drops below this minimum value Diode clamping on the bi directional lines ensures that signals from the energy modules do not exceed the supply voltage of the microcontroller The 1 wire pull up resistor facilitates 1 wire communications by pulling up the 1 wire bus line Pull down resistors on the address lines ensure that the multiplexers revert to address 0 for low power monitoring when the microcontroller is not actively commu nicating This also has the effect that the 1 wire pull up resistor is isolated when the device is not active thus the power consumption of the module is minimised Give
218. or node and a site survey carried out to determine the available environmental energy including light vibration and potential temperature difference 2 The transducer communication and sensor node requirements should be identified The overall energy demand of the sensor node should be estimated appropriate energy modules selected and the complete design s feasibility should be verified in combination with the results of the site survey 3 The microcontroller on the sensor node can be programmed with its ports being configured as appropriate to interface with its sensors and energy subsystem Chapter 3 Development Towards a Reconfigurable Energy Subsystem 61 to deliver an appropriate sensing scheme The system has a software energy stack which means that there is no need to mandate particular energy modules 4 The system can now be deployed The appropriate energy modules should be connected at the time of installation The microcontroller will now start up automatically and auto detect the energy hardware connected 5 The system will monitor and manage its energy resources with information about the energy status being made available to the application running on the microcontroller If desired the system will also be able to monitor the efficacy of energy devices reporting as appropriate in order to detect faults and monitor the energy status 6 If necessary modules on the energy subsystem can be a
219. ork e ee ee eee 154 6 3 1 Overview 154 6 3 2 Hardware development 154 6 3 3 Software development 155 6 3 4 Towards standardisation 0000084 155 6 4 A look to the future 156 A Module schematics 157 B Energy Electronic Data Sheet Contents 163 C Selected Publications 165 Bibliography 191 List of Figures eek Aaa we ee se ee 1 Lobras 2 a a ds A e AE kage alk 3 paag ee a 5 1 5 Trio wireless sensor node o oa a a e o 5 EENEG 10 KEE 14 HA 15 2 3 Efficiencies of PV cells relative to SIO o 19 abs pu a ia e eS 19 2 5 Characteristics of Schott Solar ASI Indoor Photovoltaic Module 20 a a 21 DEE 22 a yee 23 meee 23 tee 24 Sh do Bech ere Boe Pe ae ate ee 25 2 12 The Micropelt Generic Power Bolt 26 2 13 Tellurex PG1 Power Generation kul 27 EE 28 eee ates 28 gts Eta ah tcp Seat AS OR ah Blas Geode Ed e Ae ae ee I eee 30 Pais Beh we eae ee a 30 Seda pena seas 32 alge So a ee 36 Shee eee Be een nee et 37 A Sai eee Wwe dine dae bat ees oe a 40 2 22 Priority balancing in IDEAL Al 2 23 The Texas Instruments CC2430EM and eZ430 RF2500 45 DEER 47 pa Bos Fash a ON Bch eG He we ee S 48 2 26 Hardware abstraction architecture implemented in TinyOS 2 01 48 2 27 The Glacsweb Mk Il architecture e 50 per eee ds 51 2 29 Photovoltaic cells harvest energy from a ceiling mounted light unit 52 2 30 The Ambimax architecture and circuitry e 53 2
220. otovoltaic module It makes use of the exponential function described above and implements the calculations described in Section 4 5 1 The result is returned as an unsigned int 142 Chapter 5 Case Study Deployment in a Prototype System Software There are a number of intricacies associated with using these functions that are consid ered at the end of this subsection Device management functions These functions are provided for system initialisation and to enable the periodic refresh of data in the microcontroller s copy of the EEDS of the energy modules e eanInit is called on system start up to initialise relevant EAN variables to zero and calls the pyeInit function e eanStartUp tests the supply voltage of the microcontroller if it is below the thresh old value it returns a 0 otherwise it reads the EEDS data from all modules into the microcontroller memory and returns a 1 e eanRefresh checks the microcontroller supply voltage if above the threshold value it checks for changes using the pyeRefresh function If no change is detected the function returns a 1 if it was unable to complete due to low supply voltage it returns a 0 otherwise if changes were found it returns a 2 e eanRescan checks the microcontroller supply voltage if above the threshold value it checks for changes using the pyeRescan function If no change is detected the function returns a 1 if it was u
221. ource constrained wireless sensor nodes due to the overheads of the management scheme 2 4 5 Discussion The IEEE 1451 standard for plug and play industrial sensors offers real benefits to sys tem installers simplifying the deployment and configuration of wired sensors This is enabled in part by the transducer electronic data sheet format which standardises the format of the configuration and interface parameters for devices the concept has been extended to CEDS and HEDS and further to xEDS meaning that electronic data sheets can be used for a range of applications The SMBus and PMBus standards facilitate a common interface for system management in desktop computers and servers No comparable interface standards or systems exist for power or system management in wireless sensor nodes particularly for sensor nodes operating from harvested energy The author of this thesis believes that there is a compelling case for the energy devices on sensor nodes to feature electronic data sheets with their operating parameters in order that the system can effectively monitor and manage its energy subsystem and so that the energy hardware of the sensor node could be connected at the time of system deployment This concept is discussed in greater detail in Chapter 2 5 Wireless communication protocols 2 5 1 Overview Wireless sensor network communications generally have a low data rate with a short transmission range typically between 10 100
222. overall power requirement of the device Data is transmitted through the radioSend command which is also shown in Listing 5 6 eZ430 RF2500 SimpliciTI Communications for this device are provided by the SimpliciTI stack which was devel oped by Texas Instruments The protocol is more capable than the SPP but has a code size of approximately 4k It supports a number of network related tasks and has three Chapter 5 Case Study Deployment in a Prototype System Software 135 Initialise radio with new address MPL_Ioctl IOCTL_OBJ_ADDR IOCTL_ACT_SET amp sourceAddress MPL Init 0 Sleep Radio MPL_Ioctl IOCTL OBJ RADIO IOCTL_ACT_RADIO_SLEEP 0 Mn un un Wake up radio SMPL_Ioctl IOCTL OBJ_RADIO IOCTL ACTRADIO AWAKE 0 Send packet SMPL_Send SMPL_LINKID USER UUD msg sizeof msg Sleep radio SMPL_lIoctl IOCTL_OBJ RADIO TIOCTLACT RADIO SLEEP 0 LISTING 5 7 Code to interface with SimpliciTI communication stack layers as shown in Figure although it is debatable whether the Lite HAL layer is truly a layer It is this sort of capable but resource efficient communication protocol which is ideally suited to resource constrained wireless sensor nodes The method of sending data and managing the transceiver is shown in Listing 5 7 Network Apps Interface with application ping link join send receive etc Network Network related operations routing send receive
223. pe System Hardware 113 before rectification Once again in common with the wind module there are no addi tional components in the power path to affect the efficiency of the circuit during normal operation The interface for the thermoelectric module is shown in Table 4 7 Pin Type Function 2 Meas Control Connect generator to fixed load 3 Measurement Analogue voltage across fixed load 4 Control None 5 Control None TABLE 4 7 Interface pins from thermoelectric module 4 5 4 Mains module The mains module permits the system to operate from mains power The PCB from the schematic in Appendix A Figure A 4 and shown in Figure 4 16 has a standard 2 1mm jack socket and hence is compatible with a range of mains power adapters The PCB will regulate a 5 0 to 11 5V DC input down to 4 5V using a Maxim MAX639 switching regulator The EEDS identifies the type of module Additional circuitry permits the microcontroller to ascertain whether the mains adapter is supplying power to the system through a simple digital flag The microcontroller cannot act to turn off this supply as it is assumed to be a zero cost resource which should be taken advantage of whenever it is available in this case the overall efficiency of this module is immaterial FIGURE 4 16 Circuit board for the mains module Pin Type Function 2 Meas Control None 3 Measurement Digital supply connected 4 Control None 5 Control Non
224. petual environmentally powered sensor networks 2005 Fourth International Symposium on Information Processing in Sensor Networks pages 463 8 2005 Panasonic Industrial Company Gold Capacitors Technical Guide www panasonic com industrial components pdf goldcap_tech guide_ 052505 pdf May 2005 Last accessed May 2010 CAP XX Australia Pty Ltd Hs208 supercapacitor datasheet http www cap xx com resources datasheets CAP XX_HS208_Datasheet_v1 2 pdf February 2009 Last accessed May 2010 S Roundy D Steingart L Frechette P Wright and J Rabaey Power sources for wireless sensor networks Wireless Sensor Networks First European Workshop EWSN 2004 Proceedings pages 1 17 2004 J D Holladay E O Jones M Phelps and J Hu Microfuel processor for use in a miniature power supply Journal of Power Sources 108 1 2 21 27 2002 Microfuel processors S Whalen A Thompson D Bahr C Richards and R Richards Design fabrica tion and testing of the Pz micro heat engine Sensors and Actuators A Physical A104 3 290 8 2003 H Li A Lal J Blanchard and D Henderson Self reciprocating radioisotope powered cantilever Journal of Applied Physics 92 2 1122 7 2002 J P Fleurial G J Snyder J Patel J A Herman T Caillat B Nesmith and E A Kolawa Miniaturized radioisotope solid state power sources In Space BIBLIOGRAPHY 193 24 26 27 28 29 30 31 32 33
225. ply connected to the regulated supply voltage of the module meaning that they are effectively shut down when the voltage on the system is too low to sustain operation 4 4 6 Overall efficiency As shown in the schematics there are no diodes or other passive components in the electrical path between the energy modules and the microcontroller Therefore the main impact from the multiplexer module on the overall efficiency of the system is the quiescent power consumption of its components and the efficiency loss through the linear regulator Having five analogue switches with a quiescent current draw of 1uA and two voltage detectors each with a current draw of 0 7yA plus the 0 8uA quiescent current draw of the linear regulator this equates to an overall quiescent current draw of approximately 7 24A The efficiency loss through the linear regulator will impact most significantly when the raw voltage is above 3V The other measurement circuitry draws no current except when measurement operations are in progress 106 Chapter 4 Case Study Deployment in a Prototype System Hardware 4 5 Energy modules 4 5 1 Photovoltaic module Functional description The photovoltaic PV cell and this PCB from the schematic in Appendix A Figure A 2 and shown in Figure 4 10 should be taken together to represent the photovoltaic mod ule An EPROM holds the EEDS data with the operating parameters of the circuitry and PV cell in combination so it
226. pment of wireless sensor network technologies as they allow algorithm simulations to be verified and the design of sensor networks to be driven by real applications and experiences The highest profile network deployments have mainly been for environmental monitoring although there have been a range of other deploy ments aiming to demonstrate large scale networks or to test the effectiveness of energy harvesting technology The University of Southampton has been involved in a number of sensor network deploy ments for environmental monitoring The Glacsweb project pictured in Figure 2 27 monitored the behaviour of the Briksdalbre arm of the Jostedal Glacier National Park in Norway until the melting in 2006 resulted in the glacier becoming too steep and dangerous to work on Nodes were deployed deep into the ice and measured pressure temperature and tilt It should be noted that many of the nodes designed for a lifetime of up to 10 years outlasted the glacier they were embedded in A glacier represents an extremely harsh environment for wireless sensors with nodes being exposed to low temperatures high pressures and having to operate in a very difficult radio propagation environment Elsewhere wireless sensors have been deployed in a range of environmental monitoring projects Nodes have been used to monitor seismic events on the Volc n Reventador volcano in northern Ecuador 3 with a network comprising 16 nodes spread over a 3km wid
