Home

Construction of a Radio Environment Map of TV White Spaces for

1. 22 MVC 23 2 4 REM based Cognitive FAQ E 23 2 0 REM II US E 24 2 6 Current REM enabled CR standards IEEE 802 22 VVRAN o e 24 Methodology project development ss 25 Ai eebe ia a a iii 37 A ETE 44 Del Labor COSSA i pap eee 44 lo E 44 IG Amortizationsi liy rt ko je a a e al mail a admetia kl 44 UNIVERSITAT POLIT CNICA 22 telecom DE CATALUNYA BARCELONATECH 6 Conclusions and future development 45 Gils esses ters tes desde nae aes nai naine eau 45 6 2 Future Development 45 Ilse a nt a kat ek eat p k e TI 46 APPOENAN CO E 48 An a l kte kane oak ss 48 API documentation 51 Interpolation methodology 53 GIOSS M 54 UNIVERSITAT POLIT CNICA 22 telecom DE CATALUNYA BARCELONATECH List of Figures Figure 1 Gantt EE MT 19 Figure 2 Simplified version of the cognitive cycle 21 Figure 2 Duty cycle statistics 22 Figure 4 Relation between the spectrum manager and other cognitive entities 24 Figure 5 REM architecture 25 Figure 6 sample GUI Screenshot coins ri ae 37 Figure 7 sample GUI heat map visualtzation sen 38 Figure 8 devices operating outside the selected channel 38 Figure 9 Device 2 in the original 500m grid 39 Figure 10 Device 2 in the updated 1km grid 40 Figure 11 app interface 41 Figure 12 app results
2. 3 Mitola III J Cognitive Radio An integrated agent architecture for software radio architecture Diss Ph D Dissertation Royal Institute of Technology KTH 2000 4 Jondral F K Cognitive Radio A Communications Engineering View Wireless Communications IEEE vol 14 no 4 pp 28 33 August 2007 DOI 10 1109 MWC 2007 4300980 5 Lopez Benitez M Umbert A Casadevall F Evaluation of Spectrum Occupancy in Spain for Cognitive Radio Applications Vehicular Technology Conference 2009 VTC Spring 2009 IEEE 69th vol no pp 1 5 26 29 April 2009 DOI 10 1109 VETECS 2009 5073544 6 van de Beek J Riihijarvi J Achtzehn A Mahonen P TV White Space in Europe Mobile Computing IEEE Transactions on vol 11 no 2 pp 178 188 Feb 2012 doi 10 1109 TMC 2011 203 7 Marques P Rodriguez J Delaere S Delahave P Lecroart B Gundlach M Triantafyllopoulou D Moessner K Noguet D Spectrum sharing in the EU and the path towards standardization Future Network and Mobile Summit FutureNetworkSummit 2013 vol no pp 1 9 3 5 July 2013 8 Zander J Rasmussen L K Ri Won Sung Mahonen P Petrova M Jantti R Kronander J On the scalability of cognitive radio assessing the commercial viability of secondary spectrum access Wireless Communications IEEE vol 20 no 2 pp 28 36 April 2013 DOI 10 1109 MWC 2013 6507391 9 Yilmaz H B Tugcu T Alag z F Bayhan S Radio
3. e An interface for visualizing and modifying the REM parameters A convenient way to add and delete simulated devices to the network The communication with the server is achieved using AJAX and Sockets io AJAX is a javascript library used for sending http requests and handling the responses asynchronously using callbacks Heres an example of an AJAX operation In this case an object containing changed parameters is sended to the backend using method POST There is no callback defined to be called when the response arrives document ready function ajax type POST url http localhost 3000 params dataType json contentType application json data JSON stringify payload Here is an example of another AJAX operation In this case a successful response executes a callback that can operate with the contents of the response ajax type GET url http localhost 3000 api success function data Sockets io is a real time connection oriented communication library over http The information exchange is event based To use it one has to open a socket connection using var socket io 35 UNIVERSITAT POLIT CNICA 22 telecom DE CATALUNYA BARCELONATECH Then the next step is setting callbacks for the different events received For example the backend will broadcast to the browser clients a device added event when a device is added through the API To receive
4. 53 UNIVERSITAT POLIT CNICA DE CATALUNYA BARCELONATECH Glossary CR Cognitive Radio REM Radio Environment Map PU Primary User SU Secondary User TV WS TV White Spaces DSP Digital Signal Processor FPGA Field Programmable Gate Array ASIC Application Specific Integrated Circuit REST REpresentational Stateless Transfer API Application Programming Interface GUI Graphical User Interface HTTP HyperText Transfer Protocol HTML HyperText Markup Language JSON JavaScript Object Notation SQL Sequential Query Language EIRP Equivalent Isotropic Radiated Power UHF Ultra High Frequency RMSE Root Mean Square Deviation UTM Universal Traverse Mercator WGS84 World Geodetic System 84 De 54 DE CATALUNYA BARCELONATECH gt telecom UNIVERSITAT POLITECNICA NU BCN 55
