Home

Final Report - College of Engineering

1. and the nodes was a large factor in data E e Pins retrieval for the nodes Many sample tests FIGURE 18 SENSOR BOARDS WITH THREE SENSORS were run under various conditions in order to observe and examine the behavior of the network The first of the tests run focused primarily on the jumper pins located on the Wireless Sensor Boards shown in Figure 19 Detailed descriptions of the tests run provide support to show failed sensors within the network which affect the health of the network HARDWARE TESTS All tests used both of the configured sensor nodes Multiple parameters were changed and modified during each test such as length of the test removal of jumper pins from specific sensors and angle at which the sensors were positioned in relation to the RF power transmitter All were completed within a short distance as discussed above in ideal environmental conditions and with the dipole antennas 12 4 2013 Wireless Sensor Network Health Diagnostic 22 The data for all tests can be found in Appendix 4 ECE 480 Design Team 2 Test Name Sensor Node Transmitter Sensor Status Test Position Height amp Distance Jumper Pins Pulled On Certain Time Angle in Relation from Sensor Nodes Sensors to Transmitter Height Width Distance Height Temperature Light Humidity Sample2 3 3 97 0 3ft 2 5inches Not Active Active Active 30 mins Sample2 4 3 97 0 3ft 2 5 inches Active Not Active
2. 9386 Node 1 TX ID 242 Temp 70 9 F Light 141 1x Time 03 11 37 dT 00 02 RSSI 2 08mw Humidity X Extrn 1548 mv Packet 9387 Node 2 TX ID 242 Temp 71 6 F Light 171 1x Time 03 11 38 dT 00 02 RSSI 3 03mw Humidity Extrnl 1643 mv Packet 9388 Node 1 TX ID Temp 70 9 F Light 141 1x Time 03 11 39 dT 00 02 RSSI 1 74mw Humidity X Extrn 1543 mv Packet 9389 Node 2 TX ID 242 Temp 71 4 F Light 179 1x Time 03 11 40 dT 00 02 RSSI 2 48mw Humidity Extrn 1637 mv Packet 9390 Node 1 TX ID Temp 70 9 F Light 141 1x Time 03 11 42 dT 00 03 RSSI 2 22mw Humidity Extrnl 1522 mv Packet 9391 Node 2 TX ID 242 Temp 71 4 F Light 182 1x Time 03 11 42 dT 00 02 RSSI 2 74mw Humidity Extrn 1640 mv Packet 9392 Node 2 TX ID Temp 71 4 F Light 177 lx Time 03 11 45 dT 00 02 RSSI 2 61mw Humidity Extrnl 1630 mv Packet 9393 Node 1 TX ID Temp 70 9 F Light 145 1x Time 03 11 45 dT 00 03 RSSI 2 48mw Humidity X Extrnl 1525 mv H D Sample 2 5 Test Data Eile Edit Format View Help Packet 11047 Node 2 TX ID Light 1x 3 Time 03 44 44 dT 00 04 RSSI 1 42mw Extrnl 982 mv Packet 411048 Node 1 TX ID 242 Light 1x Time 03 44 46 dT 00 03 RSSI 2 33mw Extrnl 897 mv Time 03 44 48 dT 00 02 RSSI 2 33mw Packet 11049 Node 1 TX ID Packet 11050 Node 2
3. specific health monitoring functions They also created a custom software application that was able to easily display node status as well as the environmental conditions being tracked They periodically polled the sensor network for the data in order to dynamically display the data collected from the network in real time The proposed design solution will utilize some of this background knowledge to aid in the rapid deployment of a wireless sensor network diagnostic tool and in designing new analysis algorithms 2 EXPLORING THE SOLUTION SPACE AND SELECTING A SPECIFIC APPROACH DESIGN SPECIFICATIONS In order to create a successful design design specifications for the project needed to be defined Upon close examination the project description and discussions with the Air Force Research Lab sponsor a successful project requires i the configuration of a wireless sensor network ii low power sensor nodes iii reliable communication within the network and iv accurate sensor node measurements Additional desirable requirements include v simple network configurability and Vi development of a graphical user interface A MUST BE SATISFIED FULLY CONFIGURED WIRELESS SENSOR NETWORK To begin on the project a wireless sensor network must be established This includes sensor nodes that monitor external and internal measurements a cluster head and working communication between the nodes over an established protocol The external measure
4. 16 Bit XLP development board via connection J7 Once both were connected the development board settings were set To make sure the development board was operating properly the team had to set switch four to PIC24FK which allowed the team to configure K series flash devices The K series was used for the development kit but if the 16 Bit XLP development board was bought by itself then the J series could be used as well FIGURE 10 MRF24J40 CONNECTED TO DEVELOPMENT BOARD Next switch seven the component power switch had to be set to the ON position which allowed the component power to be selected only by the corresponding component select jumper After switch seven was set to ON jumper 12 had to be set to EXT PS USB Setting jumper 12 to EXT PS USB allows the 16 Bit XLP development 12 4 2013 Wireless Sensor Network Health Diagnostic 15 ECE 480 Design Team 2 board to be powered via USB The USB connection provides a nominal 5V power source but by using a Schottky diode and a Low Dropout regulator circuit the voltage was dropped down to 3 3 volts for the microcontroller and board components After this power source was selected the green LED on the development board was illuminated which showed the team the board was being set up correctly Other power sources could have been used such as two AAA batteries a CR2032 coin cell battery or an external regulated DC power supply Once the power was set up a few jumpers had to be connec
5. 30 mins Active Sample2 5 3 97 0 3ft 2 5inches Not Active Active Not Active 30 mins Sample2 6 3 97 0 3ft 2 5inches Not Active Not Not Active 30 mins Active Sample2 7 15 25 0 8ft 4ft Active Active Active 45 mins Sample2 8 0 0 2ft Oft Active Active Active 2hr Sample2 9 0 48 5 2 67ft Oft Active Active Active 30 mins TABLE 5 SUMMARY OF TEST PROCEDURES Tests sample 2 3 through 2 6 all displayed similar results Figure 18 below gives a nice visual representation of the transmitter and sensor positions The only difference that occurred between each test was that the data for sensors that were not active could not be displayed via the HyperTerminal This unreadable data was displayed as such and can be seen in the test data supplied in Appendix 4 The Received Signal Strength RSSI varied between 0 5mW 3 5mW and time between packets dT averaged around 2 seconds In order to determine the current draw of the sensors from the evaluation board the voltage across JP2 on the Wireless Sensor Board was measured and divided by a 100 resistor that was in parallel for current measuring purposes An oscilloscope Infiniium DSO90644A was used to collect screen captures that show the voltage across the pin These screen captures can be found in Appendix 4 It was found that regardless of active or non active sensors on the Wireless Sensor Board the current draw was constant and un
6. 4 2013 5 days 11 days 3 days 1 day Mon 11 18 13 Tue 11 5 13 Wed 11 13 13 Thu 11 28 13 Fri 12 6 13 Fri 17 11 22 13 Tue 14 11 12 13 Wed 20 11 27 13 Sun 21 12 1 13 Wed 12 4 13 Fri 22 12 6 13 ECE 480 Design Team 2 Brad Stu David Kelly Brad Stu David Kelly Stu Brad David Kelly Stu Brad David Kelly Brad Stu David Kelly Stu David Brad Kelly Wireless Sensor Network Health Diagnostic NA Fri 11 8 13 Fri 11 22 13 Sun 12 1 13 NA Fri 12 6 13 APPENDIX 4 TEST DATA amp SCREEN CAPTURES 12 4 2013 Packet 5371 Time 01 19 53 Packet amp 5372 Time 01 19 54 Packet 5373 Time 01 19 56 Packet 5374 Time 01 19 56 Packet 5375 Time 01 19 59 Packet amp 5376 Time 01 20 01 Packet 5377 Time 01 20 01 Packet amp 5378 Time 01 20 04 Packet amp 5379 Time 01 20 05 Packet 5380 Time 01 20 06 Packet 5381 Time 01 20 08 Packet amp 5382 Time 01 20 09 Packet 5383 Time 01 20 10 Packet amp 5384 Time 01 20 13 Packet amp 5385 Time 01 20 15 Packet amp 5386 Time 01 20 15 4 Node 1 dT 00 04 Node 2 dT 00 03 Node 2 dT 00 02 Node 1 dT 00 03 Node 2 dT 00 03 Node 1 dT 00 04 Node 2 dT 00 02 Node 2 dT 00 03 Node 1 dT 00 04 Node 2 dT 00 02 Node 2 dT 00 02 Node 1 dT 00 04 Node 2 dT 00 02 Node 2 dT 00 03 Node 1 dT 00 05 Node 2 dT 00 02
7. 7 9 and required a few additional libraries including pySerial 10 matplotlib 11 and wxPython 12 Links to more information on each of these libraries can be found in Appendix 2 12 4 2013 Wireless Sensor Network Health Diagnostic 19 ECE 480 Design Team 2 Additional algorithms were implemented for data analysis This was done in both Python and Matlab where the packets were read into Python which then converted the data into a comma separated value csv file s to be read from Matlab When inputted into Matlab correlations in the data could be viewed and graphs were easily made Doing this allowed for easier determination of what metrics could help determine a failing sensor assisted in looking at the data easily and also made calculations easier to compute An example of this data analysis from Matlab was shown below in Figure 14 Recieved Signal Strength External Voltage 0 25 1800 A Node Nodet z 0 fS SCAM RA Node2 1600 Node2 t VISIT ICI PA ITA t S 015 V RS Ss 1400 2 F B 3 1200 S 0 1F ef 2 Ka Mi e 1 3 1 2 1 441 e E En 43153 3 4 E 671 474 3 46 J TOUR Eu grex Fe x 994 T ea idus n3 6a 1323 04 30 09 04 35 07 04 40 05 04 45 03 04 50 01 04 54 59 04 59 57 05 04 56 04 30 09 04 35 07 04 40 05 04 45 03 04 50 01 04 54 59 04 59 57 05 04 56 Time Stamp HH MM SS Time Stamp HH MM SS p Time Between Packets Sent Humidity
8. ID 242 RSSI 3 96mw TX ID 242 RSSI 5 21mw TX ID RSSI 3 47mw TX ID 242 RSSI 5 40mw TX ID 242 RSSI 5 60mw TX ID 242 RSSI 3 32mw Wireless Sensor Network Health Diagnostic Temp 69 9 Humidity 29 Temp 70 3 Humidity 28 Temp 69 9 Humidity 28 Temp 69 9 Humidity 28 Temp 70 1 Humidity 28 Temp 69 9 Humidity 28 Temp 70 3 Humidity 28 Temp 70 1 Humidity 28 Temp 70 1 Humidity 28 Temp 70 1 Humidity 28 Temp 70 3 Humidity 28 Temp 70 1 Humidity 28 Temp 70 3 Humidity 28 Temp 69 9 Humidity 28 Temp 70 1 Humidity 28 Temp 70 3 Humidity 28 Sample 2 8 Test Data en en en m en en m en am m RN RN RN RN RN EA Light Extrnl Light Extrnl Light Extrnl Light Extrnl Light Extrnl Light Extrnl Light Extrnl Light Extrnl Light Extrnl Light Extrnl Light Extrnl Light Extrnl Light Extrnl Light Extrnl Light Extrnl Light Extrnl 258 1690 287 1096 298 1093 298 1093 291 1093 274 1692 302 1090 265 1687 269 1690 270 1695 275 1683 304 1090 269 1683 265 1686 274 1692 302 1093 261 1711 276 1626 262 1718 259 1715 280 1623 266 1715 280 1626 270 1712 259 1715 258 1704 276 1626 261 1708 272 1628 262 1712 258 1701 276 1620 41 ECE 480 Design Team 2 Packet 12843 TX ID 242 Temp 70 3 Light 261 Time 02 15 50 dT RSSI 1 37mw Humidity 26 Extrnl 1683 Light