227. portion of time Chapter 3 Development Towards a Reconfigurable Energy Subsystem 77 remaining expressed as a percentage where 100 is fully charged 0 is when the store is at the minimum voltage and 50 is half way along the time axis between the two values with the simplifying assumption that the capacitor is discharged through a resistive load Clearly a value for the energy proportion is not particularly helpful if the load across the store behaves as a current or resistance dominant consumer as in these cases the power requirements of the node vary with the amount of energy stored and hence the store voltage The energy fraction may be calculated by means of Equation 3 1 2 B min SRT max E min E 3 1 Where E is the amount of energy stored by the system at a given time min is the energy at the store s minimum voltage and max F is the energy at its maximum voltage This gives a rough idea of the energy status of the node but is reliant on the assumption that the load is a power dominated consumer 0 030 Me X 0 09 0 025 S 0 08 xo 8 S 0 07 02 x 7 SS pE 0 06 a A D eee z 0 015 A 0 05 Y 5 ee S O SH 0 04 0 010 0 03 0 02 0 005 Current A 0 01 Power W 0 000 0 00 27 22 23 24 25 26 27 28 29 30 31 32 33 34 35 3 6 Voltage V FIGURE 3 6 Current and power consumption of CC2430EM at various supply voltages Investigations carried out under t
228. pply voltage from the USB interface by default the USB interface provides a regulated 3 6V supply to the device In many cases it was also necessary to connect the mains module to the energy subsystem in order to guarantee a certain power supply to the device for Chapter 5 Case Study Deployment in a Prototype System Software 123 debugging purposes halting a device in its active state for debugging uses a relatively high level of power especially when the transceiver is active which quickly depletes the energy stored in a supercapacitor However both systems have also been tested in a stand alone mode operating from harvested energy ZP S a e a E Chipcon FIGURE 5 1 SmartRF04EB programming debugging platform used to interface with the CC2430EM 5 2 3 Low power modes and energy characteristics The TI MSP430 has five low power states in addition to its active mode In the deepest low power mode the CPU and all clocks and peripherals are disabled meaning that the system can only be woken by an external interrupt In shallower sleep modes certain peripherals such as clocks and DC DC converters are disabled The depth of sleep has implications for the length of time in transitioning to the active mode and so the overall power consumption and responsiveness of the processor A comparison of the current draw of the various power modes is shown in Figure The MSP430 module used in this project has a separate transceive
229. pport up to six simultaneously connected energy devices energy sources or stores It allows each device to be individually monitored by the microcontroller in order to ascer tain the amount of energy stored or power generated and where appropriate permits the device to be managed Energy modules can be attached to and removed from the system in a plug and play manner and incorporate electronic data sheets that store op 151 152 Chapter 6 Conclusions and Future Work erational data for each module The embedded software on the sensor node is structured to allow the system to interface flexibly with a range of energy devices and to present a single interface to the application running on the sensor node The contributions of this research are threefold firstly the system is enabled by a new hardware interface between the energy devices and microcontroller secondly an embedded software structure was implemented to interface with the energy hardware and thirdly energy aware modules compliant with the scheme have been produced The scheme has been evaluated by way of a prototype which accommodates a range of energy devices The end result is an energy subsystem for micropower wireless sensor nodes that supports a range of energy devices and enables energy aware operation The proposed scheme was described in detail in Chapter 8 which outlined the features of the energy electronic data sheet common hardware interface and the associated
230. proximately 50s and that the regulated voltage continues to rise until it is regulated to 3 0V The raw voltage reached a maximum of 4 5V before the mains supply was manually turned off and the system was discharged through a 1800 resistor The regulated voltage follows the decay of the raw voltage until it falls below 2 0V after which time the microcontroller is disconnected from the supply and the output falls to approximately 0 0V The test system is able to operate autonomously as a sensor node In the second test with the Hyperterminal output shown in Figure 5 9 being taken from the PC connected to the receiver node the system was charged up from cold using the photovoltaic mod ule which took less than one hour under an indoor light intensity of 900 Lux in the configuration shown earlier The system was then reset and the initial sweep of the energy subsystem resulted in the first group of transmissions of the serial number of each of the energy module EPROMs As the voltage on the supercapacitor is 2 8V the system is in PP_3 and wakes up every two seconds to perform a re scan of its energy hardware Approximately 20s after start up the vibration module was connected to socket 5 of the multiplexer module This resulted in an additional transmission showing 148 Chapter 5 Case Study Deployment in a Prototype System Software that a device had been added to socket 5 Approximately 10s after this the photovoltaic module was
231. r The hardware abstraction layer will be different for each type of microcontroller however this enables the rest of the en ergy stack to remain hardware agnostic The mechanism for reading the electronic data sheets on energy modules and storing them in memory means that the microcontroller can rapidly access information on the modules for use in computation The energy stack provides the shared application layer with a standardised interface of energy priority values and a number of other interfaces that allow the energy status of the node to be calculated monitored and manipulated 5 6 4 Comparison against state of the art systems The software architecture described and evaluated in this thesis offers a number of key benefits over existing systems In particular the software implemented on the limited number of state of the art systems which incorporate multiple energy resources is highly tailored to specific types of energy device While the existing solutions deliver a limited Chapter 5 Case Study Deployment in a Prototype System Software 147 amount of energy awareness the act of altering the energy hardware which is connected to these devices also necessitates adaptation of the embedded software This entails the parameters of the new energy hardware being obtained and the embedded software being altered accordingly Depending on the characteristics of the device this may need new routines to be coded to enable the calcula
232. r and the power consumption of the radio in receive mode is often equal to or higher than its consumption in transmit mode For this reason in order to conserve valuable energy it is essential to sleep the radio rather than leaving it in receive mode when idle In this project the MSP430 switches between active mode and LPM3 and is generally configured to be woken by its sleep timer which is controlled by its very low power 32kHz oscillator The CC2430 has a comparable current draw to the MSP430 in its active and sleep modes The system developed under this project leaves the CC2430 device in power mode 2 which similar to the MSP430 LPM3 leaves the 32kHz oscillator active and draws less than 14A allowing the device to be woken by its sleep timer The CC2430 has an integrated radio transceiver and this is also stopped whenever the device is not actively transmitting 124 Chapter 5 Case Study Deployment in a Prototype System Software Active Mode 8MHz Active Mode 1MHz Active Mode 100kHz LPM 0 1MHz LPM 0 100kHz LPM 2 LPM3 Low Frequency Crystal LPM3 Very Low Power Oscillator LPM4 0 0001 0 001 0 01 ol 1 10 100 1000 10000 Current Consumption uA FIGURE 5 2 Current draw of the Texas Instruments MSP430F 2274 in its various power modes Data obtained from 127 5 2 4 Hardware abstraction layer functions As part of the embedded software structure implemented under this project it was nec ess
233. r as in Figure 3 7 123 and mapped to straight lines through a piecewise linear approximation to simplify implementation in microcontrollers as shown in Table TEST CONDITIONS 70 F 21 C ORNS mA 62 20 VOLTAGE V 0 20 40 60 80 100 120 140 160 SERVICE HOURS FIGURE 3 7 Typical discharge profile of Duracell MN1500 alkaline cell under various impedances 123 Point Voltage V Service Hrs Rem Lifetime 1 1 52 0 100 2 1 37 15 86 3 1 22 79 27 L 1 00 108 0 4 0 80 140 29 TABLE 3 8 Simplified discharge profile Duracell alkaline AA cell through resistive load corresponds to 622 discharge curve Any discharge curve can be approximated to a series of straight lines and stored in memory The choice of curve will depend on the type of load normally attached to the battery It must be noted that use of a resistive discharge curve does not imply that the microcontroller behaves as a constant resistance at all times purely that its dynamic power consumption due to different voltages varies in line with that of a resistor The interrogation process is straightforward but imposes a further energy requirement on the battery For example if an alkaline AA battery is tested once per hour for one year and each test takes one second with a 10Q load this will consume around 2 2 of the battery s total capacity The system designer must trade off
234. r being generated by the device This calculation was timed as taking 7 16ms to complete Being the most demanding calculation it can be inferred that the energy or power calculations for the remaining modules will take place more quickly However as a worst case one can state that the ADC measurement and associ ated calculation for each module will complete within a 10ms period Given that there are potentially six modules connected to the system it follows that the update of the energy status of the node will be completed within 60ms Given that this refresh will be carried out periodically this imposes a minimal resource requirement on the sensor node The update rate can be tailored to take account of the last known EP value of the node in order that the system update frequency can adapt to the available energy and deliver an increased update rate when energy is plentiful which is desirable as in this state it will normally be used more rapidly 5 6 3 Embedded software features The embedded software stack has been implemented to be applicable to a range of mi crocontrollers The basic hardware requirements have already been stated however the full system has been implemented on an MSP430 microcontroller with a minimal amount of RAM and flash The system incorporates a full communication stack and energy stack The interface with the pins on the microcontroller and with its internal ADC is enabled through a hardware abstraction laye
235. r signal processing gaining emphasis Indeed one could say that sensing and power management for self powered sensors justify having their own individual stacks rather than being forced into the top layer of the communication stack A notable system which breaks with this conventional structure is TinyOS 2 0 which has separate interfaces based on the type of peripherals it interfaces with and is discussed in detail in Section 2 8 2 Alternative systems for structuring the energy management capabilities of the sensor nodes are outlined by Jiang et al with their Energy Management Architecture EMA 101 which is shown in Figure 2 25 The scheme permits users to set policies about the operation of the sensor node meaning that the sensor node can manage its resources to achieve these aims Examples are given for a range of deployment types for example an Arctic monitoring scenario prioritises tasks in the following order 1 a one year network lifetime 2 a 1Hz sensor sampling frequency 3 mesh networking data storage and 4 maximised sampling rate 2 8 2 Operating systems The TinyOS operating system is open source and designed for wireless embedded sen sor networks It is geared towards Crossbow motes and written in nesC 102 It has a component based architecture and is claimed to make the design of wireless sensor network solutions straightforward It is not however well respected in the wider com munity with it bein
236. record of one being connected the function also returns a 1 The function returns 0 if no changes are detected e pyeRefresh is similar to the pyeRescan function but instead of simply issuing a reset command and listening for a response it also queries the device ID of each 138 Chapter 5 Case Study Deployment in a Prototype System Software energy module connected to the multiplexer If a difference is detected from the ID recorded in the microcontroller memory the function returns a 1 otherwise it returns a 0 The pyeRefresh or pyeRescan functions are expected to be called periodically in order to check for changes in the energy subsystem They both allow newly connected devices to be recognised and similarly allow the disconnection of devices to be detected A change triggers a full re scan of the energy hardware of the system in order to update the EEDS table held in the memory of the microcontroller An important distinction between the functions is that pyeRescan just checks for the presence of modules at addresses on the multiplexer module while pyeRefresh actually checks that the modules have the same ID code as those recorded in memory Hence pyeRefreah is a more robust function as it will detect if modules have been swapped on a port since the previous update however it consumes more energy as it invokes a read of the 1 Wire memory on each port as opposed to a straightforward reset and i