5. The methods exposed by the REM module are the following e addDevice lat Ing bandwidth callback err data e Tries to insert a new device in the network using the insertDevice lat Ing bandwidth method from the spectrum broker module If the insertion is successful stores the new device into the devices table of the database and passes the device to the provided callback Callback signature is callback err data deleteDevice device id callback err data Tries to delete the device that has the dame device id property than the device id argument If successful executes the callback deleteDevices callback err data Wipes completely the devices table then executes the callback getDevices callback err data Obtain all the devices stored in the devices table Retuns a vector with all the stored device objects and passes it to the callback updateDevices callback err data Called when a REM parammeter is changed retrieves the devices table and then then recalculates all devices properties and updates them in the database When finished passes the updated devices to the callback 31 UNIVERSITAT POLIT CNICA 12 telecom DE CATALUNYA BARCELONATECH spectrum broker js gt Is the module that assigns channels and tx_power to the new devices submitted to the REM It uses the REM data to allocate the new devices and minimise interference for the primary users This functionality was not required at the st
6. The purpose of this section is to describe the functionality of the GUI for the REM Latter a step by step guide explaining how to perform the various use cases for the GUI will be provided The content of the GUI is the following aaa raue f FEU ET EL de Castellet 5 d 14 FAE YOU Y A ay Primary User Power Level Channel 21 v 1 kilometer L ci f de P as ax de Ron ana TE ona g ga v d Si f Carnets Minimum Primary MLETS User CIR 7 ats 73 Corr d Avall Map Cor Ve an Castellar Canovelles tanya 125 ki era delValles Sentmenat li x A m 7 ranollers econdary User gt Y it T me CIR BF ef d DOE s ue de val 16 A e gt ER ANM errassa i ma 1 at Alpha Viladecavalls pg Device Tooltip Poliny EE 4 A C e2RAdal 0 E lets Max Tx power ge LS a Y n lav 25 Ullastrell Des Fonts Device idt4 L 1 fas Channels CHANNEL21 Base Station TxPower 23 204363772379537dBm ZAN Hei 4 lius 1133 El Mas Coll l md Height d Coverage radius 1133 7 meters LE Blocks available 10 Alella Mobile devices amore Castellbisbal bi E From 21 to 21 g Height N a E I mos From 25 to 25 IA Ga t 15 LLA From 28 to 28 Submitl es 4 From 30 to 30 om From 32 to 32 Select a device id Wa From 38 to 42 to remove it from Weg From 45 to 46 th
7. isis EE ee eer siwi sauce masia o id p RR nb mirat A ai 42 Figure 13 results of the synthetic benchmark teen ee eneees 42 Figure 14 Alarm Probability versus sample atze ee eee eter eee e nen 43 Figure 15 Primary Not Detected versus sample size 43 Figure 16 Elements of the GU 48 Figure 17 channel selection 49 Figure 18 grid selection emen m ne enne nnne 49 Figure 20 grid style selection 50 Figure 21 display selection na nn nn 50 Figure 22 REM oarameterg conca se se se see ehe nennen nnne 50 Figure 23 delete devices 2 cece cece cece eee eee ss 50 Figure 24 insert devices 50 UNIVERSITAT POLIT CNICA 22 telecom DE CATALUNYA BARCELONATECH List of Tables Table 1 Work package 1 15 Table 2 Work package 2 csse nenne enhn RR Re nans 15 Table 3 Work package 3 16 Table 4 Work package A AAR NNN ENEE ENNEN 16 Table 5 Work package 5 17 Table 6 Work package pn 18 Table 7 Work package ANEREN ENNER NENNEN ENEE ERR ak kaa senksan ane 18 Table 8 Labour costs ugesi tas eee eee tke tas cee pe E dada 44 Table 9 Software BO Ek vi titak kt atq ME Puta tz aman 44 Table 10 Deprecaton cece cece eee e cece eect ee eae ease eee eeaeeeaeeaeeeeaees 44 10 UNIVERSITAT POLIT CNICA 12 telecom DE CATALUNYA BARCELONATECH 1 Introduction 1 1 Purpose The transition from analog TV to digital TV has left a great deal of t
8. 1ng2 08775 Bandwidth MH2 8 Insert the device Pwr 77 84146 Dades del mapa 2015 Google basado en BCN IGN Espa a Termes i condiciodiBninformeu d un erer al mapa Y ve lacarisses Palau solit P A i i Plegamans Lli de Vall La Roca RS y del Vall s Parets 3 eat Vall Vilanova La Conreria Sant del Vall s Mateu C llecs o Valltomanes g Li E errassa SE Poliny Sa abadell c N 150 Ullastrell c an La Florida agosta Viladecavall B Mollet 4 del Valle lartorelles Oe El Mas Coll L Secondary User CR 16 We Alpha 1 EE Max Tx power 25 Base Station Height 10 Mobile devices Height 1 De Submit Ke s Select a device A at S l 7 Torrelles de 1d to remove it Llobregat der from the REM 7 santClimenti Tf left void de Llobregat removes all devices Cabrils Premia d Dalt de Mar Coffer Channels CHANNEL21 Llobree TxPower 7 905348642830646dBm Coverage radius 111 7 meters Vallirana dels Bi Viladecans Gava Select a Device id for deletetion Delete Castelldef ls p Figure 8 devices operating outside the selected channel 4 Load Grid Load Samples Display Primary User Power Level Channel 39 v 0 5 kilometers v Argento a Montnegre iel Corredor de M nt a E 2 Sant Andreu enys de e fide Llavanel gt Fro