9. RSSI 0 19mw TX ID 242 RSSI 0 20mw TX ID RSSI 0 20mw TX ID RSSI 0 18mw TX ID 242 RSSI 0 10mw TX ID 242 RSSI 0 20mw TX ID 242 RSSI 0 20mw TX ID 242 RSSI 0 19mw TX ID RSSI 0 09mw Temp 70 1 Humidity 36 Temp 69 9 Humidity 37 Temp 69 9 Humidity 36 Temp 69 9 Humidity 36 Temp 69 9 Humidity 36 Temp 69 9 Humidity 36 Temp 70 1 Humidity 36 Temp 69 9 Humidity 36 Temp 69 9 Humidity 36 Temp 69 9 Humidity 36 Temp 69 7 Humidity 36 Temp 69 9 Humidity 36 T 69 7 humidity 36 Temp 69 7 Humidity 36 Temp 69 8 Humidity 36 Temp 69 7 Humidity 37 n KTM am KM 3m KM KTM KTM KTM m KT RN en wen wen en Light Extrnl Light Extrnl Light Extrnl Light Extrnl Light Extrnl Light Extrnl Light Extrnl Light Extrnl Light Extrnl Light Extrnl Light Extrnl Light Extrnl Light Extrnl Light Extrnl Light Extrnl Light Extrnl Node 2 dT 00 01 Node 1 dT 00 01 Node 2 dT 00 01 Node 2 dT 00 01 Node 1 dT 00 01 Node 2 dT 00 01 Node 1 dT 00 01 Node 2 dT 00 01 Node 2 dT 00 01 Node 2 dT 00 01 Node 1 dT 00 03 Node 2 dT 00 01 Node 1 dT 00 01 Node 2 dT 00 01 Node 2 dr 00 01 Node 1 dT 00 01 TX ID RSSI TX ID RSSI TX ID RSSI TX ID RSSI 4 84mw TX ID 242 RSSI 4 14mw TX ID 242 RSSI 5 21mw TX ID RSSI 3 79mw TX ID RSSI 5 21mw TE ID Soc RSSI 5 60mw TX ID 242 RSSI 5 21mw TX
10. RssI TX ID 242 RSSI 1 85mw TX ID 242 2 33mw TX ID RSSI 2 48mw TX ID 242 RSSI 0 91mw TX ID 242 RSSI 3 32mw TX ID RSSI 1 63mw TX ID 242 RSSI 2 08mw TX ID 242 RSSI 3 32mw TX ID RSSI 1 27mw TX ID 242 RSSI 2 22mw TX ID RSSI 2 61mw TX ID 242 RSSI 1 37mw Tk ID RSSI 2 33mw TX ID ege RSSI 2 22mw TX ID RSSI 1 32mw TX ID 242 RSSI 2 33mw F 6 Light 254 Extrnl 1567 Light 232 Extrnl 1710 Light 237 Extrnl 1727 Light 250 Extrnl 1567 Light 240 Extrnl 1690 Light 258 Extrnl 1093 Light 235 Extrnl 1713 Light 232 Extrnl 1707 Light 250 Extrnl 1096 Light 237 Extrnl 1717 Light 233 Extrnl 1727 Light 258 Extrnl 1093 Light 233 Extrnl 1721 Light 235 Extrnl 1704 Light 250 Extrnl 1096 Light 240 Extrnl 1712 ECE 480 Design Team 2 Wireless Sensor Network Health Diagnostic Sample 2 3 Test Data sample2 4 Notepa perra File Edit Format View Help Packet 7927 Node 2 TX ID Temp 71 0 F Light 1x 3 Time 02 37 19 dT 00 04 RSSI 1 74mw Humidity 36 X Extrn 1084 mv Packet 7928 Node 1 TX ID 242 Temp 70 7 F Light 1x Time 02 37 20 dT 00 02 RSSI 2 48mw Humidity 36 Extrnl 985 mv Packet 4 7929 Node 1 TX ID 242 Temp 70 7 F Light 1x Time 02 37 22 dT 00 02 RSSI 2 48mw Humidity 36 X Extrnl 987 mv Packet 7930 Node 2 TX ID 242 Temp 71 0 F Lig
11. easy to set up TABLE 1 HARDWARE FEASIBILITY MATRIX As seen from the feasibility matrix on the previous page using the full sensor development kit would provide the most accuracy and functionality with quick turnaround time in order to start developing a diagnostic tool The decision was made to do further research on full sensor network development kits 12 4 2013 Wireless Sensor Network Health Diagnostic 9 ECE 480 Design Team 2 since there is a wide range available and they vary in components ease of use and price Another way to choose designs was a selection matrix The selection matrix is similar to the feasibility matrix except the basic design is already known and instead actual parts are compared Table2 shows the comparison of multiple development kits Each development kit was ranked on a 1 3 or 9 basis per feature then summed and multiplied by a weighted value Selection Matrix Conceptual Design Rankings Sun SPOT Crossbow MTS400 Powercast National Rev8 P2210 EvaL01 Instruments AND Wireless Sensor Network Starter Kit Complete Sensor Board Complete Wireless Wireless Sensor men Network Network adig I sm p m mmm Language Type of Sensors Temperature Temperature Temperature Temperature Light Humidity Barometric Humidity Light and Voltage Accelerometer Pressure Light and Voltage Acceleration and EE Wireless RF to HERI Raus Totals 090 5 m 59 The t
12. et aol 5 ii Node1 E m Nodet ZS sol d x Node2 38 Node2 c A vA AA U E _ ES d f SC Ted NC Ade RS S 20 t bd f e ar Ei E Sloe EE EAE E jp pep pun qr et org u n A A 4 94 30 09 04 35 07 04 40 05 04 45 03 04 50 01 045459 04 59 57 05 04 56 04 30 09 04 35 07 04 40 05 04 45 03 04 50 01 04 54 59 04 59 57 05 04 56 Time Stamp HH MM SS Time Stamp HH MM SS Light WS Temperature Z r Nadel S ek Nodet 3r Pv Node2 Node2 x u EE iert LL ME By Z Zb S ene V X Ve N Y d geg E 7 2u0 Seel E bt th 04 30 09 04 35 07 04 40 05 04 45 03 04 50 01 04 54 59 04 59 57 05 04 56 04 30 09 04 35 07 04 40 05 04 45 03 04 50 01 04 54 59 04 59 57 05 04 56 Time Stamp HH MM SS Time Stamp HH MM SS FIGURE 15 MATLAB ANALYSIS The final part of the software that was implemented into this project was the failure analysis portion also done in Python The failures were based off of three separate functions all determined from studying the different data types and Light 320 how they react to failure These E Average over last few samples 2 included a short term analysis a long 280 g term analysis and a zero data analysis ral n E UU CHEN ke 740 The short term analysis was used for me the external sensors measurement of Bir 180 light humidity and temperature which 04 31 31 04 36 01 04 40 32 04 45 02 04 49 33 04 54 03
13. hardware for the design process and being there for help with certain tasks 12 4 2013 Wireless Sensor Network Health Diagnostic 2 ECE 480 Design Team 2 TABLE OF CONTENTS Executive S mmialty EE 1 Acknowledgements EE 2 1 introduction and Backe ron oer voee eta EE KEEN RR E Ree Se 5 ug get Tee TEE 5 Background MEER IEEE DE OD T DI DI LL LOL LO LI QM 5 2 Exploring the solution space and selecting a specific approach 6 RETTEN Eege ele Le CN 6 perge Mt 6 i Fully configured wireless sensor petwork eese neni nnns nnn 6 Il Low power sensor nodes enne nnns enne neni rn nass s sisse seras a assises entran sanis nnn 6 Ill Reliable communication within the network nnne 7 IV Accurate sensor node measurerments nennen enne tentent nennen ns 7 B Increases design desirability ice AEeSdER SEENEN ENEE eee E RA Te RR SERRE Va eR ERE 7 V Simple network configuration eese sees nennen enne nnne nnns nennt nina nass eser seti aE NEE essa an 7 VI Development of a graphical user pnterface eene enne nnne nnn nnne nna 7 FAST DIABLE EE 7 Coriceptual deslgfis crit ee oe t ed o hesitate see e v LE E eye DI Eee ev ERES Y 8 A Build Entire Sensor Network 8 B Zigbee Network Kit irruit rrt ee et eese etna ca e Eve eua d e Ee ere b exa a reet E EENE 8 C Sensor Network development Kit 9 Chosen Design Sol tlOh ceci oae ec rta euer dria ta eset EES ed Longi deu ul o ERR ER
14. http www powercastco com PDF P2110 EVAL 01 manual pdf Powercast P2110 915 MHz RF Powerharvester Receiver Datasheet Powercast 2013 http www powercastco com PDF P2110 datasheet pdf Powercast P2110 TX91501 915 MHz Powercast Transmitter User Manual Powercast 2013 http www powercastco com PDF TX91501 manual pdf Python Programming Language Python Software Foundation 1990 2013 http www Python or pySerial Documentation Chris Liechti 2001 2013 http serial sourceforge net wxPython Documentation 2013 http www wxPython or matplotlib Introduction John Hunter Darren Dale Eric Firing Michael Droettboom 2012 2013 http matplotlib or 12 4 2013 Wireless Sensor Network Health Diagnostic 32 APPENDIX 3 GANTT CHARTS Gantt Chart Week 4 Task Name Duration Project Overview amp 9 days Tasks First Group Meeting Project 1 day Assignment amp Initial Tasks Meeting Times day amp Scheduling First Meeting w Group 1 day Facilitator Air Force Research 7 days Labortory AFRL Proposal Submit AFRL 1 day Proposal GANTT Chart 2 days Pre Proposal 5 days Due Team Webpage 6 days Started 12 4 2013 Start Wed 9 4 13 Wed 9 4 13 Wed 9 4 13 Tue 9 10 13 Thu 9 5 13 Mon 9 16 13 Mon 9 16 13 Mon 9 16 13 Mon 9 16 13 Finish Mon 9 16 13 Wed 9 4 13 Wed 9 4 13 Tue 9 10 13 Fri 9 13 13 Mon 9 16 13 Tue 9 17 13
15. microcontroller the current used by the sensors or knowing what jumpers to pull for each sensor There were a multitude of switches and jumpers on the development and evaluation boards which Kelly was able to learn and effectively teach to the team 12 4 2013 Wireless Sensor Network Health Diagnostic 28 ECE 480 Design Team 2 STU ANDRZEJEWSKI FAULT DETERMINATION AND SYSTEMATIC FAILURE OF NODES Stu s technical role comprised of network testing and fault determination He created multiple test procedures and implemented node failures in order to analyze the network health These test procedures included but were not limited to targeted sensor failures sensor distance testing and RF signal strength in relation to network health The forced failures of specific sensors on the Wireless Sensor Boards were done with the express interest to effects it would have on the network During each test Stu recorded many different parameters important to the testing procedure These different parameters included the number of active nodes the number and type Light Temperature and Humidity of active and failed sensors the RF power transmitter angle in relation to the sensor nodes the distance of the sensors from the RF power transmitter and the antenna type that were attached to each sensor node Due to the limits of the project budget a total of 2 sensor nodes were available and Stu utilized both of these sensor nodes for a vast majority