237. refore prompt further interrogation Functions have also been written to permit bits bytes and blocks of data to be read from the 1 Wire memory and for command bytes to be written to the bus These can prompt responses from the 1 Wire device including its device ID or allow specific memory addresses to be read Due to the presence of a pull up resistor to enable the 1 Wire bus the microcontroller normally leaves the pin used for 1 Wire communications in its high impedance input state Communications from the microcontroller are actuated by the microcontroller changing the status of the pin to a low output as opposed to a high impedance input which pulls the line low Responses from the 1 Wire device will also pull the line low e onewireDelay2us is a delay function that implements a multiple of 2us delay which is necessary due to the time sensitive nature of 1 Wire communications This function is trimmed for specific microcontrollers and clock speeds e onewireReset sends a reset signal over the 1 Wire bus If a presence pulse is detected i e if a 1 Wire device responds to the reset signal the function returns a non zero result e onewireWriteByte writes a byte of data to the 1 Wire bus which will typically be a command as data cannot actually be written to the EPROM memory in this system due to the high voltages required for this operation e onewireWriteBit writes a single bit to the 1 Wire bus It is generally called by
238. rement and control of the energy subsystem to take place This board interfaces directly with the microcontroller providing its regulated power supply and having a number of additional interface pins to facilitate communication with the microcontroller The multiplexer module also has an EPROM memory which can be interrogated by the microcontroller storing basic information such as how many sockets the multiplexer module has and what voltages it supports 2 Multiple energy modules which may be used for either the generation or storage of energy An energy module includes the actual energy device and its power conditioning and management circuitry It also has an EPROM memory which stores its operating parameters which is readable by the microcontroller These parameters permit the microcontroller to interpret measurements from the module in order to estimate the amount of power being generated or energy stored and also to learn how the device may be controlled if applicable The energy modules are connected to the multiplexer module through sockets the power outputs from Chapter 3 Development Towards a Reconfigurable Energy Subsystem 59 the energy modules are effectively connected together through the multiplexer module thus forming a common raw or unregulated voltage rail 3 A microcontroller that interfaces with and obtains its power supply from the energy subsystem There is a certain resource commitment from the mic
239. resentation to application layer Network Software modules for communications functionality Energy Control High level view of energy subsystem Medium Access Control Control of message transmission over radio channel Energy Analysis Calculation of stored energy using models Sensor Processing Compensation for offset and gradient d E E wbd bi 2 bel E E D Energy Management Intelligent Sensing FIGURE 5 5 A combined stack comprising stacks for communications energy man agement and sensing initialise radio radioInit frequency myAddr send data radioSend sendBuffer sizeof sendBuffer remoteAddr DONOT_ACK STOP_RADIO LISTING 5 6 Code to interface with Simple Packet Protocol communication stack CC2430EM Simple Packet Protocol The communication stack implemented in the system is based on the Texas Instruments Simple Packet Protocol SPP The protocol permits data payloads of up to 125 bytes and supports source destination addressing and broadcasting along with sequence bits and acknowledgements Data frames received through the SPP stack have a received signal strength indicator RSSI byte The communication system is set up through the radioInit command shown in Listing 5 6 In the demonstrator transmissions are made on a fixed channel and are unacknowledged This minimises the amount of time that the transceiver must be active for and thus reduces the
240. retention capacitor The current consumption in the case that the microcontroller and push button are in conflict would be below 5004A If the user erroneously pushes both buttons simultaneously the resulting current would be below 2004A and the status control would revert to either the on or off state after the buttons are released sai burns arge FIGURE 4 17 Circuit board for the primary battery module The circuit facilitates two types of state of charge determination its open circuit volt age or its closed circuit voltage across a known load The measured parameter is buffered by an operational amplifier in a unity gain buffer arrangement The type of test is controlled by the value of resistors in the voltage divider large MQ scale resistors will approximate to open circuit while smaller resistors kQ scale will apply a fixed known load across the cell At the time of the test it is important that the discharge output is off to ensure that readings are accurate Table 4 9 shows the management Chapter 4 Case Study Deployment in a Prototype System Hardware 115 connections from this module Pin Type Function 2 Meas Control Connect battery to voltage divider 3 Measurement Analogue battery voltage through fixed load 4 Control Enable discharge from battery 5 Control None TABLE 4 9 Interface pins from primary battery module Performance evaluation The provision
241. returns a 1 otherwise it returns a 0 This function is useful to verify the Chapter 5 Case Study Deployment in a Prototype System Software 139 state of the bistable multivibrator on energy storage modules which may or may not be overridable using buttons on the module pyeSetChargeStatus sets the charge status of the module which will be retained by the bistable multivibrator on the energy module e pyeGetDischargeStatus reads the discharge status of the module if discharging is enabled it returns a 1 otherwise it returns a 0 This function is useful to verify the state of the bistable multivibrator on energy storage modules which may or may not be overridable using buttons on the module e pyeSetDischargeStatus sets the discharge status of the module which will be retained by the bistable multivibrator on the energy module Together they enable the physical interface with the energy devices permitting the charge and discharge of energy storage devices to be enabled and disabled and mea surements to be carried out in order to assess the energy status of the node Flash write requirements An intricacy of interfacing with the flash memory on microcontroller devices is that the supply voltage must be above a certain voltage frequency threshold as shown for the MSP430 in Figure 5 7 for flash write operations to be carried out reliably This imposes an important constraint on the system as th
242. rgy devices to ensure that they are operating correctly and as designed In the scheme outlined in this thesis the EEDS stores the operating parameters of the energy modules in order that the amount of stored energy or generated power can be measured Typically the method for determining the amount of power being generated will use one of the following methods 1 Closed circuit Current Measurement with the harvester or normally the rectified regulated output from the harvester connected across a known typically resistive load the power delivered across that load can be calculated Clearly this is dependent on the assumption that the power being delivered by the harvester across the test load is similar to that being delivered to the actual load 2 Open circuit Voltage Measurement with the harvester not its regulat ed rectified output disconnected from any load its open circuit voltage can be measured From this the nominal output power may be calculated This is most suitable for harvesting devices with a DC output such as photovoltaic cells or ther moelectric generators but the absence of any load means that the actual output power must be estimated in software 3 In line Current Measurement this method uses a small in line resistor to measure the actual amount of current being delivered by the harvester to the load While this is theoretically the most accurate type of measurement as it reflects the actual amoun
243. rminal output from second test Further experimentation has been carried out with the complete energy stack and mul tiple connections to energy modules While this system level testing has not evaluated the accuracy of energy estimates of each module per se it allowed measurements of the processing time to be carried out as presented earlier in this chapter and for the operation of the overall system to be observed Sanity checks were carried out with the energy and power estimates being checked to ensure that they were within reasonable ranges further the system was used to power the microcontrollers starting up from cold and operating from a range of energy modules Modules were exchanged during the system s operation and estimates of energy and power were automatically revised to take account of the hardware changes Chapter 5 Case Study Deployment in a Prototype System Software 149 5 8 Summary This chapter carried forward the case study outlined in Chapter 4 by describing and evaluating the software interface for the system which is co located with the communi cation stack on the system s microcontroller The microcontroller platforms used in this investigation along with their important features of relevance to this work were de scribed The implementation of the EEDS for the deployed modules has been outlined including examples for the data stored in the modules and detail on the implementation and energy costs of
244. rocon troller in order that it may effectively manage its energy subsystem for the pur poses of the system developed in this project the requirement is the equivalent of eight I O pins with at least one being connected to the internal ADC There is also an impact on the embedded software of the node unlike conventional systems a dedicated software stack interfaces with the energy hardware While these are extra resource requirements for an already resource constrained system they en able a number of benefits including the ability to react to changes in the energy hardware and allow the system to be reconfigured in a plug and play manner The requirements of the proposed scheme can realistically be achieved with common microcontrollers used in wireless sensor nodes Modern communication stacks such as ZigBee already impose a substantial overhead in terms of memory code size and proces sor time and the proposed energy scheme has a low level of complexity in comparison While this scheme requires an additional circuit board the multiplexer module and a standardised way of interfacing with the energy devices it is anticipated that in the future should the scheme become widely adopted it will be trivial to integrate the multiplexer functionality onto the same PCB as the microcontroller While there is a requirement for a standard voltage and communication scheme between the energy modules and the microcontroller there is l
245. rough a fixed load are used by this system The voltage measurement is normally divided by means of a voltage divider in order to bring it into the acceptable range for the microcontroller The resistor values on the voltage divider are set to provide an appropriate level of impedance to the measured device e g for open circuit measure ments the impedance will be very high to simulate open circuit for closed circuit they will be smaller and appropriate for the load the device has been modelled at The parameters stored on the electronic data sheet on the module enable conversion between the measured voltage and the estimate of power output or energy stored An example of a voltage test circuit is shown in Figure 4 6 For purposes of clarity pull up resistors are not shown in this schematic It may be observed that the operational amplifier is in a unity gain buffer arrangement and its power supply is the regulated microcontroller supply Its power supply will normally be controlled by Mu in order to reduce the quiescent power consumption of the system but this is not shown in the diagram The voltage divider circuit is switched by the Maa line but the effective impedance of the MOSFET M1 is negligible compared to those of Ry and Ra The output Chapter 4 Case Study Deployment in a Prototype System Hardware 101 Vin Vio IRLML 6401 TAL R1 3M9 vap uae o V meas 1 TS941 ANA