9. 32 to 32 From 37 to 38 From 40 to 43 From 46 to 46 From 49 to 60 From 62 to 63 From 65 to 66 Bo Base Station Mobile devices Height 1 I Submit at tosa Select a device id to remove it from the REM Ifleft void removes all devices Select a Device id for deletetion 1 5 Delete C in Coll lat 41 43372 1ng2 11246 Figure 6 sample GUI screenshot The primary user data is represented as a grid containing the primary user power readings for each channel The grid is painted red if the power reading for Primary Users is above the detection threshold in this case 85 dBm and blue otherwise indicating that the this channel in this cell can be accessed The visualized channel can be changed using the toolbar located at the upper edge of the GUI This tool allows the GUI user to check the geographical availability of each TV WS The Primary user readings can also be observed as a heat map if the user click the checkbox Display Primary User Power Levels The colour styling of the grid will change into a gradient representing the detected PU power level An example of this view is shown at figure 7 The markers in figure 6 and figure 7 represent cognitive devices introduced in the network The markers will be shown green with an approximate coverage circle if the devices are using the selected channel In case they are not using it the markers will be shown grey figure 8 If
10. 500 1000 56 67 96 1500 2000 12 08 96 5 89 96 Figure 3 Dutv cvcle statistics Casadevall F Umbert A REM based real time testbed A proof of concept on the benefits of using REM for improving radio resource management capabilities Telecommunications ICT 2013 20th International Conference on vol no pp 1 5 6 8 Mav 2013 Cognitive radio has been proposed for optimizing the usage of the spectral blanks 1 3 do to it s awareness features and adaptabilitv that enable a cognitive device to detect when a certain part of the spectrum is not occupied and change its transmission or reception characteristics to make use of this unused channel or spectrum blank becoming this device a Secondary User for that spectrum Moreover a cognitive device would be capable of detecting if a licenced user or primary user is using the channel and move its activity to another spectral blank minimizing the interference endured by primary users 22 UNIVERSITAT POLIT CNICA D telecom DE CATALUNYA BARCELONATECH 2 3 TV WS One case of underutilized spectrum is the case of TV white spaces TV WS TV WS are the outcome of the transition from analogical TV systems to the more spectral efficient DVB T standard A study of the benefits advantages in terms of spectrum usage for Germany and other European countries can be found at 6 The result of the transition is that certain parts of the allocated TV spectrum was left unused with
11. Euros Table 9 Software costs 5 3 Amortizations The development of this project only used a single laptop computer for development tasks Laptop for the developer Purchase cost 900 euros Amortization period 4 years residual value 25 Used for 143 days Depreciation 88 15 euros Table 10 Depreciation 44 UNIVERSITAT POLIT CNICA 12 telecom DE CATALUNYA BARCELONATECH 6 Conclusions and future development 6 1 Summary The technical result of this project is clear A REM for cognitive utilization of TV WS can be implemented through generic technologies in a cost effective manner using mostly open source tools On a personal level completing this thesis has helped me to develop my skills in two key areas The first of them is bibliographical research which has been key to gather sufficient information in a timely manner in order to develop this thesis The other area in which my knowledge has improved is web development At the start of this thesis my knowledge of web technologies was superficial at most but during the development of the project learned about front end development back end development and database management all done in a programming language was not very proficient at the start of the project Javascript 6 2 Future Development The product of this thesis can be expanded in the following ways 1 Extending the functionality of the
12. begin to implement REM based cognitive networks Even though the concept of cognitive radio and in particular the use of a REM for providing Spectrum Awareness to the cognitive devices is something general the TV band ranging from 470MHz to 790 MHz in Spain is specially useful for implementing the cognitive radio concept The opportunity for implementing cognitive radio technology lies in the transition between analog and Terrestrial Digital Television Digital channels occupy 4 times less band that analog ones and also the need to keep certain channels empty to avoid interference between TV repeaters is gone since Terrestrial Digital TV is a synchronous network These factors have vacated most of the space between 470 790 MHz the unused channels constitute TV White Spaces TV WS This white spaces would be useful in a series of wireless applications and services such as Wireless Sensor Networks WSN indoor WLANs or rural broadband access all of which can be deployed under the concept of cognitive radio The main challenge for TV WS utilization is that they are not geographically uniform even within a certain region This fact is due that some repeaters transmit local channels and the coverage of these repeaters varies In summary the usage of TV white spaces would be very beneficial from the perspective of technological and economical development of these new services but requires the implementation of new techniques such as cogni