16. of his tests During his testing he was able to successfully determine the current draw for the Wireless Sensor Boards attached to each node and collected multiple screenshots to confirm this The most challenging part for Stu during testing was the fault determination Because of the RF energy source testing had to be done as close distances If nodes were not within range of the RF power transmitter complete failure of the node would occur and no data would be received Stu did look at the possibility of physically damaging the sensor nodes because this would likely give data irregularities but it was decided after a team discussion to not approach any testing in this fashion Stu also assisted Kelly with some of the minor elements of the network configuration during the project 12 4 2013 Wireless Sensor Network Health Diagnostic 29 ECE 480 Design Team 2 DAVID ROGERS SOFTWARE ARCHITECT AND GRAPHICAL USER INTERFACE DEVELOPER David s technical role was designing and implementing the software associated with the diagnostic tool This role required a solid foundation of object oriented programming knowledge David s decision to use a model view controller architecture provided a foundation for rapid development of the application He was able to successfully find and integrate the necessary libraries in order to deliver a high quality software product with the proper documentation for further development His controller impleme
17. 04 58 34 05 03 05 Time Stamp HH MM SS determined if the very last one seemed out of normal To determine what normal was a calculated threshold was based off of 2096 of the average of the frame except for the last packet This formula was 12 4 2013 Wireless Sensor Network Health Diagnostic 20 ECE 480 Design Team 2 calculated under the assumption that if one of these packets fluctuated that largely that fast when an external factor such as that would typically not do so and generally have slow changing values then something was not operating correctly External Voltage 1100 Average over previous data 900 Calculated threshold Voltage mV 850 800 750 M i L L A L n 04 31 31 04 36 01 04 40 32 04 45 02 04 49 33 04 54 03 04 58 34 05 03 05 Time Stamp HH MM SS FIGURE 177 LONG TERM ANALYSIS The long term analysis was particularly designed for the internal sensors which could fluctuate a good amount however it would generally stay around the same value that it initially was Therefore the average of the entire data set was taken and if a packet came in that was a certain threshold away from the calculated mean then something occurred to make this irregularity occur Temperature Non zero elements Temperature Fahrenheit D S Y S II Zero valued elements i inat 04 45 02 04 49 33 04 54 03 04 58 34 0503 99 FIGURE 18 ZERO VALUE ANALYSIS 04 36 01 Finally a zero data
18. 2583 Time 04 31 31 Packet 12584 Time 04 31 56 Packet 12585 Time 04 32 29 Packet 12586 Time 04 33 00 Packet 12587 Time 04 33 36 Packet 12588 Time 04 33 56 Packet 12589 Time 04 34 05 Packet 12590 Time 04 34 38 Packet 12591 Time 04 35 09 Packet 12592 Time 04 35 45 Packet 12593 Time 04 35 47 Packet 412595 Time 04 36 52 Packet 12596 Time 04 37 23 Packet 12597 Time 04 37 36 H ECE 480 Design Team 2 Node 2 dr 02 11 Node 1 dT 05 09 Node 1 dT 00 25 Node 1 dT 00 33 Node 1 dT 00 31 Node 2 dT 03 27 Node 1 dT 00 56 Node 2 dT 00 29 Node 2 dT 00 33 Node 2 dT 00 31 Node 2 dT 00 36 Node 1 dT 01 50 Node 2 dT 00 34 Node 2 dT 00 33 Node 2 dT 00 31 Node 1 dT 01 49 Ele Edit Format Views Help Packet 479 Time 00 09 23 Packet 480 Time 00 09 23 Packet 481 Time 00 09 24 Packet 482 Time 00 09 25 Packet 483 Time 00 09 25 Packet 484 Time 00 09 26 Packet 485 Time 00 09 26 Packet 486 Time 00 09 27 Packet 487 Time 00 09 28 Packet 488 Time 00 09 29 Packet 489 Time 00 09 30 Packet 490 Time 00 09 30 Packet 491 Time 00 09 31 Packet 492 Time 00 09 31 Packet 493 Time 00 09 32 X ID RSSI 0 07mw TX ID 242 RSSI 0 19mw TX ID RSSI 0 22mw TX ID RSSI 0 20mw TX ID 242 RSSI 0 19mw TX ID 242 RSSI 0 20mw TX ID RSSI 0 10mw TX ID
19. 283 Extrnl 1093 Light 265 Extrnl 1690 Light 269 Extrnl 1684 Light 280 Extrnl 1096 Light 269 Extrnl 1684 Light 272 Extrnl 1096 Light 266 Extrnl 1692 Light 266 Extrnl 1692 Light 272 Extrnl 1096 Light 270 Extrnl 1704 Light 269 Extrnl 1684 Light 276 Extrnl 1099 Light 274 Extrnl 1692 Light 270 Extrnl 1692 Packet 12844 TX ID 242 Temp 70 8 Time 02 15 53 s RSSI 1 13mw Humidity 25 Packet 12845 TX ID Temp 70 3 Time 02 15 54 RSSI 0 18mw Humidity 24 Packet 12846 TX ID Temp 70 3 Time 02 15 58 RSSI 1 37mw Humidity 25 Packet 12847 TX ID 242 Ti 70 8 Time 02 15 58 umi RSSI 0 82mw Humidity 25 e Packet 12848 TX 2D T 70 3 Time 02 16 02 umi RSSI 1 42mw Humidity 21 TX ID Temp 70 8 RSSI 0 74mw Humidity 26 Se Packet 12850 TX ID 242 Ti 70 3 Time 02 16 06 umi RSSI 1 32mw Humidity 26 5 Packet 12851 TX ID 242 Ti 70 3 Time 02 16 10 umi RSSI 1 42mw Humidity 26 Was Packet 12852 TX ID 242 Ti 70 8 Time 02 16 13 un RSSI 0 82mw Humidity 26 Se Packet 12853 TX ID 242 Temp 70 3 Time 02 16 13 RSSI 1 42mw Humidity 25 z UN Packet 12854 TX ID Temp 70 3 Time 02 16 17 RSSI 1 37mw Humidity 25 Bu Packet 12855 TX ID 242 Temp 70 8 Time 02 16 21 RSSI 0 74mw Humidity 25 e Packet 12856 Time 02 16 21 TX ID 242 Temp 70 3 RSSI 1 37mw Humidity 26 Packet 12857
20. A POWER AND DATA TRANSMITTER TX91501 3W ID The transmitter was manufactured by Powercast Co It is powered at 3 watts and uses a data integrated 8dBi antenna at a center FIGURE 4 POWER AND DATA TRANSMITTER frequency of 915 MHz The device sends a pre programmed transmitter ID that was received by the Powercast chip P2110 and decoded by the microcontroller unit MCU on the Wireless Sensor Board The transmitter provides a 60 degree beam pattern for width and height meaning that as long as the sensor nodes are placed within the angles defined by the transmitter the RF signal will be able to power the nodes efficiently B MICROCHIP 16 BIT XLP DEVELOPMENT BOARD DM240311 This board included in the Powercast P2110 Eval 01 kit was a development platform featuring Microchip s PIC24F MCU that was pre programmed to operate as an access point for receiving FIGURE 5 16 BIT XLP DEVELOPMENT BOARD 12 4 2013 Wireless Sensor Network Health Diagnostic 13 ECE 480 Design Team 2 data from the included Wireless Sensor Boards This board was connected directly to the computer via a USB cable and acts as the cluster head for all sensor nodes C MICROCHIP MRF24J40 PICTAIL PICTAIL PLUS DAUGHTER BOARD AC164134 1 A radio 2 4GHz IEEE 802 15 4 that plugs into the 16 bit XLP 3 8 Development Board for receiving data from the Wireless Sensor Boards D P2110 EVALUATION BOARD P2110 EVB Thi
21. Direct Path No Angle Lt Sensor Nodes Transmitter FIGURE 202 NETWORK TEST SETUP SAMPLE2 8 The last test to be completed was sample 2 9 This test focused on the Teer maximum width angle that was allowable from the Powercast transmitter If the sensor nodes were place outside of the maximum allowable angle the capacitor would Sensor not charge and no data transmission Nodes would take place This situation would result in a complete node failure in the network and the fact that no new data would be obtained was the clear indicator of that This issue could easily be seen as a power issue similar to battery life of currently FIGURE 21 NETWORK TEST SETUP SAMPLE2 9 used sensor nodes in the industry Again along with all other tests no strange data was obtained via the HyperTerminal All sensors were operating correctly reaffirming a good health for the network 12 4 2013 Wireless Sensor Network Health Diagnostic 25 ECE 480 Design Team 2 5 FINAL COST SCHEDULE SUMMARY AND CONCLUSIONS FINAL COSTS As discussed before Team 2 was granted an increase budget of 2000 for this project Although the budget was 2000 the team was able to successfully complete the project with less than two thirds of that amount As you can see in Table 6 below only 1 250 of the 2000 the team received was necessary Luckily the team was able to acquire everything they needed in one development kit This helped the team ke
22. ES EE EEEE 9 irl M 10 Project timeline Gantt Chart terri v Funera in eee eR EVE ER PE ERA VR ERE ERR NUR 11 12 4 2013 Wireless Sensor Network Health Diagnostic 3 ECE 480 Design Team 2 3 Technical description of work performed ccccccecssscccssssececsesececsessececsesaececsesaececeesaececsesaseeesesaeeeeseaaes 11 Hardware Desigri eh et eter o e eee reete eer e ee EE Ser eret tua e EE UE M eges 11 A Power and Data Transmitter TXO1SO 2WID nnnn nnne nnnn nnn 13 B Microchip 16 bit XLP Development Board DM240311 ssssssesseeseeeeenneer nennen 13 C Microchip MRF24J40 PlICtail PlCtail Plus Daughter Board AC164134 1 sees 14 D P2110 Evaluation Board P3IIO ENBI enne enne ne entes nesnten tnnt tte te nnns enne 14 Patchiand Dipole ANtenNaS e e err eite er tet en ive regi eyes 14 E Wireless Sensor Board WSN EVAL O1 cccccscccecssssececsssseceessssececessaececeesseceessaaeceesesaeceeseaaeeeeseaaes 14 F PICkit 3 Programmer Debugger DGiGA120 eene enne nnn entren 15 Hardware Implementaton esses nennen enne nenne neta nass s essi teta assa sies enses asas sss ss sena nan 15 Wireless Sensor Network Flow Chart 18 Software designirequiremehnts eerte ere een T exe ere e Tx Rees ede tessa 19 Software implementations cer Eee eee tent e ete E Pe eeu e Ee ee 19 4 Test data with proof of functional design 22 Har
23. Fri 9 20 13 Sun 9 22 13 ECE 480 Design Team 2 Predecessors Resource Names Deadline NA Wed Brad Stu David Kelly 9 4 13 Brad Stu David Kelly NA Brad Stu David Kelly NA 2 Brad Stu David Kelly Fri 9 13 13 Mon 5 Stu 9 16 13 Stu NA Stu David Brad Kelly NA David NA Wireless Sensor Network Health Diagnostic 33 First Contact Wed 1 day w Sponsor 10 2 13 Mon Propsal 15 days 9 23 13 Configure Sensor Network Fri 16 days amp Verify Correct 10 4 13 Sensor Readings Research amp Fri Order 2 days 10 4 13 Sensors SOC Study amp Configuration of Thu IEEE 802 15 4 7 days 10 10 13 MiWi P2P Protocol Verification of Sat Correct Sensor 6 days 10 19 13 Readings Design Day Mon 10 days Team Page Work 9 23 13 Oral Proposal Mon Presentation 10 days 9 23 13 Practice Team Progress Mon 5 days Report 1 10 28 13 Design Issues 24 days Tue 12 4 2013 Wed 10 2 13 Fri 8 10 11 13 Fri 6 10 25 13 Mon 10 7 13 Fri 10 18 13 Fri 10 25 13 Fri 10 4 13 Fri 10 4 13 Fri 11 1 13 Fri ECE 480 Design Team 2 Stu David Brad Kelly Stu David Brad Kelly Stu David Brad Kelly Brad David Stu Brad Kelly Kelly David Kelly Brad Stu David Kelly Brad Stu David Kelly Brad Stu David Kelly Wireless Sensor Network Health Diagnostic Wed 10 2 13 NA NA Mon 10 7 13 Fri 10 18 13 Fri 10 25 13 NA NA NA NA 34 ECE 480 Design T