246. rther features such as sensor processing and energy management to what was previously just the communications module This has added pressure to integrate the embedded software for interfacing with further external devices into the communication stack The result of this is a relatively complex and unstructured stack with the application layer hosting a number of disparate functions The reader can refer back to Figure 2 20 which shows the ZigBee stack with the application layer being subdivided into the application framework application support sublayer and ZigBee device object Chapter 2 Background Energy Sensing and Wireless Communication 47 Application Application e g HTTP FTP Presentation Application Application ER Session FMS and FAS Transport e g TCP UDP Network Internetworking Network e g IP R MAC Data Link H1 Data Link 802 15 4 Physical 802 15 4 Transport Physical H1 Physical OSI Basic Internet Reference Foundation Reference Model Model Fieldbus H1 FIGURE 2 24 The OSI BRM IRM Foundation Fieldbus H1 and ZigBee communica tion stacks reproduced from Merrett et al 100 ZigBee The increasing complexity of the communication stack is a result of evolution rather than forward planning It may be argued that communications software stacks are becoming less important with moves towards implementing many communication tasks in hardware with operations such as senso
247. rvesting devices it is quite an inelegant solution each energy device has to have a go between PCB which then connects to the multiplexer module Instead it may be desirable for each energy harvesting device to have its power conditioning cir cuitry integrated into its package so that a simple connector could go from the energy module meaning the energy harvester and its interface circuitry as one block to the multiplexer module without the need for a go between This could interface directly with the stacking system proposed above Chapter 6 Conclusions and Future Work 155 6 3 3 Software development The work carried out thus far has focussed on the development of the energy stack while the sensor stack was proposed development work carried out has been minimal indeed the sensing application targeted by the case study was straightforward In future work it may be useful to look at how the sensing stack can be developed to deliver a similar plug and play capability for the sensing hardware of wireless sensor nodes Indeed the sensing stack could provide standard functions for data processing and event detection and could conceivably be controlled through a simplified interface by the application layer with the application layer only having minimal involvement in the collection and processing of data The application layer could then act as a simple go between and scheduler acting to collect data from the sensing stack and move i
248. s 227 38 2004 S Bapat V Kulathumani and A Arora Analyzing the yield of ExScal a large scale wireless sensor network experiment 13th IEEE International Conference on Network Protocols page 10 pp 2006 200 BIBLIOGRAPHY 108 109 110 111 112 113 114 115 116 117 118 K Langendoen A Baggio and O Visser Murphy loves potatoes experiences from a pilot sensor network deployment in precision agriculture Proceedings 20th International Parallel and Distributed Processing Symposium page 8 pp 2006 F Chiti A De Cristofaro R Fantacci D Tarchi G Collodo G Giorgetti and A Manes Energy efficient routing algorithms for application to agro food wire less sensor networks 2005 IEEE International Conference on Communications 5 3063 7 2005 D Estrin Reflections on wireless sensing systems from ecosystems to human systems 2007 IEEE Radio and Wireless Symposium pages 1 4 2007 V Raghunathan A Kansal J Hsu J Friedman and M Srivastava Design considerations for solar energy harvesting wireless embedded systems 2005 Fourth International Symposium on Information Processing in Sensor Networks pages 457 62 2005 P Corke P Valencia P Sikka T Wark and L Overs Long duration solar powered wireless sensor networks Proceedings of the 4th Workshop on Embedded Networked Sensors EmNets 2007 pages 33 37 2007 J Eliasson P Lindgren J D
249. s case the device type field identifies it as a secondary battery module which can be controlled by the microcontroller to enable 132 Chapter 5 Case Study Deployment in a Prototype System Software typedef struct _TParamComplex TFloat a TFloat b TFloat c unsigned char d TParamComplex typedef struct TPrimaryParams unsigned char EndOfLife TUInt MaxCapacity TPrimaryParams typedef struct _TParamCurve TParamCurveSingle Point 4 TUlnt TotalEnergy TParamCurve typedef struct _TParamCurveSingle unsigned char HalfVolts unsigned char Percent TParamCurveSingle LISTING 5 5 Parameters to enable energy awareness for energy modules both charging and discharging It states that the measurement of the state of charge of the device is via a curve whose points are given e g 100 charge corresponds to a voltage of 2 88V double 1 44V It also shows that its maximum voltage is 2 9V and that the measured values from the ADC in the range 0 1V must be multiplied by 4 9 to arrive at the actual battery voltage NiMH Battery Module Device Type 01 0010 11 0x47 Measurement Curve 0x06 Multiplier 4 9 49 0x31 Max Output 2 9 29 0x1D Total Energy 6912 0x00 00 1B 00 144 100 0x90 0x64 f 132 88 0x84 0x58 SE E SE 100 0 0x64 0x00 TABLE 5 1 Electronic Data Sheet for NiMH battery module Similarly an example of a data sheet for the photovolt
250. s conversion generator circuitry with 3 6 V output S Displacement A weight e _ Displacement well RLP smart energy beam Power A Device B Components FIGURE 2 10 The AdaptivEnergy piezoelectric vibration energy harvester a in its packaging and b disassembled showing its components reproduced from AdaptivEn ergy promotional materials 38 Electrostatic Electrostatic based vibration energy harvesting requires two conductors moving relative to each other and separated by a dielectric effectively acting as a capacitor Move ment in the conductors causes the amount of energy stored in the capacitor to change which can be exploited to extract energy There are two main modes of energy con version charge constrained and voltage constrained The two methods are compared by Meninger et al 40 with the conclusion that charge constrained conversion delivers less energy by a factor of Vstart Vmax where Vstart is the initial voltage applied and Vmax is the maximum voltage reached although there are hybrid alternatives which may deliver improved efficiency Figure 2 11 shows the operating principles of the three main types of electrostatic generator The in plane gap closing type is believed to be the most efficient 29 The performance of this device is expressed by Equation which gives the voltage V stored across a rectangular parallel plate capacitor Here q represents the charge on the capacitor d is the distan
251. s its performance at a range of wind speeds reproduced from 45 c is an off the shelf turbine used by Park et al reproduced from 46 2 3 6 Human power Human power may be categorised as being either active or passive An actively pow ered device requires the user to do a form of work that would not otherwise be done specifically in order to power the device Conversely a passively powered device har vests energy from the user s normal bodily motion and properties not requiring any additional work Among the earliest human powered wearable devices have been wrist watches for example the Seiko Kinetic range harvests power from arm movement while Seiko s Thermic technology can power the wristwatch from the temperature difference between the skin and the surrounding air by means of the thermoelectric technology discussed in Section 2 34 The ability to harness thermal energy at any point on the body adds an element of flexibility in that a device need not be located at a point of maximum movement There are drawbacks however in that the skin may naturally act to restrict blood flow to areas in contact with cold objects which obviously causes the skin temperature and hence the magnitude of the heat gradient to decrease and Chapter 2 Background Energy Sensing and Wireless Communication 29 limit the amount of energy that can be harvested It may be observed that the most promising and potentially unobtrusive source of en
252. s of their memories and frequent resets led to excessive amounts of network traffic in some areas Another sensor network deployment powered from light is reported by Voigt et al 82 which also has a focus on energy aware routing and network management Separately Corke et al describe a long term deployment of solar energy harvesting nodes based on batteries and conclude that supercapacitors are essential for delivering extended system lifetimes Eliasson et al present a system which harvests energy by means of a photovoltaic cell with energy being buffered in a supercapacitor and occasionally switches in a non rechargeable battery to provide power for high current drain high priority tasks Further developments on photovoltaic energy harvesting motes include the HydroWatch system at University of California Berkeley which attempts to consider the impact of various circuit elements during the development of the system 114 Several prototype systems incorporating vibration energy harvesting have been devel oped For example the S NAP 6 pictured earlier in Figure uses a commercially available electromagnetic vibration energy harvester to power an accelerometer based condition monitoring system In a separate development shown in Figure 2 8 a micro generator with a volume of less than 1cm has been developed in conjunction with a custom sensor node 36 In both systems energy harvested from vibrations is buffered in supercapac
253. s temperature sensor sensing formatting messages and transmitting them through the communication stack at a duty cycle appropriate for the energy status 5 5 Energy stack 5 5 1 Physical Energy PYE layer Datasheet functions The layer implements two separate functions for reading in data from the EEDS on energy modules and the multiplexer module The reason that two functions are needed is that the structure of the multiplexer EEDS is significantly different to that of the other energy modules and the data from them may be stored in different locations on Chapter 5 Case Study Deployment in a Prototype System Software 137 the microcontroller In the case of the MSP430 implementation of the system data on the energy modules is stored in flash while data on the multiplexer system is stored in RAM This is due to the limited amount of RAM available on the MSP430 node compared with the amount of flash e pyeGetMuxData sets the address as 7 and interrogates the 1 Wire device on the multiplexer module Data about the multiplexer module are stored in a struct as defined in Listing 5 3 e pyeGetModuleData erases the stored EEDS for all energy modules and sequentially interrogates each energy module from address 1 to the last address maximum 6 The data are read into structs as defined in Listing 5 5 These functions utilise the procedures outlined in Section 5 3 to interrogate the 1 Wire memory EPROMs and store d
254. s that the load responds to changing voltages in a similar way to a power source For a node with a given activity level the power drawn is constant but the current drawn depends on the relationship J P V Therefore at 3V the node draws 33 less current than at 2V 3 6 4 Supercapacitor state of charge and capacity There will typically be a range of voltages over which a microcontroller can operate for the CC2430 this is between 2 0 and 3 6V Obviously if such a microcontroller is driven directly from an energy store the energy stored when the voltage is below 2 0V is effectively unusable For a capacitor the following methods for determining the energy status p the fraction of operating time remaining compared to when the store is full are applicable and largely rely on the relationship E SCHT 1 Energy fraction pg this is simply a measure of how full the energy store is and is most useful when the load imposes a power dominant requirement on the store It is calculated by determining the usable energy stored and dividing by the maximum usable energy if the store was full 2 Voltage fraction yy this value assumes that the load imposes a current dominant requirement on the store It is calculated by dividing the usable voltage range by the maximum usable voltage range 3 Logarithmic discharge fraction er this value assumes a resistive dominant load is present across the store It gives an indication of the pro
255. sely secondary bat teries can be recharged and offer broadly similar energy densities to those of primary batteries however they are sensitive to the number of charge discharge cycles to which they are exposed and must be charged in a controlled manner to maximise their life time Finally supercapacitors offer an energy density which is approximately an order of magnitude lower than those of batteries exhibit capacitor like behaviour and are relatively insensitive to the method of charge or discharge The choice of energy storage technology is highly dependent on the other considerations for the energy subsystem of the sensor node but due to their favourable characteristics supercapacitors are most commonly used in systems which feature energy harvesting technology 18 Chapter 2 Background Energy Sensing and Wireless Communication 2 3 Energy harvesting and wireless power transfer 2 3 1 Motivations and overview In this thesis energy harvesting also known as power harvesting or perhaps most appropriately as energy scavenging is defined as being the process in which electrical energy is produced through the exploitation of naturally occurring environmental or hu man energy Devices with a finite fuel supply such as miniature fuel cells fall outside the scope of this section and have already been considered in Section 2 2 The technolo gies behind the harvesting of power from light vibration and thermal di