13. environment map as enabler for practical cognitive radio networks Communications Magazine IEEE vol 51 no 12 pp 162 169 December 2013 DOI 10 1109 MCOM 2013 6685772 10 Farnham T REM based approach for hidden node detection and avoidance in cognitive radio networks Global Communications Conference GLOBECOM 2012 IEEE vol no pp 1391 1397 3 7 Dec 2012 DOI 10 1109 GLOCOM 2012 6503308 11 Jian Wu Zhang Qi Zhao Jing yuan Zou The IEEE802 22 WRAN System Based on Radio Environment Map REM Education Technology and Computer Science 2009 ETCS 09 First International Workshop on vol 1 no pp 98 101 7 8 March 2009 DOI 10 1109 ETCS 2009 31 12 Youping Zhao Reed J H Shiwen Mao Bae Kyung K Overhead Analysis for Radio Environment Mapenabled Cognitive Radio Networks Networking Technologies for Software Defined Radio Networks 2006 SDR 06 1st IEEE 46 6 13 UNIVERSITAT POLIT CNICA 22 telecom DE CATALUNYA BARCELONATECH Workshop on vol no pp 18 25 25 25 Sept 2006 DOI 10 1109 SDR 2006 4286322 Stevenson C Chouinard G Zhongding Lei Wendong Hu Shellhammer S J Caldwell W IEEE 802 22 The first cognitive radio wireless regional area network standard Communications Magazine IEEE vol 47 no 1 pp 130 138 January 2009 DOI 10 1109 MCOM 2009 4752688 47 UNIVERSITAT POLIT CNICA D telecom DE CATALUNYA BARCELONATECH Appendices GUI User s manual
14. formatting the information from the grid to be used in the web application blocks Concretelv what the program does is reading the sample information from a certain folder and then outputting a set of CSV files accompanied with VRT files that are used to create the GeoJsons used in the web application Procedure Firstly the basis for this project were a series of data from the TV spectrum occupancy taken in various points inside Barcelona metropolitan area in two prior degree thesis The data was organized in data points each of them referenced five different files that contained the GPS location the date in which the sample was taken a vector containing spectrometer power readings a vector with the frequencv axis for the readings and finallv a vector containing the power levels for each TV channel 28 UNIVERSITAT POLIT CNICA 1 telecom DE CATALUNYA BARCELONATECH For the construction of the REM the only relevant fields were the GPS position and the TV channel levels The first step was to create a script that from a folder that contained the various readings extracted all relevant information and pulled it into a Matlab structure that contained the point name the position and the channel readings data function loadData Now the next step is using this data to interpolate a grid that contains the power readings for each channel but first the GPS data must be projected in order to translate the spheric
15. gran medida la eficiencia espectral de estas emisiones dejando gran parte del espectro asignado a usos televisivos sin utilizar Este espectro podr a utilizarse de forma oportunista por parte de usuarios sin licencia usuarios cognitivos para establecer conexiones entre ellos a trav s de una red cognitiva basada en REM Para la construcci n del REM se han usado muestras de campo tomadas en el rea metropolitana de Barcelona que han sido posteriormente procesadas con el objetivo de adaptarlas a la estructura REM Se puede acceder a la informaci n contenida en el REM a trav s de una API REST que adem s permite administrarlo La GUI ha sido implementada a trav s de una p gina web que permite visualizar y administrar el REM Finalmente se ha a adido una interfaz que permite utilizar el REM para administrar dispositivos cognitivos simulados con el objetivo de demostrar las capacidades de esta tecnolog a DE CATALUNYA BARCELONATECH e UNIVERSITAT POLIT CNICA K em Per la Laura Sense tu res tindria sentit UNIVERSITAT POLIT CNICA 22 telecom DE CATALUNYA BARCELONATECH Acknowledgements Firstly want to thank my supervisor Ferran Casadevall for his thoughtful advice and the editorial assistance in the development of this thesis also want to thank Sergio Garcia for his advice and helping me put my ideas in order for the web development part of the project UNIVERSITAT POLIT CNICA 2 telecom DE CATALUNY