24. TX ID Light 1x Time 03 44 49 dT 00 05 RSSI 1 53mw Extrnl 982 mv Packet 11051 Node 1 TX ID 242 Light 1x Time 03 44 51 dT 00 03 RSSI 2 08mw Extrnl Packet 411052 Node 2 TX ID Light Time 03 44 53 dT 00 04 RSSI 1 42mw Extrnl Light dx Time 03 5 dT 00 04 RSSI 2 22mw Extrnl 903 mv Packet 11054 Node 2 Time 03 44 57 dT 00 04 TX ID RSSI 1 63mw l Packet 11053 Node 1 TX ID Light 1x Extrnl 980 mv Packet 11055 Node 1 Time 03 44 58 dT 00 03 TX ID 242 RSSI 2 33mw Light 1x Time 03 45 00 dT 00 02 RSSI 2 74mw Extrn 891 mv Packet 11057 Node 2 Time 03 45 01 dT 00 04 TX ID 242 RSSI 1 63mw Light 1x Extrnl 985 mv l Packet 11056 Node 1 TX ID 242 Packet 11058 Node 1 TX ID 242 I Time 03 45 03 dT 00 03 RSSI 2 08mw Packet 411059 Node 2 TX ID Time 03 45 04 dT 00 03 RSSI 1 63mw Packet 11060 Node 1 TX ID Light 1x Time 03 45 05 dT 00 02 RSSI 2 33mw Extrn 903 mv Packet 11061 Node 1 TX ID 242 Light 1x Time 03 45 08 dT 00 03 RSSI 2 22mw Extrn 888 mv Packet 11062 Node 2 TX ID Temp F Light_ ax Time 03 45 09 dT 00 05 RSSI 0 13mw Humidity Extrnl 982 mv H D Sample 2 6 Test Data 12 4 2013 Wireless Sensor Network Health Diagnostic 40 12 4 2013 File Edit Format View Help Packet 12582 Time 04 30 09 Packet 1
25. TX ID Temp 70 3 Time 02 16 25 RSSI 1 32mw Humidity 25 P1 WN m KTM wn KTM m wen KTM wn KTM wn KT KTM wn KTM wn RN Packet 12858 Time 02 16 28 Dr TX ID Temp 70 8 Light 276 RSSI 0 78mw Humidity 25 Extrnl 1096 Sample 2 9 Test Data 12 4 2013 Wireless Sensor Network Health Diagnostic 42 ECE 480 Design Team 2 File Control Setup Trigger Measure Analyze Utilities Help 11 Nov 2013 1 43 PM Acquisition is stopped 10 0 MSa s 10 0 Mpts V max 1 Current 3 5336 V 3 5293 V Mean 3 5336 V 3 5293 V Min 3 5336 V 3 5293 V Max 3 5336 V 3 5293 V Sample 2 3 JP2 Voltage Screen Capture File Control Setup Trigger Measure Analyze Utilities Help 11 Nov 2013 2 15 PM 2 V max 1 Current 3 5220 V 3 5338 V Mean 1 37528 V 2 16501 V Min 172 2 mV 360 7 mV Max 3 5507 V SN Sample 2 4 JP2 Voltage Screen Capture 12 4 2013 Wireless Sensor Network Health Diagnostic 43 ECE 480 Design Team 2 More 10f2 Delete All File Control Setup Trigger Measure Analyze Utilities Help Current Mean Min Max ax 2 V max 1 3 5383 V SN 1 98484 V 1 90690 V 172 2 mV 360 7 mV 3 6087 V 3 5941 V 11 Noy 2013 2 46 PM Sample 2 5 JP2 Voltage Screen Capture More 1 of 2 Delete All File Control Setup Trigger Measure Analyze U
26. VERSION 1 0 DECEMBER 4 2013 Bae AFRL THE AIR FORCE RESEARCH LABORATORY LEAD DISCOVER DEVELOP DELIVER WIRELESS SENSOR NETWORK HEALTH DIAGNOSTIC FINAL REPORT PRESENTED BY DAVID ROGERS KELLY DESMOND STU ANDRZEJEWSKI BRAD GARROD MICHIGAN STATE UNIVERSITY COLLEGE OF ENGINEERING ECE 480 DESIGN TEAM 2 ECE 480 Design Team 2 EXECUTIVE SUMMARY The Air Force Research Laboratory has proposed a project of developing a diagnostic tool to best determine the health of a wireless sensor network The main objective of the project was to scientifically determine the best set of metrics that indicate that a node was about to malfunction was malfunctioning or has malfunctioned In order to accomplish this objective a wireless sensor network was configured to collect external metrics about the environment being monitored and internal metrics about the sensor nodes themselves External metrics included temperature relative humidity and light internal metrics included received signal strength and current draw Over the course of the project the design team configured a wireless sensor network developed software to process the sensor network data in real time display the data in a user friendly manner and alert operators of problems in the network A graphical user interface was written from scratch in Python with the help of external libraries wxPython and matplotlib The software architecture leveraged the model view contr
27. able below displays therankings of our conceptual designs based on theratings from the selection matrix Design Powercast Sun SPOT National Instruments Crossbow MTS 400 P2210 Eval 01 Rev 8 WSN Starter Kit 4 Worst o fe s L5 TABLE 2 SENSOR NETWORK DEVELOPMENT KIT SELECTION MATRIX BUDGET At the beginning of the semester the team was given a 500 budget This budget was for all essential hardware and software components for the design project Luckily the team had the Air Force Research 12 4 2013 Wireless Sensor Network Health Diagnostic 10 ECE 480 Design Team 2 Lab as a sponsor and was able to request a larger budget within the proposal As you can see below in Table 3 the team requested a 2 000 budget After receiving the project proposal the Air Force Research Lab accepted it as well as the increased budget Hardware Components Price Powercast P2110 EVAL 01 Development kit Additional Sensors Temperature Light Humidity Additional Node Engineering Shop Services Fees TABLE 3 ESTIMATED BUDGET PROJECT TIMELINE GANTT CHART Attached at the end of the report in Appendix 3 is the original Gantt Chart for the project and an updated Gantt Chart for the actual tasks accomplished The major difference between the two different timetables was the available time spent on different portions of the project These time changes aside a great deal of the project followed the original timelin
28. acitor that was ideal for the development kit Once both nodes were built and configured and the power transmitter was plugged in it was time to go back to the PC and make sure the development board and PC were working correctly reading the data being sent from the wireless sensor nodes HyperTerminal was opened and the team witnessed all the data being sent from each sensor on each node to the PC via the development board successfully See Figure 12 to view the HyperTerminal reading in data from the wireless sensor nodes 12 4 2013 Wireless Sensor Network Health Diagnostic 17 ECE 480 Design Team 2 38 Node 2 TX ID Temp 71 8 F Light 145 Ix Time 00 04 16 dT 00 00 RSSI 2 22mW Humidity 31 X Extrnl 1728 mV 39 Node 2 Temp 71 8 Light 138 Ix Time 00 04 17 dT 00 01 Rest 6 82mi mie 31 Extrnl 1720 mV Node 2 TX ID 242 Temp Light 153 Ix Extrnl 1742 mV Node 2 Light 120 1x dT 00 01 Extrnl 1751 aV FI xi F xi FI xi Node 2 TX ID Tem F Light 3113 Ix dT 00 01 RSSI 4 84mi un 3 X Extend 1711 nV ki xi Ei xi Fi xi iT i dT 00 01 RSSI 6 40mM is Rest KU Hombdity 3 Node 2 TX ID 242 Temp Light 117 lx dT 00 01 RSSI 7 28mi Ve E Extrnl 1668 mV Node 2 TX ID 242 Temp Light 117 lx dT 00 02 RSSI 1 74mM eh Extrnl 1736 nV i 45 Node SI Tem Light 117 lx Time 00 04 26 dT 00 03 Rest 1 Aw Ve Extrnl 1718 mV SCROLL CAPS NUM C
29. analysis was a simple analysis that was implemented to see if data came in at a zero value where it was not near zero before This was very similar to the short term analysis in terms of determining when a packet might actually read a zero value however if it was unusual an assumption could be made of malfunction Upon further research into sensor failure an open circuit in the nodes resulted in no data being transferred but rather a string of which will then be switched to a 0 value and look like an open circuit failure 12 4 2013 Wireless Sensor Network Health Diagnostic 21 ECE 480 Design Team 2 4 TEST DATA WITH PROOF OF FUNCTIONAL DESIGN After the successful configuration of the wireless sensor network metric analysis of the network was necessary to determine the health Testing was split into 2 different categories hardware functionality and software testing HARDWARE FUNCTIONALITY The tests carried out for the hardware concern the receiving of correct data packet information The HyperTerminal shown in Hardware Implementation displays the readings of the sensors located on the Wireless Sensor Board If there are issues with a particular sensor or sensors those readings will fluctuate or become unreadable Because of the limitations of RF energy short distance testing was done when confirming hardware functionality lo e Temp Light and Additionally line of sight for the transmitter E Humidity Jumper
30. apture Preit echo FIGURE 12 HYPERTERMINAL SESSION WITH STREAMING SENSOR NETWORK DATA WIRELESS SENSOR NETWORK FLOW CHART Power Transmitter Temperature Sensor Light Sensor Packet Humidity Sensor Temperature Sensor Light Sensor Power Power Receiver Receiver Humidity Sensor Sensor Sensor Board Board Packet 16 Bit XLP Development Board Cluster Head FIGURE 13 WIRELESS SENSOR NETWORK FLOWCHART 12 4 2013 Wireless Sensor Network Health Diagnostic 18 ECE 480 Design Team 2 SOFTWARE DESIGN REQUIREMENTS The node sensor network that was purchased for this project came with the necessary software to simply run and gather data from each of the sensors in a simple manner making setting up the sensor network less time consuming After the network was set up to collect sensor data packets the data had to run through processing manipulation and display programs This was set up into a model view controller software pattern to accomplish this The model portion does the backend data analysis list concatenation of data and set up all parameters to run into the view portion which is a graphical user interface GUI to view the sensor data in real time The main software requirement embedded into the scope of the project was data representation in combination with analysis and failure detection This visual data representation that is shown in the GUI is essential in viewing nodal failures or misrepresenta
31. changing The average voltage of 3 5V for the capacitor can be seen for all tests which gives a calculated average current draw of 350mA These results were very consistent during all the tests and even with certain sensors not active no significant change could be observed 12 4 2013 Wireless Sensor Network Health Diagnostic 23 ECE 480 Design Team 2 Transmitter 2 5 Inches FIGURE 20 NETWORK TEST SETUP SAMPLE2 3 THRU SAMPLE2 6 Test sample 2 7 increased the distance and height that the sensor nodes were from the transmitter and by doing so RSSI greatly decreased and dT greatly increased These factors were expected considering the limitations of RF energy but the information obtained from the sensors was very consistent and matched the current conditions in the testing lab Even though the time to receive the data had been increased the network health was still in good despite the strength decrease Transmitter Sensor Nodes 14 8 Feet FIGURE 19 NETWORK TEST SETUP SAMPLE2 7 12 4 2013 Wireless Sensor Network Health Diagnostic 24 ECE 480 Design Team 2 Test sample 2 8 was completed specifically for the software that had not been implemented yet It was the longest test so it could provide a large sample of data that could be displayed by the GUI interface that was being developed for it No irregularities were found during testing and all nodes functioned properly throughout the testing period