256. system from a primary battery when it is first installed In this situation it is important that the microcontroller is able to both read and set the status of the module and justifies the 98 Chapter 4 Case Study Deployment in a Prototype System Hardware control lines being bidirectional The quiescent power consumption of this block can be minimised by choosing a micropower comparator and using high value resistors With an LMC7215 micropower comparator configured as shown in Figure 4 1 the quies cent current draw of this circuit was found to be 1 8uA at 3V This compares favourably with the data sheet current draw 0 7uA of the comparator alone The circuit was verified as operating without oscillation and starts up in the off position This circuit configuration is used for many of the modules described later in this chapter As an aside it must be ensured that the that the state retention capacitor is suitably large to overcome other capacitances in the control circuit which includes the multiplexer module and microcontroller otherwise the act of interrogating the module by the mi crocontroller may cause the output state to be toggled The correct operation of this block with an MSP430 and CC2430EM was verified 4 3 5 Over Undervoltage protection and regulation As discussed in Section 1t is essential to provide undervoltage protection to the microcontroller to ensure that it starts up correctly and overvoltage prote
257. t Towards a Reconfigurable Energy Subsystem layer deals with processing of this information and the management layer deals with high level processing For the example of the energy management stack the physical layer carries out switching to take measurements and control the energy modules the analysis layer translates those measurements using the device models and the manage ment layer takes this information to provide high level data to the application The chosen architecture permits modules in each layer to be swapped without affecting the rest of the stack layers For example a highly developed management layer may be implemented on more capable microcontrollers which may enable prediction to be used to dictate the behaviour of the node Shared Application Management Layer Medium Layer Interface Layer FIGURE 3 9 A basic template stack reproduced from 124 o s o E o ke x a E Ka 3 7 2 The Energy Stack Overview The energy management software tasks are modularised and arranged into the energy stack shown in Figure 3 10 which resides on the microcontroller Here the energy management process is divided into three layers with an emphasis on modularisation and re usability e Energy Control Layer ECO takes a high level view of the energy subsystem presenting information about the overall energy status of the node and responding to queries It can make decisions to maintain the
258. t across to the communication stack dependent on the amount of energy available indicated by the energy stack This would deliver a true plug and play solution for sensor nodes It may also be interesting to look at the network level interactions between sensor nodes with the complete plug and play system Mechanisms could be put in place for them to co ordinate their sensing tasking and scheduling in order that they can make best use of their individual sensing hardware and energy resources This would require the development of network interaction capabilities which could conceivably be devolved to their own stack along similar lines to the energy and sensing stacks that have been proposed under this work 6 3 4 Towards standardisation At the time of writing this thesis a proposal was out for ballot with the International Society of Automation ISA to form a Power Sources Working Group under the ISA100 family of standards 129 The aim of the provisional working group is to develop standards to enable users to compare specify and interface power energy sources for non line powered low power wireless sensor nodes The ultimate aim is to promote the interchangeability of energy devices for sensor nodes In short this will standardise the quoted power capabilities of energy sources including energy harvesting devices and batteries and the quoted power requirements of sensor nodes The standard will also look a
259. t mains and transmitted power supply options A deliverable may also be a common connector standard although the number of pins required is currently open to debate and is related to the expected capabilities of connected system with regard to the energy awareness and device management capabilities The aims of the proposed standard have a number of parallels with the work described in this thesis The fact that industry has recently realised that interchangeability and interfacing between wireless sensor nodes and their energy resources is important is in effect a validation of the original motivation for this work Indeed the system described 156 Chapter 6 Conclusions and Future Work in this thesis goes beyond what is being proposed in that it introduces an electronic data sheet concept and provides a method for multiple energy resources to be connected to a single sensor node in a plug and play manner It is hoped that as this standard develops the work carried out under this thesis can feed into discussions to enable the capabilities realised under this prototype system to be transferred to commercial applications 6 4 A look to the future In the last few years there has been a rapid pace of development in energy harvesting and storage leading to an increase in the amount of power than can be generated and the efficiency with which it can be stored along with a continued march of integrated circuit development leading t
260. t of power being delivered it has a number of drawbacks Firstly the efficiency of the system will be affected if the sense resistor is left in line at all times or complex switching hardware must be used to enable the connection of the sense resistor during measurements Secondly the resistor must be small in order to minimise the effects of the resistance this means that the resulting voltage measurement will be small Thirdly the resistor must be connected on the positive supply line meaning that the measured voltage will be differential again requiring additional processing and introducing additional error Chapter 3 Development Towards a Reconfigurable Energy Subsystem 81 3 7 Software structure 3 7 1 Overall software structure Given the aim of developing a consistent scheme for resource management for wireless sensor nodes it follows that the embedded software should have a defined structure Section gave a justification for the development of a software structure in which energy and sensor management are devolved into their own individual stacks rather than being integrated into the application layer of the communication stack A software structurd has been developed comprising separate communications energy and sensor stacks linked through a shared application layer and is shown in Figure A consistent solution has been presented in which each stack features interface medium and management layers in li
261. t they only draw current when measurements are in progress In addition the operational amplifier is powered from the regulated digital line provided by the multiplexer module while the voltage divider takes the rectified output from the generator this works to limit the output from the module to within the bounds of the regulated digital supply of the microcontroller in case of malfunction 112 Chapter 4 Case Study Deployment in a Prototype System Hardware Pin Type Function 2 Meas Control Connect generator to fixed load 3 Measurement Analogue measured voltage across fixed load 4 Control None 5 Control None TABLE 4 5 Interface pins from vibration module Impact of energy awareness circuitry Unlike the example of the photovoltaic module in order to deliver energy awareness there is no requirement for additional in line circuitry between the generator and the output from the circuit The enable input to the switching converter is used to disconnect the generator from its normal load in order that measurements can be carried out Therefore the only impact of energy awareness circuitry on this circuit is apparent when measurement activities are in progress and these are only likely to take place every few seconds or minutes and take a few milliseconds to complete 4 5 3 Other energy harvesting modules The wind energy harvesting module full wave rectifies the turbine output which is then buffere
262. tage Modules Pin 7 Detector Sp Meee Detector 1 to 6 Ping Energy E S7 Meas Ctrl Voltage ta Modules RawV S Meas Regulator 1 to 6 Meas Pin 8 uC Pin 10 FIGURE 4 9 Simplified schematic of multiplexer module showing power connections Pin numbers refer to Common Hardware Interface pin numbers 104 Chapter 4 Case Study Deployment in a Prototype System Hardware Pin Type Function 2 Meas Control None 3 Measurement Digital threshold crossing detection A Control None 5 Control None TABLE 4 1 Interface pins from multiplexer module address 071 Pin Type Function 2 Meas Control Perform raw voltage measurement 3 Measurement Analogue raw voltage measurement 4 Control None 5 Control None TABLE 4 2 Interface pins from multiplexer module address 7 4 4 2 Energy awareness circuitry In order that the microcontroller can monitor the unregulated up to 4 5V voltage on the multiplexer module a resistor divider and operational amplifier based unity gain buffer arrangement is provided This brings the potential 0 4 5V unregulated voltage range into the 0 1V range supported by most microcontroller ADCs This facility is managed and monitored through address 7 and allows the microcontroller to moni tor the unregulated voltage on demand There is also a resistor providing the pull up functionality for the 1 wire bus The
263. tage approximately 6004A will be dissipated This is clearly a substantial level of power but given that the line is pulled low for relatively short periods the overall energy requirement is minimised In order to further minimise the energy demand from the system the entire data sheet is read from each device and stored in memory on the microcontroller This means that the parameters can be accessed by the energy stack at any time without delay or further energy cost The timings of the 1 Wire operations used by the system are with recommended timings shown in brackets 128 e Reset Hold bus low for 480 640us 480s sample bus after 63 78us 70s wait for at least 410s 410ys e Read Bit Hold bus low for 5 15us 6s sample bus after 5 12us 9yus wait for at least Blus 55 us e Write 1 Hold bus low for 5 15us 6s wait for at least 59us 64ys e Write 0 Hold bus low for 60 120us 60us wait for at least Bus us For the system implemented here the following process is followed to read the device ID and EEDS data which are then stored in memory on the microcontroller 1 Reset bus and check for response If a response is received proceed 2 Write Byte 33h read ROM command 3 Read Byte family code of 1 Wire EPROM discard 4 6 x Read Byte save as six byte device ID and store in microcontroller memory 5 Write Byte CCh skip ROM command 6 Write Byte F0h read memor
264. tatus will however be off 3 5 3 Microcontroller requirements To provide the interface with the energy subsystem the microcontroller needs to have eight I O pins available One of these must be configurable as an ADC input four Chapter 3 Development Towards a Reconfigurable Energy Subsystem 73 as digital outputs and a further three as bidirectional reconfigurable digital inputs or outputs The microcontroller should be capable of entering a sleep mode with a current draw of around 14A and be woken by interrupts either external or timer driven Ideally the microcontroller will incorporate a sleep timer which will permit it to wake up after a defined time period The microcontroller should incorporate at least a basic 8 bit processor with sufficient memory to include both the interface with the energy subsystem and the other processing and communication functions 3 5 4 Under and over voltage protection Wireless sensor systems based on microcontrollers are sensitive to their supply voltage For systems which include energy harvesting this can be problematic as a character istic of these is that their output voltage will typically vary over time dependent on the charge state of the energy store the amount of energy being harvested and the dynamics of energy usage by the sensor node It is therefore important that the power management electronics act to prevent the microcontroller from being supplied with an