16. la Barca al Can Coll ra de Fontpineda Pallej regat WI ta Rosa de C x Device id 2 Channels CHANNEL21 TxPower 13 399382564341 721dBm Cervell Coverage radius 80 2 meters PP FUI ana Can Pi Figure 9 Device 2 in the original 500m grid 39 DE CATALUNYA BARCELONATECH e UNIVERSITAT POLIT CNICA D telecom Load Grid Yi Load Samples Display Primary User Power Level Channel 21 v 1 kilometer Jan Coll ade Fontpineda pates i gat ya Pali si le Cerv Rosa Device id 2 Channels CHANNEL23 TxPower 25dBm Cervell Coverage radius 778 1 meters l Can Bik To0 G O TT f Torrelles de Figure 10 Device 2 in the updated 1km grid It is important to note that all changes in the GUI are synchronized with the REM manager in real time and also between different GUls connected to the same REM manager The effect of this feature is that if there were more than one user connected via web the changes to the REM done in one of the GUIs would be updated to the other GUI immediatelv On the particular case of a grid resolution change reintroducing the devices in the REM is especiallv important since the geographical availabilitv of the channel will actuallv change In zones where the signal level is close to the threshold a finer granularitv on the interpolation allows to approximate with more precision the activitv of the primarv users It is important to
17. per a usuaris sense llicencia usuaris cognitius per a establir comunicacions entre ells mitjancant una xarxa cognitiva basada en REM Per a la construcci del REM s han utilitzat mesures de camp preses a l rea metropolitana de Barcelona que han estat processades posteriorment per adaptar se a l estructura REM La informaci continguda al REM es pot consultar a trav s de una API REST que a m s a m s es pot utilitzar per administrar el REM La interf cie de consulta del REM finalment ha estat construida mitjan ant una p gina web que permet visualitzar i administrar el REM A m s a m s s ha creat una interf cie que permet utilitzar el REM per administrar dispositius cognitius simulats amb l objectiu de demostrar les capacitats de la tecnologia BARCELONATECH UNIVERSITAT POLIT CNICA 22 telecom DE CATALUNYA Resumen Esta tesis describe la construcci n de un REM Radio Enviroment Map de TV WS TV White Spaces usando medidas de canal tomadas en la rea metropolitana de Barcelona adem s de una GUI para facilitar su consulta Un REM es una pieza clave para la implementaci n de una red cognitiva ya que proporciona a los dispositivos informaci n del medio radio en el que se encuentran permiti ndoles que canales utilizar y cuando utilizaros para operar de forma ptima Los TV WS son las zonas espectrales asignadas a televisi n pero que no se usan para ese cometido desde la transici n al est ndar DVB T que aument en
18. performs excellently and has a variety of algorithms available Using Matlab for this block eased the burden of implementing an interpolator in some compiled language to perform the REM construction tasks Also it made possible to develop benchmark for the various scattered interpolation algorithms available in order to pick the optimal one for the REM construction REM manager back end The back end contains the server logic for the REM manager Its tasks include managing cognitive devices keeping the database up to date and also updating and serving the webpage that acts as the GUI This block is implemented using Node js Node js is a Javascript framework for developing the server part of web applications It also features an extensive library of modules which allow for implementing geospatial processing applications such as this project At the start of the project this part of the REM manager was to be implemented using Google Maps Engine a Google service for managing geospatial data but unexpectedly Google announced that Google Maps Engine would be discontinued and a dedicated back end software module had to be created to comply with the project requirements Database The storage has to perform with two functions Store the interpolated grid data that contains the power readings of the primary users at various locations and maintaining an up to date database of the cognitive devices The first type that references prima
19. technical problem if the current growth of mobile communications has to be maintained As it is mentioned before the majority of the useful spectrum is already allocated this means that is reserved for a certain use and that has a specific licensed user Only the licensed users and no one else are allowed to make use of their allocated spectrum because if someone else makes use of it the licensed user could be interfered However that a certain part of the spectrum is allocated does not mean that it s owner is using it all the time or in every geographic location The lack of usage of certain radio channels means that there is room for improvement of the usage of the channel One 11 UNIVERSITAT POLIT CNICA 1 telecom DE CATALUNYA BARCELONATECH could think that a third party or secondary user could make use of allocated spectral bands when or where their rightful owner or primary user is not making use of them This constitutes the opportunistic usage of the spectrum Opportunistic usage of spectrum is one of the tools that will allow the capacity of wireless networks to grow and face the expected increase of wireless traffic as it maximizes the usage of the available spectrum The technology that allows for the dynamic usage of spectrum to happen is Cognitive Radio which is listed as one of the technologies that will allow 5G wireless communications to be developed A cognitive radio is a terminal that is aware of its immediate