32. de 1 TX ID 242 Temp 70 7 F Light 1x Time 02 37 37 dT 00 02 RSSI 3 03mw Humidity 36 Extrnl 985 mv Packet 7942 Node 2 TX ID 242 Temp 71 0 F Light_ 1x Time 02 37 38 dT 00 02 RSSI 1 74mw Humidity 36 extrnl 1100 mv D Sample 2 4 Test Data 39 ECE 480 Design Team 2 semple2 5 Notepa S mim Die Edit Format View Help Packet 9378 Node 2 TX ID Temp 71 6 F Light 177 1x 3 Time 03 11 28 dT 00 02 RSSI 2 61mw Humidity X Extrn 1636 mv Packet 9379 Node 1 TX ID 242 Temp 71 0 F Light 141 lx Time 03 11 28 dT 00 02 RSSI 2 08mw Humidity Extrnl 1528 mv Packet 9380 Node 2 TX ID 242 Temp 71 6 F Light 185 1x Time 03 11 30 dT 00 02 RSSI 2 61mw Humidity X Extrn 1634 mv Packet 9381 Node 2 TX ID 242 Temp 71 6 F Light 175 1x Time 03 11 32 dT 00 02 RSSI 2 74mw Humidity Extrnl 1637 mv Packet 9382 Node 1 TX ID 242 Temp 70 9 F Light 141 1x Time 03 11 32 dT 00 04 RSSI 2 08mw Humidity Extrnl 1540 mv Packet amp 9383 Node 2 TX ID 242 Temp 71 6 F Light 177 1x Time 03 11 34 dT 00 02 RSSI 2 61mw Humidity Extrnl 1634 mv Packet 9384 Node 1 TX ID 242 Temp 70 9 F Light 141 lx Time 03 11 35 dT 00 02 RSSI 2 33mw Humidity Extrnl 1540 mv Packet 9385 Node 2 TX ID 242 Temp 71 6 F Light 377 1x Time 03 11 36 dT 00 02 RSSI 2 48mw Humidity Extrnl 1633 mv Packet
33. der to ensure they are functioning properly The number of properly functioning nodes has a direct impact on the health of the wireless sensor network This project consists of configuring a wireless sensor network and monitoring a few external parameters such as environmental metrics like light temperature and humidity as well as internal network parameters such as sensor node current voltage received signal strength RF transmission power and channel availability Developing a diagnostic tool to monitor the health of a wireless sensor network is an application of particular interest to the United States Air Force BACKGROUND The Air Force has used wireless sensor networks for many years but has just recently been examining ways to better monitor the health of their networks Their work on developing a health diagnostic for wireless sensor networks did not lead to a definitive solution which led them to create an open design project for senior capstone teams They worked with multiple sensor networks of Sun SPOT and Crossbow sensor nodes organized in a mesh network topology as shown in Figure 1 In a mesh network topology each node must be able to collect and distribute its own data and serve as a relay for other nodes in order to propagate the data throughout the network Wireless Mesh Network The main advantage of this topology includes its robustness to fall E Hall failing nodes but a drawback was that the nodes consume 7 N la
34. dware Functionality cesses nnne sn ennt nana sisse sentita dass sse satia sans ases ensi gan 22 Hardware Tests ettet mte to ie ota ati a i et dtt tate ete nera ec iat 22 5 Final cost schedule summary and conchuslons eene enne nh nnne ennt 26 dare ice M 26 Schedules i ete e ri ox Per ERI Fe IM E 26 CONCIUSION EE 26 Appendix 1 Technical roles responsibilities and work accomplished sees 28 Appendix 2 Literature and website references ccccccccccesssssssscecececesseseaeseceeecesseseeaeeeseesseeseasaeeeesesseeseaa 32 Appendix 3 Gantt Charts ete rtr hee tr esie needed rte tuse 33 Appendix 4 Test Data amp Screen Capture eene nnne enti nnns s esse setas a nasse entia 39 12 4 2013 Wireless Sensor Network Health Diagnostic 4 ECE 480 Design Team 2 1 INTRODUCTION AND BACKGROUND INTRODUCTION Wireless sensor networks are commonly used to monitor important environmental information such as temperature or light level which may alert users of hazardous conditions for themselves or machinery However wireless sensors typically have very limited power and memory and consequently node malfunction or failure was common A network of largely malfunctioning nodes can mislead users analyzing the data of the network and may lead to dire consequences Thus it is very important to monitor the health of the nodes in the network in or
35. e and is well documented and engineered for future expansion More features on graphical user interface can be added by fully leveraging the capabilities of the wxPython and matplolib libraries Additionally new algorithms for network health analysis can be implemented in the model module Overall the project is readily suitable for future development The project that was chosen by the team resulted in an outstanding learning experience for each member The team was fully immersed in the design process and each member will take with them a new set of skills to industry Some of these soft skills include learning how to work in a team setting outstanding written and oral communication overcoming and learning from failures and technical skills such as programming and configuring hardware Ultimately this design class taught each member how to become a professional and skills to utilize in their future careers 12 4 2013 Wireless Sensor Network Health Diagnostic 27 ECE 480 Design Team 2 APPENDIX 1 TECHNICAL ROLES RESPONSIBILITIES AND WORK ACCOMPLISHED KELLY DESMOND SENSOR NETWORK CONFIGURATION AND MAINTENANCE Kelly s technical role this semester was Sensor Network Configuration and Maintenance With this role Kelly had to make sure the team was ordering the correct wireless sensor network development kit as well as be in charge of all the hardware of that network throughout the entire semester His first role was to mak
36. e sure that the development kit that was being ex 1 purchased by the team had could successfully fulfill all the requirements from the Air Force Research Laboratory These requirements included low r power must include but not limited to temperature humidity and light sensors and the team must be able to monitor internal and external parameters Kelly was able to find the P2110 EVAL 01 Lifetime Power Energy Harvesting Development Kit for Wireless Sensors manufactured by Powercast and Microchip Once the development kit arrived Kelly had to learn everything about it He had to know what each part was and how each part connected so that the group could easily construct and deconstruct the network quick and efficiently for tests Before the team could start the tests Kelly had to configure the network He had to ensure proper construction of the network and make sure it could communicate correctly with the PC A terminal emulator needed to be downloaded installed and setup for this purpose HyperTerminal a terminal emulator from Powercast was recommended and after some research deemed to be the best option Kelly used HyperTerminal to set up the COM3 port on the PC so that the network would be able to send its data to the PC to be recorded Once the network was configured the team needed to know how to interact with the hardware to help create different tests and read different information such as current being used by the
37. e with little to no issues 3 TECHNICAL DESCRIPTION OF WORK PERFORMED HARDWARE DESIGN With regards to reliability time and scope of the project the hardware design of the sensors used in the project were bought rather than designed and built This allowed for more time to focus on developing a diagnostic tool for determining when a sensor node was malfunctioning However this did not make things simple since there were plenty of options when purchasing the sensor network and many hardware considerations had to be made These included choice of power cost getting the correct data input accuracy of sensors and ease of use When looking upon the choice of power there were three main types that were found to be common power adapter plugged into a source battery powered and RF powered The first option of having a sensor that was plugged into a source of power was immediately abandoned since this would eliminate 12 4 2013 Wireless Sensor Network Health Diagnostic 11 ECE 480 Design Team 2 the wireless factor in our project Although the reliability of the sensor nodes would increase without the need of monitoring voltage levels their usability would greatly decrease and therefore not desirable Battery powered sensors are very common amongst wireless sensors and provide a reliable form of power source however this adds additional complexity in detecting failed nodes As the battery decays over time it may not provide enough v
38. eam 2 Paper 9 17 13 10 18 13 Team Progress Report 2 amp Mon Fri 5 days 18 Brad Stu David Kelly NA Project 11 18 13 11 22 13 Demonstration Identify Simple Sat Fri Network Health 11 days 15 Brad Stu David Kelly Fri 11 8 13 10 26 13 11 8 13 Metrics Develop Graphical User Sat Fri Fri Interface GUI 11 days 21 Stu Brad David Kelly 11 9 13 11 22 13 11 22 13 for Configuring Sensor Nodes Confirm Health Diagnostics amp Implement Sat Sun Sun Network 7 days 22 Stu Brad David Kelly 11 23 13 12 1 13 12 1 13 Security Functionality If Time Allows Wed Final Reports Brad Stu David Kelly NA 12 4 13 Fri Fri Design Day 1 day 23 Stu David Brad Kelly Fri 12 6 13 12 6 13 12 6 13 12 4 2013 Wireless Sensor Network Health Diagnostic 35 ECE 480 Design Team 2 Gantt Chart Final Task Name Duration Start Finish Predecessors Resource Names Deadline Project Wed Mon Overview amp 9 days NA 9 4 13 9 16 13 Tasks First Group Meeting Wed Wed Wed Project 1 day Brad Stu David Kelly 9 4 13 9 4 13 9 4 13 Assignment amp Initial Tasks Meeting Wed Wed Times amp 1 day Brad Stu David Kelly NA 9 4 13 9 4 13 Scheduling First Meeting Tue Tue w Group 1 day Brad Stu David Kelly NA 9 10 13 9 10 13 Facilitator Air Force Research Thu Fri Fri 7 days 2 Brad Stu David Kelly Labortory 9 5 13 9 13 13 9 13 13 AFRL Proposal Submit AFRL Mon Mon Mon 1 day 5 Stu Proposal 9 16 13 9 16 13 9 16 13 Mon Tue GANTT Chart 2 da