265. te for their technical supervision and guidance throughout this PhD and Geoff Merrett for his helpful and thorough contributions to collaborative work I would also like to express my appreciation to Neil Grabham for his contributions in connection with the Adaptive Energy Aware Sensor Network AEASN project Bashir Al Hashimi for his feedback on technical writing and to Neil Ross for his analogue circuit design advice Thanks also to Alex Rogers and Kirk Martinez for their helpful input as internal examiners over the course of this work I also wish to thank the Engineering and Physical Sciences Research Council EPSRC who provided funding to me through a doctoral training award and under Platform grant EP D042917 1 New Directions for Intelligent Sensors and the Data and Information Fusion Defence Technology Centre consortium for their financial assistance under the AEASN project Thanks also to the School of Electronics and Computer Science for the provision of excellent research facilities and support My gratitude also goes to the other members of the Electronic Systems and Devices ESD group who have made me feel at home including but not limited to Adam Lewis Andrew Frood Cheryl Metcalf Dirk de Jager Evangelos Mazomenos Matthew Swabey Mustafa Imran Ali and Russel Torah Special thanks must go to my parents for their love support and encouragement over many years Finally I would like to thank my wife Jenny for her lo
266. terface 2005 IEEE Sensors page 4 pp 2005 D Dondi D Brunelli L Benini P Pavan A Bertacchini and L Larcher Pho tovoltaic cell modeling for solar energy powered sensor networks Advances in Sensors and Interface 2007 2nd International Workshop on pages 1 6 June 2007 Texas Instruments Inc eZ430 RF2500 SEH Solar Energy Harvesting Develop ment Tool User s Guide http focus ti com lit ug slau273 s1au273 pdf January 2010 Last accessed May 2010 C Park and P H Chou Power utility maximization for multiple supply systems by a load matching switch Proceedings of the 2004 International Symposium on Low Power Electronics and Design pages 168 73 2004 Duracell Duracell alkaline manganese dioxide technical bulletin duracell com oem Pdf others ATB full pdf 1997 Last accessed May 2010 G V Merrett A S Weddell N R Harris B M Al Hashimi and N M White A structured hardware software architecture for embedded sensor nodes In 17th International Conference on Computer Communications and Networks August 2008 S G Hageman SPICE models a solar array Electronics Design News page 220 May 7 1992 T Markvart and L Castaner editors Practical Handbook of Photovoltaics Fun damentals and Applications Elsevier Science Oxford 2003 Texas Instruments Inc MSP430F2274 Mixed Signal Microcontroller focus ti com lit ds symlink msp430f2274 ep pdf 2008 Last accessed May 2010 Maxim Integrate
267. ternal 12V input which provides the required programming voltages The programming set up is shown in Figure 5 3 In situ programming however was not possible for the EEDS device on the multiplexer module this is due to the fact that the 1 Wire EPROM on this module is connected through a multiplexer IC which cannot be controlled easily when the device is not under the command of the microcontroller An in system programming facility may be desirable for future iterations of the system FIGURE 5 3 Set up for EEDS on an energy module to be programmed over a 1 Wire interface via a DS9097E COM port adapter connected to a PC serial port 5 3 2 Functions for 1 Wire communications The functions implemented for the 1 Wire interface are shown in the list below The 1 Wire interface functions are incorporated into the hardware abstraction layer of the energy stack as there is the possibility that alternative types of memory may be used to implement the EEDS functionality However much of the plug and play functionality of the system is dependent on features provided by the 1 Wire IC One of the most important functions is the onewireReset function which resets the 1 Wire bus and Chapter 5 Case Study Deployment in a Prototype System Software 127 listens for a presence pulse from a 1 Wire device on the bus This is used by the functions in the energy stack to detect whether an energy module is connected to a port and can the
268. tform 1 HW Platform 2 HW Platform 3 HW Platform 4 FIGURE 2 26 Hardware abstraction architecture implemented in TinyOS 2 0 repro duced from 104 2 8 3 Discussion The original TinyOS was not well respected in the wider community but the new soft ware structure implemented in TinyOS 2 0 is promising as it will deliver platform independence to applications running on sensor nodes The programming languages and compilers used for embedded software development are highly dependent on the support given by the microcontroller manufacturers although ANSI C is the dominant Chapter 2 Background Energy Sensing and Wireless Communication 49 language for most low power platforms The historical dominance of the communication stack has resulted in a number of schemes including ZigBee being heavily structured around this stack however some modern structures have given greater importance to energy management and sensor interfacing Chapter 3 of this thesis introduces a new hardware and embedded software structure to formalise this relationship allowing the sensor node to manage its energy and sensing resources alongside its communication activities 2 9 Existing systems 2 9 1 Wireless sensor system deployments A limited amount of useful information can be gained from simulation and a growing number of sensor network deployments are being documented Deployments are impor tant for the successful develo
269. the end user will want to select the appropriate energy hardware for the deployment location rather than the other way around and choose an appropriate node platform first before considering its energy resources 2 If designed for a specific configuration the embedded software on the mi crocontroller has to be extensively modified to interface with a modified energy subsystem Energy aware systems are at present highly tailored towards the energy hardware they are designed for and are difficult to adapt this means that the software of the sensor node along with the hardware limits the range of energy devices that can be accommodated while remaining aware of its energy sta tus This limitation means that applications for energy harvesting capable devices are limited to specific niche scenarios where the deployment fits with the energy device that the energy subsystem design has been developed for 3 As mentioned earlier the energy awareness of many existing systems ex Chapter 3 Development Towards a Reconfigurable Energy Subsystem 57 tends only to the energy stores and no active monitoring or management of the energy sources such as energy harvesting devices is carried out In systems with multiple energy harvesters this means that the relative performance of each device cannot be monitored for efficacy that any patterns in the generation of energy cannot be reliably detected and that the diagnosis of any energy related f
270. tion and peak fre quency carried out in the USA reproduced from 29 Vibration energy harvesters can be coarsely modelled by the linear system shown in Figure 2 6 Here m is the mass of the oscillating object k is the spring constant and cr is the damping coefficient Also y is the input displacement and z is the resultant spring displacement The term cr is a combination of electrical and mechanical damping In simple electromagnetic generators the electrical damping is linear but the damping effect becomes more complex for alternative generator types 30 This simple model can be expressed by Equation 2 1 FIGURE 2 6 Model of a linear inertial generator reproduced from m t cr t kz t mij t 2 1 The detailed modelling of the behaviour of vibration energy harvesters is outside the scope of this thesis however it is important to note that the dynamics of the load attached to the generator can affect the electrical damping and hence the overall per formance of the generator Most generators are tuned to specific frequencies normally mains frequency or twice mains frequency which are the most prevalent frequencies in electrical machinery Some techniques have been developed to modify the resonant 22 Chapter 2 Background Energy Sensing and Wireless Communication frequency of vibration energy harvesters either by mechanical tuning or by adjusting the dynamics of the electri
271. tion of the energy status After this the software would need to be re compiled and the microcontroller reprogrammed In the case of systems which are deployed in the field this would be a time consuming and unreliable process Indeed where a number of energy devices may be trialled in a situation in order to find the best way of powering a sensor node when they are changed frequently the act of reconfiguring the energy hardware would be arduous and impractical The software solution reported in this thesis greatly simplifies the process of reconfiguring the energy hardware of the device ensuring that the embedded software remains aware of its energy status without the need for code alteration compilation or device reprogramming This is believed to be a compelling argument and is arguably an essential feature to enable the energy hardware of sensor nodes to be selected as appropriate to the deployment environment and activity of the sensor node 5 7 System testing and results The system has been tested with the configuration shown in F igure 4 19 The first test performed involved starting with an empty supercapacitor and allowing it to charge from the mains module The result of this test is shown in Figure 4 5 In this test Channel l is the raw voltage on the multiplexer module and Channel 2 is the regulated voltage which the microcontroller is connected to It may be observed that the node reaches the 2 0V turn on voltage after ap
272. to be measured for a sustained period and monitored remotely with a higher spatial and often temporal resolution The Crossbow Telos mote as shown in Figure is a popular wireless sensor plat form that integrates sensing hardware with a low power microcontroller and a radio transceiver This device is normally powered by a pair of AA non rechargeable batter ies and the lifetime of the node is ultimately dependent on the the characteristics of the batteries and activity of the node i e its duty cycle and the power consumption of its sensing hardware when performing measurements Obviously the batteries must be replaced when depleted which imposes a maintenance requirement that can prove costly over the course of long deployments FIGURE 1 3 The Telos wireless sensor mote reproduced from 5 A common facet of wireless sensor nodes is that they are highly resource constrained They will typically have an 8 bit or limited 16 bit microcontroller with a relatively small amount of memory and be restricted by their energy resources operating from batteries or harvested environmental energy They have low power radios which often operate in relatively crowded radio frequencies and have a limited transmission range The cost and physical size of devices are also common constraints as are their weight and deploy ment location For these reasons the careful design installation and management of these systems is essential
273. to convert temperature dif ferentials into an electro motive force The maximum efficiency of power conversion from a temperature difference is represented by the Carnot efficiency 7 shown in Equa tion 2 4 where Thigh and Tiow are the temperatures on the hot and cold sides of the thermocouple and are measured in degrees Kelvin 29 Thigh Tiow 2 4 Thigh Geer The maximum amount of power available may be estimated by quantifying the heat flow through a material Roundy et al state that convection and radiation may be ignored at low temperature differentials The amount of power flow of heat through a material is given by Equation Here o is the heat flow through the material k 26 Chapter 2 Background Energy Sensing and Wireless Communication is the thermal conductivity of the material AT is the temperature difference across the thermocouple and is the length of the material Ongoing work in the field of thermoelectric energy generation appears to have the aim of exploiting thick film techniques in combination with IC etching in order to create complex structures comprising thousands of leg thermocouples Indeed Fleurial et al claim that for relatively small temperature differences such as 10 to 20K high specific power outputs in the 1 to 10 W cm range are potentially achievable provided that the legs be no thicker than 50 to 100 um 41 Bottner et al have produced a micro thermogenerator using t
274. tp eprints ecs soton ac uk 15342 Appendix C Selected Publications 179 Weddell A S Merrett G V Harris N R and Al Hashimi B M 2008 Energy Har vesting and Management for Wireless Autonomous Sensors Measurement Control 41 A This publication is not available in the online version of this thesis but may be down loaded from http eprints ecs soton ac uk 15342 180 Appendix C Selected Publications Weddell A S Merrett G V Harris N R and Al Hashimi B M 2008 Energy Har vesting and Management for Wireless Autonomous Sensors Measurement Control 41 4 This publication is not available in the online version of this thesis but may be down loaded from http eprints ecs soton ac uk 15342 Appendix C Selected Publications 181 Weddell A S Merrett G V Harris N R and Al Hashimi B M 2008 Energy Har vesting and Management for Wireless Autonomous Sensors Measurement Control 41 A This publication is not available in the online version of this thesis but may be down loaded from http eprints ecs soton ac uk 15342 182 Appendix C Selected Publications Weddell A S Merrett G V Harris N R and Al Hashimi B M 2008 Energy Har vesting and Management for Wireless Autonomous Sensors Measurement Control 41 4 This publication is not available in the online version of this thesis but may be down loaded from http eprints ecs soton ac u