20. A BARCELONATECH Revision history and approval record poi Pr C RN DOCUMENT DISTRIBUTION LIST Mart Cervi Caball marti cerviaQalu upc edu Ferran Casadevall ferrancQtsc upc edu Written by Reviewed and approved by Date 19 1 2015 Date 5 1 2015 Name Marti Cervia Name Ferran Casadevall Position Project Author Position Project Supervisor UNIVERSITAT POLIT CNICA 22 telecom DE CATALUNYA BARCELONATECH Table of contents The table of contents must be detailed Each chapter and main section in the thesis must be listed in the Table of Contents and each must be given a page number for the location of a particular text ASMA D 1 FOS W NI kt lee 2 EE 3 ene Ve E CET RA 5 Revision history and approval record 6 TABS DECO TEE a l D tn eat dent 7 BO ONA ti ta TTC TP 9 ES 6 Ol PADIS qe T 10 Vs INtrOCLICHOM M 11 VA PUNPOSOE ix b ai tas lauds 11 CN e de te Une ME 11 1 3 Requirements and Objectives ucraniana ree 13 1 4 Previous WOPKS a ee een 13 l o Work plain ita les di dias della 15 1 6 Gantt RE KE 19 1 7 Deviations and incidences nn nn nn HH 20 2 State of the art of the technology used or applied in this thesis 21 2 1 Cognitive Radio an introduction ss 21 2 2 Cognitive Radio use cases
21. AL open source utility is used to convert from the CSV files to GeoJson using its VRT descriptors An external program was selected because a suitable implementation in Matlab was not found and the actual conversion to GeoJson can take a long amount of development time to implement GDAL is implemented in C and actually faster than any Matlab script so using this program is not a performance concern The program is called in the same Matlab script using the system command and then the newly generated files are moved to the public folder of the web app to be served by the back end 4 REM Manager back end This software was written as a Node js based web application and consists of two sub sections One is a normal web server that serves the webpage that acts as the GUI and other static content such as the GeoJsons that contain the primary user information The other part consists in a REST API that is used to access REM functions such as adding devices getting or deleting cognitive devices and administrating the REM working parameters The back end is also tasked with updating the GUI information at real time using sockets io Implementation The back end is implemented using the express framework that is a simple and lightweight framework for creating web apps for node js Express was chosen because it automates the most tedious processes in building a web app or a REST API All the code is divided in modules In Javascript modules
22. DE CATALUNYA BARCELONATECH example alpha 2 The response JSON is a vector containing the updated version of the REM recalculated with the new parameters or an error if invalid information is passed Interpolation methodology Choosing a suitable interpolation methodology is an important factor in the construction of a REM An adequate interpolation methodology will provide the REM with accurate estimations of the state of the channel in most geographical areas enabling the cognitive network to make the best possible use of the channel without interfering its owners The mathematical problem that the interpolator has to face is reconstructing a complete grid for the Radio Environment Map from a set of scattered non uniform distributed values In this case the aim to reconstruct or interpolate the values inside the region or convex hull of the dataset The solution must form a surface where two dimensions correspond to the geographical positions an the other to the field intensity for each channel To solve this problem the Matlab class Scatteredinterpolator three different algorithms that will be briefly explained below 1 Nearest Neighbour This is the simplest method where an interpolated point is given the same value as the nearest data sample using a certain definition of distance It s the fastest from the three methods available but also happens to be the least accurate 2 Natural Neighbour This
23. IVERSITAT POLIT CNICA 22 telecom DE CATALUNYA BARCELONATECH The results in Figure 13 show the relevant figures of merit of the diverse interpolation algorithms versus the number of data samples used in the interpolation The lower left corner figure shows the original synthetic power distribution in dBw The RSME figure shows that for a moderate amount of samples higher than 100 linear interpolation and natural neighbour interpolation perform almost identically with a marginal advantage to linear interpolation The figure for False Alarm Probability show the same tendency benefiting nearest neighbour interpolation for a small amount of samples detail at figure 14 and linear and natural neighbour interpolation performing almost equally for large amounts of samples Finally in the probability of Primary not detected graph it can be seen that Natural neighbour interpolation performs better than the other algorithms especially for a limited amount of samples detail at figure 15 Detecting the primary user correctly is the key parameter for a cognitive radio as cognitive radios operate on the premise of minimizing interference with Primary Users From the results of this test and taking into account the prevalence of Primary User detection in the operation of cognitive radio the algorithm that was chosen to construct the REM data was Natural Neighbour Interpolation as it performed better or equally detecting the prima