39. ecessary tools to program the System On Chip the total cost was approximately 100 for a node 75 for a cluster head Creating a complete accurate sensor network would require a semester of work itself Communication kit with sensor configuration Be effective in communicating easily between node and cluster head with simple analog signals Still need a circuit design for internal metrics of sensor The communication kits would reduce what is needed circuit wise however with a price of the cost of product With this kit only sensors are needed Cost approximately 300 for one development kit 50 for all the sensors needed Feasible 8 10 Having this development kit would allow for less time settingup and more time determining the metrics that cause failure Only extra time is configuring to send correct information from sensors Full sensor network development kit Already measures internal and external metrics with easy programmable communication chip All sensors are verified to factory specs and internal node measurements are able to be tracked One of the full sensor development kits would fulfill all hardware requirements Extra sensors for robust testing would be obtained Cost is approximately 1200 for development kit 20 for extra sensors to test Feasible 6 10 The full sensor kit would eliminate a lot of configuration time Additional sensors could be added Very
40. ed selecting network hardware choosing a flexible programming language identifying a proper software design pattern and pinpointing reliable metric analysis algorithms Other project obstacles involved writing an extra proposal for the United States Air Force Research Laboratory and being unable to contact and speak with sponsor due to job related 12 4 2013 Wireless Sensor Network Health Diagnostic 26 ECE 480 Design Team 2 activities and the US government shutdown Due to the shutdown the team was unable to receive any design components previously used by the Air Force Research Laboratory Starting delays aside the team was still able to proficiently analyze multiple design choices The team decided to purchase a full wireless sensor network development kit in order to strengthen its focus on analyzing metrics within the wireless sensor network that would best determine its health Furthermore using a development kit allowed for quick setup and configuration In choosing this option the team made a great decision and was able to create a full working prototype on time and on budget Along with fulfilling all of the design requirements the team was able to design and develop a project with a solid foundation for future improvements Since the wireless sensor network is easily configurable a future team can easily pick up where this design team left off The software is written in Python a simple and flexible programming languag
41. ep within their budget Product Cost Powercast P2110 EVAL 01 Development Kit 1 250 Total 1 250 TABLE 6 FINAL COSTS SCHEDULE The schedule consisted of the initial project acceptance phase research and design choices and lastly network implementation At the beginning of the semester the team had to decide what direction to take the project in regards to the design choices This was due to the fact that the Air Force Research Laboratory required an additional proposal separate from that of the ECE 480 course which needed to be accepted before the team could start the design process After acceptance the team was able to obtain the necessary budget which allowed them to enter the research and design choices phase of the project During this phase the team obtained a suitable wireless sensor network development kit for network health monitoring purposes Once the network was configured the team was able to test the network implement their software and introduce a graphical user interface This allowed them to monitor the health of the wireless sensor network in real time The team was pleased that that they were able to stick to the schedule and meet all deadlines The phases described above can be seen in more detail in the Final Gantt Chart in Appendix 3 CONCLUSION Overall the project was a success Various design issues were identified earlier in the process and properly addressed These design issues includ
42. he nodes such as current and received signal strength These metrics will monitor the environment the network was deployed in and also track relevant data about the health of each sensor node itself These metrics are the crux of the project goal and are a crucial design parameter B INCREASES DESIGN DESIRABILITY V SIMPLE NETWORK CONFIGURATION It was particularly desirable to choose a network design that was easy to setup and configure A network that can scale up to handle more sensor nodes was another desirable feature This will allow the customer to customize the sensor network at any point during its lifetime Easy set up and customization of the network was an important part of the project and very desirable to the customer VI DEVELOPMENT OF A GRAPHICAL USER INTERFACE A visually appealing graphical user interface that displays the data obtained from the sensor network should be developed to allow an operator to quickly solve problems when they arise A graphical user interface will make it easier to pin point what is going on in the network at all times This will help when failing nodes are identified This part of the project isn t necessarily required but a user friendly visual would significantly increase the desirability A diagnostic tool is only useful if it is easy to us and with a handy graphical user interface it will allow an operator to easily diagnose problems within a network FAST DIAGRAM The Function Analysis Sys
43. her components would be needed to measure current draw and voltage drop in the internal circuit at each individual node B ZIGBEE NETWORK KIT The next group of considerations was buying ZigBee communication network kit This would require adding additional circuits for the sensors that are fed into the analog input output port on each network node Similarly this group would need additional circuitry to determine internal metrics of the sensor node 12 4 2013 Wireless Sensor Network Health Diagnostic 8 ECE 480 Design Team 2 C SENSOR NETWORK DEVELOPMENT KIT The last group of considerations was a full sensor development kit with multiple sensor nodes and a cluster head These kits had full configured sensors as well as analog input output ports for adding additional sensors In addition the kits that were considered measured internal node metrics CHOSEN DESIGN SOLUTION The team created a feasibility matrix to help identify the best hardware design The feasibility matrix compares multiple designs to different vital parameters such as functionality cost and time Each design was ranked from 1 10 for each parameter then totaled for a complete score out of 30 See Table 1 below sensor network Functionality Highly dependent on quality of design in the given amount of time Generally would not be as accurate because of the lack oftime given and able to be put into the design Lacked a cluster head design With the n
44. ht 1x Time 02 37 23 dT 00 04 RSSI 1 63mw Humidity 35 X extrnl 1084 mv Packet 7931 Node 1 TX ID 242 Temp 70 7 F Light 1x Time 02 37 24 dT 00 01 RssI 2 74mw Humidity 36 Extrnl 985 mv Packet amp 7932 Node 2 TX ID 242 Tem 71 0 F Light_ 1x Time 02 37 25 dT 00 02 RSSI 1 63mw Humidity 35 X Extrnl 1102 mv Packet 7933 Node 1 TX ID 242 Temp 70 7 F Light 1x Time 02 37 26 dT 00 02 RSSI 2 48mw Humidity 36 Extrnl 982 mv Packet 7934 Node 1 TX ID Temp 70 7 F Light 1x Time 02 37 28 dT 00 01 RSSI 2 48mw Humidity 36 Extrnl 987 mv Packet 7935 Node 2 TX ID 242 Tem 71 0 F Light 1x Time 02 37 29 dT 00 04 RSSI 1 53mw Humidity 35 X extrnl 1087 mv Packet 7936 Node 1 TX ID 242 Temp 70 7 F Light 1x Time 02 37 30 dT 00 02 RSSI 2 48mw Humidity 36 Extrnl 985 mv Packet 7937 Node 2 TX ID 242 Temp 71 0 F Light dx Time 02 37 32 dT 00 03 RSSI 1 74mw Humidity 35 X Extrn 1087 mv Packet 7938 Node 1 TX ID 242 Temp 70 7 F Light 1x Time 02 37 32 dT 00 02 RSSI 2 48mw Humidity 36 X Extrnl 982 mv Packet 7939 Node 1 TX ID Temp 70 7 F Light 1x Time 02 37 35 dT 00 02 RSSI 2 61mw Humidity 36 Extrnl 979 mv Packet 7940 Node 2 TX ID 242 Temp 71 0 F Light 1x Time 02 37 35 dT 00 03 RSSI 1 74mw Humidity 36 Extrn 1096 mv Packet 7941 No
45. ments that will be monitored must include but not limited to temperature humidity and light Also a network topology must be decided This parameter of the project was absolutely necessary Il LOW POWER SENSOR NODES Anode must be able last sufficiently long while deployed which includes tasks such as collecting accurate data and transmitting a high fidelity signal to the cluster head In order to accomplish these tasks for a 12 4 2013 Wireless Sensor Network Health Diagnostic 6 ECE 480 Design Team 2 long period of time the node must be low powered Sensor nodes with long lifetime will decrease collection errors and ultimately allow for greater accuracy in nodal failure detection This parameter was very important Ill RELIABLE COMMUNICATION WITHIN THE NETWORK To assist in determining if a node or sensor is failing the data that is retrieved by the cluster head and analyzed in the mainframe computer must be accurate Signal accuracy is a combination of signal transmission power received signal strength and communication protocol Data fidelity is highly important in order to develop reliable metrics that diagnose the health of the network This parameter was also very important to the customer IV ACCURATE SENSOR NODE MEASUREMENTS The sensor node must be able to communicate not only reliable but also with accurate data This includes external metrics such as temperature light humidity etc but also internal metrics about t