275. tputs and two bidirectional digital I O pins These resource requirements are conservative the CC2430 and MSP430F2274 microcontrollers introduced in Section 2 7 1 have 21 and 32 general purpose I O pins with 8 and 12 of these being ADC enabled pins respectively Chapter 3 Development Towards a Reconfigurable Energy Subsystem 69 Pin Connector Description 1 Mux Address 0 2 Mux Address 1 Digital lines to control all multiplexers 3 Mux Address 2 4 1 wire EEDS line Digital 1 wire interface 5 Measurement Control Initiates measurement 6 Measurement Analogue measurement output 7 Device Control Control outputs to module these are bi directional 8 Device Control control state can be determined 9 Vreg Supply voltage output direct to 10 GND microcontroller TABLE 3 5 Common Hardware Interface connections between microcontroller and multiplexer module Multiplexer to Energy Modules The interface between the multiplexer module and each energy module is comprised of eight lines as shown in Table The 1 wire EEDS line one measurement control line one measurement line and two digital bidirectional I O lines are passed through to each energy module after multiplexing The ground and supply terminals from each module directly connect to the ground and raw voltage rails on the multiplexer module Furthermore the regulated voltage from the multiplexer module is fed back to each energy module to fa
276. tteries up to tens of kilojoules down to supercapacitors typically a few joules with an unsigned int which has range 0 65 535 e eanGetVoltEnergy is used to estimate the amount of energy that could be stored in the energy storage devices with the raw voltage on the multiplexer module at a given voltage It sequentially analyses each device connected to the multiplexer module and if the device is an energy store as opposed to an energy source it uses data on the maximum voltage from the device EEDS and the voltage on the multiplexer module given as an input and takes the lower of the two It then estimates the amount of energy that would be stored by the device connected to Chapter 5 Case Study Deployment in a Prototype System Software 141 the system at the given voltage The function returns an estimate of energy as an unsigned int and is measured in J 100 i e tens of millijoules in a similar way to the eanGetEnergy function In combination the data from the eanGetVoltEnergy and eanGetEnergy functions allow the useable proportion of energy stored on the system to be calculated This information is used by the EAN layer to calculate the EP value for the node Mathematical functions The calculations of power and energy carried out in this layer rely on a number of mathematical functions e eanCalculateCurve estimates the energy stored given the input voltage along a piecewise linear approximated discharge
277. uced from 52 2 3 8 Hybrid energy harvesting technologies A limited number of systems have combined multiple energy harvesting sources such as photovoltaic and wind as in AmbiMax 46 or photovoltaic wind and water flow as in Morais et al H These existing systems are discussed in more detail in Sec tion 2 9 There are also some modules that integrate two types of energy harvesting in a single device The Mid Volture hybrid energy havester HeH combines a piezoelectric vibration energy harvester with an encapsulated photovoltaic module 53 as shown in Figure Further some hybrid photovoltaic thermoelectric systems which aim to improve on the performance of conventional photovoltaic modules have been proposed and are currently under development at MIT 54 FIGURE 2 17 Mid Volture hybrid energy harvester with piezoelectric energy har vester and encapsulated photovoltaic module reproduced from 53 2 3 9 Summary and discussion A range of energy harvesting technologies have been developed and some are available commercially Table summarises the power densities and readiness levels of a range of prominent energy harvesting technologies Due to the maturity of the technology and the presence of significant levels of light in many environments photovoltaic energy harvesting is the most established form of technology however the selection of the ap Chapter 2 Background Energy Sensing and Wireless Communication
278. uch as primary 60 Chapter 3 Development Towards a Reconfigurable Energy Subsystem Thermo electric Vibration Photovoltaic Microcontroller amp Transceiver Li Primary Supercap Secondary FIGURE 3 1 A modular energy subsystem connected to a sensor node Energy modules are connected through the multiplexer module to provide the node s power supply batteries the default behaviour for these types of resource is to not use them unless this is requested by the microcontroller these types of energy module may also be fitted with push button switches to allow the system installer to override this temporarily until the system starts up and the microcontroller takes over management operations For example this may be useful to allow a primary battery to provide initial start up power for a system after it is first installed 3 2 3 Usage scenario Ultimately the usage scenario for this system is for the energy hardware to be connected together at the time of system deployment and for it to be reconfigurable afterwards through the addition removal or exchange of energy modules The process of system design and installation however will take into account the sensing tasks for the device as well as the available environmental energy It is expected to follow these steps 1 The deployment location and sensing methodology should be decided on An appropriate location should be chosen for deploying the sens
279. uels There are a number of possible approaches to design of the heat engine including piston based combustion engines Wankel internal combustion engines and micro gas turbine engines The P3 micro heat engine makes use of thermal expansion to generate electrical power by means of a piezoelectric material 20 Conversely radioactive sources could potentially offer massive energy densities for example the DIN isotope has been used by Li and Lal to actuate a conductive cantilever Beta particles emitted from the isotope collect on the cantilever which ef fects an electrostatic attraction eventually causing the cantilever to contact the isotope and discharge and the process restarts Along similar lines Fleurial 22 presents a sur vey of technologies including the combination of Radioactive Heat Units RHUs with thermoelectric generators 2 2 6 Discussion This section has covered the available technologies for energy storage focussing on pri mary and secondary batteries and supercapacitors with some additional consideration of other energy storage mechanisms In summary primary batteries have the advantage that they have a high energy density long shelf life and relatively low cost however they are a limited resource that must be replaced when expended Some of the long life technologies also have a stable discharge characteristic which limits the resolution of state of charge determination or end of life identification Conver
280. uetooth Low Energy are established or likely to become established in the field of low power wireless sensor nodes Other schemes are of peripheral interest but have not yet achieved a critical level of acceptance or have much higher power requirements The work carried Chapter 2 Background Energy Sensing and Wireless Communication 39 out in this thesis is implemented on hardware which is capable of 2 45GHz IEEE 802 15 4 or similar radio transmission schemes 2 6 Energy aware operation 2 6 1 Energy aware routing Leading on from Section a limited number of schemes exist to achieve energy aware routing In conventional non energy aware protocols continuously using the most energy efficient paths to route data to a sink can lead to a disproportionate amount of energy being used by nodes along the preferred path A number of routing schemes are energy aware and aim to extend the life of the network through managing the decline of the network by sharing routing tasks evenly between nodes 79 Few schemes are able to cope with the rapidly changing energy status of an energy harvesting node Notable examples however include the scheme developed by Shah and Rabaey 80 which associates probabilities with route paths dependent on residual energy and amount of energy required to route data through that path By associating probabilities with each path the energy expenditure can be equalised across the network as pack
281. uiescent current consumption of these compo nents must be minimised and the on resistance of components in the power path must Chapter 4 Case Study Deployment in a Prototype System Hardware 95 also be kept to a minimum For this reason the following components are used extensively in this case study as they offer good levels of performance against these metrics The selected data are given to allow the reader to appreciate the dynamics and importance of these carefully selected system components Measurement and control The operational amplifiers comparators and multiplexers used in the prototype modules are used to control module states and take measurements of system parameters e Comparator National Semiconductor LMC7215 This is a push pull output comparator which operates from a 2 8V supply and draws 0 71 A quiescent current e Operational Amplifier ST Microelectronics T5941 This is a rail rail output op amp which operates from a 2 5 10V supply but has been tested by the author down to 2V with minimal loss of resolution and draws a 1 24 A quiescent current e Data multiplexer Analog Devices ADG708 This low voltage 8 channel ana logue switch operates between 1 8 5 5V It has a maximum quiescent current con sumption of 1uA and on resistance of 211 Power switching and rectification These switching and rectification components are used at many points in the system acting to inhibit the reverse flow
282. ul indication of charge status as it assumes a resistive discharge of the energy store i e the amount of energy consumed by the system is dependent on the store voltage which is in turn dependent on the energy stored This calculation can be implemented on microcontrollers using a Taylor expansion The expansion of the natural logarithm is shown in Equation Taking the expansion to its fourth term results in a typical error of approximately 3 and to its fourth term the error is around 1 This will normally be sufficiently accurate to give a good idea of the state of charge of the system e lec z 1 Ing zx 1 pi gt ae 3 5 3 6 5 Battery state of charge and capacity For conventional primary cells such as alkaline manganese dioxide batteries the closed circuit voltage CCV of the cell must be measured to give an accurate idea of their state of charge Measuring the open circuit voltage OCV of the cell will not give an accurate indication of the service life remaining as it is subject to recovery effects and other phenomena The CCV can be determined by placing the battery under load and measuring its voltage The load is dependent on battery size but Energizer recommends Chapter 3 Development Towards a Reconfigurable Energy Subsystem 79 a 100 load is used for a single 1 5V AA size cell 88 The resultant CCV can then be traced across to the voltage profile of the cell such as that given by the manufacture
283. ult of the work carried out under this project 1 Weddell A S Harris N R and White N M 2008 Alternative Energy Sources for Sensor Nodes Rationalized Design for Long Term Deployment In Inter national Instrumentation and Measurement Technology Conference May 12 15 2008 Victoria British Columbia Canada 2 Weddell A S Harris N R and White N M 2008 An Efficient Indoor Pho tovoltaic Power Harvesting System for Energy Aware Wireless Sensor Nodes In Eurosensors 2008 7 11 September 2008 Dresden Germany 3 Merrett G V Weddell A S Lewis A P Harris N R Al Hashimi B M and White N M 2008 An Empirical Energy Model for Supercapacitor Powered Wireless Sensor Nodes In 17th International IEEE Conference on Computer Communications and Networks 3 7 August 2008 St Thomas Virgin Islands USA 4 Weddell A S Merrett G V Harris N R and Al Hashimi B M 2008 Energy Harvesting and Management for Wireless Autonomous Sensors Measurement Control 41 4 pp 104 108 ISSN 0020 2940 Chapter 1 Introduction 9 5 Weddell A S Grabham N J Harris N R and White N M 2008 Flexible Integration of Alternative Energy Sources for Autonomous Sensing In Electron ics System Integration Technology Conference September 1 4 2008 Greenwich UK 6 Merrett G V Weddell A S Harris N R Al Hashimi D M and White N M 2008 A Structur
284. urable system Section 4 2 outlines the situation of the case study deployment and the energy devices that have been chosen for the evaluation The basic circuitry which has been designed to enable energy awareness voltage regulation and other functionality for the system is described in Section 4 3 The hardware design of the multiplexer module is detailed in Section 4 4 and the designs of the energy modules are discussed in Section 4 5 The integrated system is discussed in Section The design of the embedded software to interface with this hardware is documented in Chapter 4 2 Overview of the case study 4 2 1 Scenario The aim of the case study is to provide a realistic representation of the types of energy resource that a sensor node may have access to on a typical industrial monitoring deploy ment These include harvestable environmental energy such as indoor artificial lighting machinery vibration temperature difference or air flow As the architecture proposed in this thesis is intended to be a comprehensive scheme for the energy subsystems of sensor nodes the case study also incorporates more conventional methods of powering the node such as primary batteries which may also be used in systems primarily using energy harvesting but as an emergency supply of energy and even a wired power 91 92 Chapter 4 Case Study Deployment in a Prototype System Hardware supply Energy is capable of being buffered i