24. POLIT CNICA 22 telecom DE CATALUNYA BARCELONATECH Qui ia 1 04 results activitv The device has been created successfully with the following parameters Device id 6 This demo shows that the requirements to use cognitive access through a REM are very limited a device only needs a way to access an approximate location and an http client to use the prototype REM described in this thesis Lat 41 38816166666667 Lng 2 11212 Max Tx power is set at 25 0 dBm Approximate coverage is 1257 1617416420315 meters Assigned channels CHANNEL21 Figure 12 App results 4 4 Interpolation algorithm validation The last objective of the thesis was the validation of an algorithm for REM construction for that task the synthetic benchmark described at section 3 2 was used The benchmark was iterated several times for greater statistical significance and also tested with different data sample sizes linear tg natural neighbor r nearest neighbor b linear tg natural neighbor r nearest neighbor b jS 70 10 607 SN 50 N w 0 X E 40 a A E M ii E 5 ND 5 50 a Yen MN 10 MAC TS w 20 e sons LL NA u 15 A fa 10 20 0 0 100 200 300 400 500 0 100 200 300 400 500 Samples Samples linear g natural neighbor ir nearest neighborib 207 1 o PoweridB Primary Not Decected Prob 26 o 0 100 200 300 400 500 Samples Figure 13 results of the synthetic benchmark 42 UN
25. UNIVERSITAT POLIT CNICA DE CATALUNYA BARCELONATECH Construction of a Radio Environment Map of TV White Spaces for Barcelona Metropolitan Area A Degree s Thesis Submitted to the Faculty of the Escola Tecnica d Enginyeria de Telecomunicaci de Barcelona Universitat Politecnica de Catalunya by Marti Cervi Caball In partial fulfilment of the requirements for the degree in Science and telecommunication Technologies ENGINEERING Advisor Ferran Casadevall Barcelona Februarv 2015 UNIVERSITAT POLIT CNICA 22 telecom DE CATALUNYA BARCELONATECH Abstract This thesis covers the construction of a REM Radio Environment Map for TV WS TV white spaces for the metropolitan area of Barcelona and a GUI for accessing it A REM is a key piece in the implementation of a cognitive network since proportions environmental information to cognitive devices allowing them to decide what channels they have to use and when they have to use them in order to perform optimally The term TV WS refers to the spectral zones that are assigned to TV broadcasting but they are not used for that purpose since the transition to the DVB T standard This new standard allowed for better spectral efficiency in TV broadcasting fact that let to the underutilization of the TV assigned spectrum This underutilized spectrum could be used opportunistically for unlicensed users cognitive users to establish a REM backed cognitive network Field meas
26. al coordinates to a set of Cartesian points Projection is a necessary step because interpolation cannot be done in angular coordinates since Euclid distance needed for the correctness of the results of the interpolation algorithm is not applicable in wgs84 coordinates GPS as they represent points in a sphere For that purpose the UTM coordinate system was chosen since it has low distortion for projections of zones inside the same UTM grid zone and the metropolitan area of Barcelona is located entirelv in UTM zone 31T A Matlab function that transformed the data points GPS coordinates into UTM coordinates was developed for that the Geodetic Transformations Toolbox developed bv Peter Wasmeier and available in the Matlab file exchange was used Function coordinates convert GPS values utm gps coordinates After that the next task was to interpolate the grid In particular a function that given the data positions in UTM coordinate svstem and the channel readings generated an interpolated grid for each of the TV channels was developed function WGS polygonal matrix WGS Samples matrix function process all data data grid size This function takes as input parameter the separation between grid samples in km The grid size is chosen so the resulting square contains inside the convex hull of the data points Next the interpolated data must be formatted before being converted to GeoJson In particular each cell of the gri