46. nd monitoring activity of the external sensors also helped in error analysis Accuracy of the sensors was also a very important in the design decision Flawed or inaccurate data can impact failure analysis algorithms This could result in false positives when a node is deemed to have failed but is actual functioning properly or false negatives when a failed node is not detected Ease of use was a big factor in selecting a sensor network The ability to set up the network and get it running quickly was a top priority This allowed for minimal wasted time in designing and building the network itself and maximizing the time in researching the metrics and developing a diagnostic tool As show previously in Table 2 the Powercast P2210 Eval 01 development kit was chosen as the best option for the AFRL network health monitoring purposes Included in the kit were components developed and built by Powercast Co and Microchip The items included are shown in Figure 3 on the next page 12 4 2013 Wireless Sensor Network Health Diagnostic 12 ECE 480 Design Team 2 FIGURE 3 POWERCAST P2110 EVAL 01 DEVELOPMENT KIT e Power and Data Transmitter TX91501 3W ID e P2110 Evaluation Board 92110 EVB e Patch Antennas 2 e Dipole Antennas 2 e Wireless Sensor Board WSN EVAL 01 e Microchip 16 bit XLP Development Board DM240311 e Microchip MRF24J40 PlCtail PICtail Plus Daughter Board AC164132 1 e PlICkit Programmer Debugger PG164130
47. ntation can be run in both real time mode and demo mode In real time mode data from the sensors is read from the cluster head over USB as it streams in This required understanding and using the pySerial library to read bytes from a serial connection In demo mode previously recorded sample data can be read in from a file and simulated as if the data is streamed in real time He also successfully created a model API for which any view can interface to in order to display the data The model is also extensible to be called for data analysis purposes Most importantly the model processes the data from the controller quickly and efficiently so that many calls to the model can be made without crashing the application David also created a graphical user interface the view so that an operator can see network data and the status of each node in real time The view required using matplotlib to plot the data and using wxPython to create the rest of the user interface The plots can be saved at any time and the time range and metric data shown can all be configured at run time This view is very extensible and there is room left for further development Ultimately David fulfilled his role with flying colors He also set up a git repository in order to facilitate the software engineering process and track changes over the course of the project Additionally he implemented a logger within the software for documentation and debugging purposes Finall
48. oller design pattern Additionally a number of algorithms were implemented in order to detect failing nodes These algorithms include short term long term and zero value analysis of real time sensor data The software was designed for use in real time and demonstration modes in order to rigorously test and verify proper detection of failing nodes Ultimately the design team successfully met requirements and developed a robust application extensible for further development 12 4 2013 Wireless Sensor Network Health Diagnostic 1 ECE 480 Design Team 2 ACKNOWLEDGEMENTS AIR FORCE RESEARCH LAB A special thanks to Mr Joseph Natarian Mr Kenneth Littlejohn and the entire student challenge team at the Air Force Research Lab located at Wright Patterson Air Force Base DR TONGTONG LI A special thanks to Dr Tongtong Li the faculty facilitator for assisting in the design process and all of encouragement along the way DR TIMOTHY GROTJOHN AND DR LALITA UDPA Thanks to Dr Grotjohn and Dr Udpa the professors for the ECE 480 course this semester Both of you were very helpful and reliable STEPHEN A ZAJAC AND DR GREGORY M WIERZBA Thanks to Mr Zajac and Dr Wierzba for all the hard work they put into the ECE 480 Lab which truly challenged design team 2 and helped us become true electrical engineers ELECTRICAL AND COMPUTER ENGINEERING SHOP Thanks to Gregg Mulder and Brian Wright of the ECE shop for providing all the necessary
49. oltage to give accurate readings and all batteries are guaranteed to die at some point in time thus the need to pay constant attention to the battery level The third form of power RF energy harvesting allows for the sensors to be powered for the lifetime of the actual sensor node hardware Energy harvesting is growing rapidly in popularity throughout low power electronics and working on the cutting edge of technology allows the opportunity to produce new and exciting results A downfall with of RF energy harvesting was the need to have the provided RF transmitter with a semi clear pathway to the sensors in order to function properly The cost of the project is another important factor in deciding which product to purchase In general as price increased so did the number features included in the sensor network and with a project geared towards the initial stages of research a reasonably priced but reliable sensor network would fulfill specifications The sensors were required to measure both internal and external metrics including but not limited to light humidity temperature and either voltage or current The sensor network that was chosen provided that plus measuring received signal strength These metrics worked very well with analyzing data and determining when a particular node was malfunctioning The received signal strength in combination with voltage could help determine when a sensor was not getting the appropriate amount of power a
50. re 12 4 2013 Wireless Sensor Network Health Diagnostic 14 ECE 480 Design Team 2 included in the development kit These are the most important pieces of the kit Analyzing the informative sent from the sensor boards to determine network health was the key objective of the project and will be discussed further throughout the report F PICKIT 3 PROGRAMMER DEBUGGER PG164130 This was a programming tool included in the kit for updating code on the Wireless Sensor Boards and the 16 bit XLP Development Board This was not utilized due to time constraints of the project HARDWARE IMPLEMENTATION After gaining familiarity with all of the hardware it was time to configure the wireless sensor network development kit Configuring the wireless network was a vital part of the design process The team undertook many steps to setup the wireless sensor network The first step of configuring the network was to download and install a terminal emulator program on the PC being used There were many options to choose from but the team decided to utilize the terminal emulator recommended by Powercast This terminal emulator was called HyperTerminal and was available to download on Powercast s website The second step was to power and configure the Microchip 16 bit XLP Development Board DM240311 as well as the Microchip MRF24J40 PICtail PlCtail Plus Daughter Board AC164134 1 As you can see in Figure 10 The Microchip MRF24J40 was connected to the
51. reated short term long term and zero analysis These functions were all applied to separate data sets which was determined from the Matlab analysis Overall because of limitations in time and prioritizing tasks in relationship to these algorithms the algorithms were not as robust as they could have been These algorithms however gave a good basic detection of failure of the nodes and in combination with a user analysis using the interface can assure that nodal failure can be determined In conclusion the research into the metrics associated with nodal failure was accomplished which was the main scope of this project 12 4 2013 Wireless Sensor Network Health Diagnostic 31 ECE 480 Design Team 2 APPENDIX 2 LITERATURE AND WEBSITE REFERENCES 1 10 11 12 Wireless Sensor Networks in Space NASA Ames Research Center March 2011 http cenic2011 cenic org program slides cenic 2011 ZigBee sensor net foster pdf Sun SPOT Rev8 Sun SPOT World 2013 http www sunspotworld com products index html Crossbow MTS 400 Crossbow Technology 2007 http www eol ucar edu isf facilities isa internal CrossBow DataSheets MTS400 420 pdf Powercast P2110 Eval 01 User Manual Powercast 2013 http www Powercastco com PDF P2110 EVAL 01 manual pdf Wireless Sensor Network WSN Starter Kit National Instruments 2013 http sine ni com nips cds print p lang en nid 206916 Powercast P2110 Eval 01 User Manual Powercast 2013
52. rge amounts of power in order to propagate data around the D i mE s Toi network Once the network was implemented the Air Force toj Au l es needed to detect anomalies tracked by collecting metrics A A Sensor Node about the network In order to detect anomalies and identify FIGURE 1 WiRELESS MESH NETWORK failing nodes the Air Force used a number of algorithms including side channel analysis and thresholding Side channel analysis consists of using information obtained from the status of the sensors themselves to correlate parameters and determine the interconnected metrics that contribute to sensor failure Training based and thresholding algorithms 12 4 2013 Wireless Sensor Network Health Diagnostic 5 ECE 480 Design Team 2 work by aggregating sensor data over time to make projections based on past data and comparing those projections against incoming data While the Air Force was not able to come to any major conclusions there has also been a lot of work done with sensor network health in industry and other branches of government Upon further research NASA s Ames Research Center also addressed the creation of intelligent wireless sensor networks Intelligence was defined as the capability for supporting dynamic ad hoc self configuring real time sensor networks able to adapt to faults while maintaining measurement accuracy and temporal integrity They developed an ad hoc sensor network that integrated additional sensors for