285. urrent charge cannot be guaranteed and the cells will be charged in a piecemeal fashion whenever surplus energy is available Indeed a helpful aspect of this applica tion is that the cells need not reach 100 charge Therefore the charge system uses a straightforward voltage threshold technique to manage the charge of the cells Firstly the raw voltage of the multiplexer module is tested to ensure it is above 3V If so a 3V linear regulator is enabled which causes current to flow from the multiplexer module towards the cell should the voltage on the multiplexer module fall below 3V the charge is terminated In this way the cell voltage is effectively regulated to 3V and the cell is continuously topped up to this voltage whenever energy is available and charging is enabled by the microcontroller The cells never reach the 1 6V 3 2V in combination full voltage so an over charge condition will never be reached The EEDS stores information on the size of cells fitted Table shows the manage ment connections for this module Pin Type Function 2 Meas Control Connect battery to voltage divider 3 Measurement Analogue battery voltage through fixed load 4 Control Enable discharge from battery 5 Control Enable recharge of battery TABLE 4 10 Interface pins from secondary battery module Performance evaluation Similar to the primary battery module the secondary battery module has two bistable multivibrators th
286. ve understanding and unwavering confidence in me x1x To Jenny with all my love xxl Nomenclature Ratio between resistors in voltage divider arrangement Voltage factor Temperature factor Cross sectional area m Capacitance F Damping coefficient Separation distance m Carnot efficiency Permittivity of free space Energy J Tlluminance Lx Current A Diode saturation current A Maximum power point current A Photogenerated current A Short circuit current A Spring constant Thermal conductivity of material WK m7 Voltage ratio for photovoltaic cells Boltzmann constant Length m Wavelength of transmission m Mass of oscillating object kg Current ratio for photovoltaic cells Velocity of fluid ms Power W Inital transmitted power W Remaining lifetime fraction Energy fraction Logarithmic discharge fraction Voltage fraction Electronic charge C xxiii xxlv NOMENCLATURE Q s t 7 S Vo Vmpp Voc Votart Vinax w y z Heat flow through material W m Density of fluid kg m Radius of transmission m Resistance Q Expected lifetime Guaranteed lifetime Temperature K Hot side temperature K Cold side temperature K Expected temperature K Rated temperature K Voltage V Starting voltage V Maximum power point voltage V Open circuit voltage V Initial voltage applied V Maximum voltage reached V Width m
287. w voltage on the multiplexer is above 3V For future revisions of the circuit it would probably be worthwhile experimenting with step down regulators which could kick in when this ceiling voltage level is exceeded However this would make little difference to the efficiency of the circuit when the raw voltage is below this level It may also be worthwhile looking into the use of a Buck Boost converter in order that voltages below 2 0V could be also be boosted to be used by the microcontroller 4 7 Summary The case study described in this chapter described a realistic scenario with a sensor node having a range of energy devices available including energy harvesting devices energy storage and mains power These represent a subset of the selection of energy resources that may be expected to be used in deployments of sensor nodes in a variety of scenarios for example in a machinery monitoring application The proposed architecture was evaluated by way of the case study Basic circuitry was described and evaluated which enables energy awareness voltage regulation and other functionality for the system The hardware designs of the multiplexer module and the other relevant energy modules were described in detail The method for connecting the system together and its default operation were also explored The efficiency of the overall system and the impact of the additional circuitry required to deliver energy awareness was evaluated The design o
288. water meter which has been operating for over 20 years is powered by a Tadiran lithium primary cell 9 The fact that lithium cells have been demonstrated operating for very long periods of time is a welcome verification of the application of this technology to long term wireless sensing applications An arguable drawback of these cells however can be their flat discharge profile As shown in Figure 2 1 end of life for these lithium cells can only be detected passively up to 3 before cut off or actively by applying a pulsed load to the cell up to 15 before cut off In a deployment of ten years this 3 warning corresponds to approximately three months or a 15 indication is equivalent to 18 months Providing that the sensor node draws the expected amount of current this in itself is not problematic three months is in most applications an acceptable amount of time to arrange battery replacement In the case of a malfunctioning sensor that is drawing a substantial amount more current than expected however the warning period would clearly be much shorter The difficulty of 14 Chapter 2 Background Energy Sensing and Wireless Communication identifying the end of life or state of charge of primary batteries is a major drawback for their application in wireless sensor nodes Primary batteries are resources that cannot be used after they have been expended except for some negligible recovery effects therefore energy must be use
289. x architecture and circuitry reproduced from 46 An alternative system is MPWiNodeX 45 which is capable of using energy from wind water flow and sunlight to power a sensor node for precision agriculture applications However the type of energy store cannot be changed and the energy sources only give a coarse indication of their status they cannot be actively managed This system has been developed for a precision agriculture application and its architecture is shown in Figure 2 31Jalong with a photograph of the deployment of a prototype system The three energy sources were capable of providing an average current of 58mA which exceeds the current requirement of the wireless sensor node in its active mode 2 9 4 Discussion A large number of sensor network deployments have been reported in the literature Some of the most common applications of the technology are for environmental investi gations and machinery condition monitoring The largest reported sensor network com prised 1 200 nodes and the largest deployment of energy harvesting nodes comprised 557 solar powered nodes The majority of nodes powered by environmental energy sim ply harvest energy from solar cells deployments of nodes using other harvested power sources are much less common A small number of projects have utilised power har vested from a range of sources however these systems have been configured for specific power sources and are relatively inflexible T
290. xt and evaluated in line with the criteria outlined in Section 3 9 2 3 2 2 Modular design With a potentially complex energy subsystem that is able to support multiple energy sources and stores it is important that a standardised modular approach is used for the design of the hardware in the energy subsystem As already expressed in Section 3 2 1 there is a need for a system that can be plugged together at the time of system de ployment with the appropriate energy and sensing hardware being attached and for the individual modules in the system to interface to deliver a reliable and robust power supply that can both be monitored and managed by the microcontroller on the wireless sensor node It has been stated that each energy module should have its own power conditioning and management interface circuitry but as yet a scheme has not been in troduced that permits multiple energy devices to be attached to a sensing system in such a way The system described here enables a plug and play energy subsystem by means of a modular architecture The modular scheme outlined in this thesis includes the following elements 1 A multiplexer module that accommodates a number of energy modules being connected through sockets on the multiplexer module circuit board The multi plexer module has no real intelligence and is simply a common path for power lines provides voltage regulation circuitry and appropriate multiplexer facilities to allow measu
291. y command to bus 7 2 x Write Byte 00h to bus start read from address zero 8 Read Byte CRC of command 9 Multiple Read Byte commands dependent on type of device read and store EEDS parameters in microcontroller memory Chapter 5 Case Study Deployment in a Prototype System Software 129 The typical execution times for these operations mean that the time taken up until the start of item p is approximately 8 24ms Each further byte read takes a further 560us For the system developed here there are up to 13 additional bytes to be read from the data sheet The read time for this would be 7 28ms which means an overall read time of 15 5ms The shortest data sheet implemented in this case study has four additional bytes which equates to an additional read time of 2 24ms or an overall read time of 10 5ms The minimum amount of time taken per address is 960s in the case that a device is not connected and therefore no further actions need to be taken beyond the reset operation This operation must be carried out for each socket on the multiplexer module In the extreme case where all sockets are occupied and devices have the maximum length of data sheet the read operation for the six devices takes approximately 93ms The multiplexer module data sheet must also be interrogated but has only five bytes of data after the CRC is returned The total read time in this case is approximately 11ms For a combined re
292. y from twice line frequency vibrations in this case 100Hz Its performance can be assumed to be similar to that of the Perpetuum device shown in Section 2 3 3 The harvesting device gives an AC output which must be rectified for use by the sensor node For the purposes of this case study it will be assumed that the device is deployable on a machine with those frequencies at a suitable amplitude in order to provide power to the system The vibration energy harvesting module is described in Section 4 5 2 Other energy harvesting sources air flow temperature difference A number of other modules were developed under the AEASN project to interface with the plug and play system These are of peripheral interest to this project and therefore are only covered briefly here The modules included 1 A thermoelectric energy harvester which was based on an off the shelf ther moelectric generator from Tellurex which permitted milliwatts of power to be generated from substantial temperature differences The module incorporated a heat sink and small fan which worked to increase the temperature gradient the fan could optionally be self powered by the thermoelectric module or driven from an external power supply to simulate a flow of air past the device The device provided a DC output which had to be regulated for use by the sensor node 2 A miniature wind turbine which used an adapted Mathmos Wind Light as its power source Again the device cou
293. z 4 4 gt 1 8V 2 2V 2 7V 33V 36V Supply Voltage V FIGURE 5 7 Voltage and system frequency frequency requirements for writing to the MSP430 flash reproduced from 127 e eanGetPower is used to estimate the amount of power being generated by the system at a given moment It sequentially analyses each device connected to the multiplexer module and if the device is an energy source as opposed to an energy store it carries out a measurement on the device The measurement is then converted to an estimate of power using the mathematical functions listed below and the parameters from its electronic data sheet The function returns an estimate of power as an unsigned int and is measured in microwatts Given the limits of the unsigned int variable this allows the system to represent incoming power between zero and 65 5mW e eanGetEnergy is used to estimate the amount of energy stored in energy storage devices It sequentially analyses each device connected to the multiplexer module and if the device is an energy store as opposed to an energy source it carries out a measurement on the device The measurement is then converted into an estimate of energy stored using the mathematical functions below The function returns an estimate of energy as an unsigned int and is measured in J 100 i e tens of millijoules The units used in this estimation are necessary to be able to represent the large range of energy that may be stored in ba

- ePrints Soton

Contents

Download Pdf Manuals

Related Search

Related Contents