53. s component was an evaluation board Rev B for the P2110 Powerharvester Receiver Located on the evaluation board there was an SMA connector in order to connect antennas for data transmission and a 10 pin connector for the included Wireless Sensor Boardef Two evaluation boards are available within the kit As mentioned above this WSN utilizes RF energy thus the FIGURE 7 P2110 EVALUATION BOARD evaluation boards are battery free The RF energy captured was converted to DC power via the P2110 Powerharvester Receiver and the DC power was stored in a 5OmF capacitor Upon reaching a voltage of 3 3V the capacitor discharges the energy and transmits the data to the host PATCH AND DIPOLE ANTENNAS Both antenna types are 915 MHz directional antennas The patch antenna has a 120 degree reception pattern whereas the dipole antenna has a 360 degree reception pattern Two of each type of FIGURE 8 PATCH AND DIPOLE ANTENNA antenna are included in the kit one for each evaluation board P2110 EVB E WIRELESS SENSOR BOARD WSN EVAL 01 This board comes fixed with 3 different sensor types temperature humidity and light It also offers an external input as well for additional sensors The wireless sensor board connects into the 10 pin connector on the P2110 Evaluation Board which then sends FIGURE 9 WIRELESS SENSOR BOARD information from the sensors to the Microchip Development Board Two wireless sensor boards a
54. ted Jumpers six nine and ten all had to be connected Jumper six enables the modular expansion header Jumper nine allows the team to measure current consumed by the PIC24F microcontroller but jumper nine interrupts the microcontroller s Vop path so when the team was not measuring the current jumper nine must be connected Jumper ten acts the same way but was used to measure the current consumed by the various board components which does not include the microcontroller ICSP header and the USB interface After all of those jumpers were connected the 16 Bit XLP development board was configured Once the development board was configured it was time to make sure the team s PC could read the data being sent to the development board from the sensor nodes For the team s PC to successfully read in data from the Microchip 16 Bit XLP development board a USB to Serial driver file had to be downloaded and installed Powercast s website had an available USB to Serial driver file on their website to download After the file was downloaded the team connected the development board to the PC and installed the hardware The team now could use the port COM3 to communicate with the wireless sensor network via USB After the COM3 port was enabled and the development board was configured and connected to the PC HyperTerminal was opened and configured to read in the data HyperTerminal had a simple set up with just a few things needed to finish configuring the net
55. tem Technique FAST Diagram is a method to determine the essential functions of a design The FAST Diagram in Figure 2 shows from left to right the primary and secondary functions 12 4 2013 Wireless Sensor Network Health Diagnostic 7 ECE 480 Design Team 2 of a great design for this project Reading from left to right the diagram explains how For example How do you monitor network health one must configure the network and communicate the data being collected by the network The FAST Diagram made creating conceptual designs and determining essential functions much simpler Purchase Sensor Hardware Configure Network Develop Software Implement Define Communication Communication i Pr Pr l Communicate otocol otoco Data Collect Sensor Identify Parameters Parameters FIGURE 2 FAST DIAGRAM CONCEPTUAL DESIGNS The primary focus of this project was developing a diagnostic tool to monitor the health of a network but first network hardware must be selected A number of hardware options were considered for the design including a building an entire sensor network from scratch b obtaining a zigbee sensor network kit and c buying a full sensor network development kit A BUILD ENTIRE SENSOR NETWORK The first design under consideration was to build a sensor network from scratch This included a PCB design attached sensors and a system on chip with a built in ZigBee communication protocol In addition to this ot
56. tilities Help Tal lelelo Delai RI Measurements Markers Status Scales 2 V max 1 3 5344 V 3 5338 V 1 77549 V 1 96803 V 172 2 mV 360 7 mV 3 6113 V 3 5941 V Current Mean Min Max 11 Nov 2013 3 27 PM 12 4 2013 Sample 2 6 JP2 Voltage Screen Capture Wireless Sensor Network Health Diagnostic 44
57. tion of data as well as monitoring the health and status of the nodes themselves This method of data representation especially assists in ease of use in the sensor nodes and also decreases the difficulty in understanding the information The data analysis to create the failure detection is also essential in the determination of failure of nodes in the system This software implemented into the model portion of the controller will allow users to not have to continuously monitor the system but can look at if a failure warning is filed into the view and thus allowing less time to be consumed in graph analysis SOFTWARE IMPLEMENTATION The software portion of the project grew to be the main part of the work associated with correctly analyzing the data input First the data is read from Node 1 oo Node 2 the USB port byte by byte into the controller module Next the data is formatted into packets and sent to the model for storage and data analysis pius icu ee Lug Humidity ShowGrid V Show X labels V The model implements an application programming Xin X max Ymin Ymax Auto Auto Auto Auto interface API which allows the view to access the O Manual 0 O Manual 60 Manual 0 Manual 100 stored data The view polls the model for new data every few seconds and updates its view This design FIGURE 14 THE GRAPHICAL USER INTERFACE pattern allows the data to be graphed in real time The software was written in Python 2
58. work First the team had to setup the connection which included naming the connection and how the team was connecting For this project as mentioned above the team connected to the network via USB port COM3 Next the team had to choose the parameters of the COM3 port With some advice from Powercast the team chose the following settings in Table 4 12 4 2013 Wireless Sensor Network Health Diagnostic 16 ECE 480 Design Team 2 Parameter Bits per second Baud Rate TABLE 4 COM3 PORT SETTINGS Once the COM3 port was set it was time to plug in the power transmitter and build the wireless sensor nodes The power transmitter did not need any configuration It was simple as plugging it into a normal power outlet Next it was time to build the wireless sensor nodes which consisted of the P2110 Evaluation Board P2110 EVB a dipole or patch antenna and the Wireless Sensor Board WSN EVAL 01 For the team s wireless sensor network two nodes were built using the dipole antenna The dipole antenna was used due to it being omni directional Each node also has a unique node ID FIGURE 11 WIRELESS SENSOR BOARD COMPONENTS AND BUILT which was set using DIP switches located on each sensor board Figure 11 shows the wireless sensor node parts as well as a completely built wireless sensor node The last thing that needed to be done to successfully build the nodes was to make sure C5 was connected using jumper 1 C5 was a 50mF cap
59. y David multithreaded the design in order to have responsive view and the application can continue to run if the view crashes or is exited on accident 12 4 2013 Wireless Sensor Network Health Diagnostic 30 ECE 480 Design Team 2 BRAD GARROD FAILURE ANALYSIS ALGORITHM IMPLEMENTATION Brad s technical role consisted of determining the associated metric data correlations to a nodal failure then writing functions that can assist in failure detection This role consisted of work in both Matlab for data analysis and python for the failure detection functions To analyze the data in Matlab failure data produced by Stu was separated and put into a comma separated value file csv to be imported into Matlab Once in Matlab the data was organized into relevant sets which could then be plotted along with averages This data analysis assisted in easier calculations of thresholds in regards to the failures This analysis helped break the failure detections into two groups external and internal sensors These two groups had separate relationships to failure in themselves but also grew in a confidence level when determining a failure Once the thresholds and correlations to failure were analyzed and determined Python functions were constructed to determine when a nodal failure occurred This in combination with the Graphical User Interface constructed by David can make failure detection much easier In python three simple functions were c
60. ys Stu NA 9 16 13 9 17 13 Pre Proposal Mon Fri 5 days Stu David Brad Kelly NA Due 9 16 13 9 20 13 Team Webpage 9 days Mon Sun David NA 12 4 2013 Wireless Sensor Network Health Diagnostic 36 ECE 480 Design Team 2 Started 9 16 13 9 22 13 First Contact Wed Wed Wed 1 day 6 Stu David Brad Kelly w Sponsor 10 2 13 10 2 13 10 2 13 Mon Fri Propsal 15 days 8 Stu David Brad Kelly NA 9 23 13 10 11 13 Configure Sensor Network Fri Fri Fri amp Verify Correct 16 days 8 Stu David Brad Kelly 10 4 13 10 25 13 10 25 13 Sensor Readings Research amp Order Sensor Fri Mon Mon 2 days Brad Development 10 4 13 10 7 13 10 7 13 Kit Verification of Tue Tue Tue Correct Sensor 6 days Kelly Stu 10 8 13 10 15 13 10 15 13 Readings Design Day Mon Fri Team Page 10 days David Kelly NA 9 23 13 10 4 13 Work Oral Proposal Mon Fri Presentation 10 days Brad Stu David Kelly NA 9 23 13 10 4 13 Practice Team Progress Mon Fri 5 days Brad Stu David Kelly NA Report 1 10 28 13 11 1 13 Design Issues Tue Fri 24 days Brad Stu David Kelly NA Paper 9 17 13 10 18 13 12 4 2013 Wireless Sensor Network Health Diagnostic 37 Team Progress Report 2 amp Project Demonstration Identify Simple Network Health 6 days Metrics Develop Graphical User Interface GUI for Configuring Sensor Nodes Confirm Health Diagnostics amp Implement Network Security Functionality If Time Allows Final Reports Design Day 12

Final Report - College of Engineering

Contents

Download Pdf Manuals

Related Search

Related Contents