27. ama y Martorelles L x PETE LI I FI a Florida r Cant Cant da Figure 17 channel selection Change the REM grid size Select a new grid size from available sizes at the dropdown menu highlighted in fig 16 4 Load Grid Load Samples Display Primary User Power Level Channel 21 v mem Dy 3 x APTE Y Pay 7 La Florida Cant Cant da Figure 18 grid selection 49 6 regatie DE CATALUNYA BARCELONATECH UNIVERSITAT POLITECNICA K woo Change grid styling Check the highlighted checkbox in Figure 20 to switch between the grid colours representing cognitive availability or primary user power levels Load Grid Load SamplescC Display Primary User Power Lex t Channel 21 y 1 kilometer v Martorelles x FATTE I Y PS La Florida NW Cant Cant da Figure 20 grid style selection Display original samples toggle grid visibility Uncheck the load grid checkbox to hide the grid overlay or toggle load samples to overlay markers that show where the original channel measures used to construct the REM were taken Load Grid 4 Load Sample gt Display Primary User Power Level Channel 21 v 1 kilometer y Martorelles LI PATTES LI ra La Florida NW Cant Cant da Figure 21 display selection Change the REM parameters crue User CIR Minimum Primary d User CIR To change the REM operational parameters alter the Af c desired fields and click the submit bu
28. are parts of the program that reside in different source files The advantage of this approach is it permits to compartmentalize the code in more maintainable units than using only one source file Node js usually performs asynchronic operations such as reading or writing intro files or databases In order to keep the execution flow the asvnchronic methods admit a callback function as a parameter which is called when the method finishes executing This characteristic marks the stvle of the code that normallv features callbacks in each operation The functionalitv of the backend is split into the following modules app js gt initializes the http server and binds the routes also sets up the generic error handlers Loaded at app start 30 UNIVERSITAT POLIT CNICA D telecom DE CATALUNYA BARCELONATECH REM js gt represents the model for the REM Contains all the methods used to access and modify the REM database Its methods are called from the API to modify the contents of the REM REM js manages an sqlite3 database with a single table called devices where every row corresponds to a device The devices table has the following columns 1 device_id a numeric identifier for the device lat latitude in decimal degrees Ing longitude in decimal degrees tx power the maximum EIRP for the device in dBm coverage radius an approximated coverage radius ous on channels the channel indexes assigned to the device
29. art of the project and is implemented in a basic way however it is a fine way to showcase the REM capabilities The methods exposed by spectrum broker are listed below e insertDevice lat Ing bandwidth callback err data e This method is the responsible of adding new devices to the network The algorithm is actually quite simple The pseudocode is found below 1 The first thing it does is to locate what cell in the grid contains the designated point for the new device If the device sits outside the grid returns an error 2 Next checks if the bandwidth selected is allocable at the selected position Normally a series of available channel blocks will result If no suitable channel block is found an error is returned 3 The next step is selecting the channel block it calls the selectChannels method of the same module and that will output the assigned channel block 4 Once the channel block is chosen the max transmission power for the device is selected lt is obtained from the getMaxTxPow method 5 The final step is to calculate the approximated coverage radius for the device for this task the method findCoverage is used 6 Once all the parameters are set a new Device object with the obtained transmission power coverage radius and selected channels is passed to the callback function and the execution ends e selectChannels num channels available blocks e This function just selects the lowest block of num chan
30. be synchronized to the GUI 2 Interpolation selection and implementation Once Matlab was selected for this task several methods for performing scattered interpolation were available through the Scatteredinterpolant class This class implements various methods of scattered interpolation natively The methods available are the following e Nearest Neighbour interpolation e Natural Neighbour interpolation e Linear Interpolation To test the relevant performance metrics of each interpolation method in a REM a synthetic benchmark was developed The benchmark In order to evaluate each interpolation algorithm a synthetic benchmark was proposed based in the one conducted on 8 In it a synthetic power distribution simulating 3 radio sources located at different positions emitting in the same channel and with propagation loss based on flat earth model is used In the experiment a number of random scattered data samples are extracted from the synthetic distribution and fed to each interpolator Then the interpolator has to reconstruct a grid from the sample data and the resulting grid is compared against the original synthetic distribution 27 UNIVERSITAT POLIT CNICA 2 telecom DE CATALUNYA BARCELONATECH In order to interpret correctly the results of the benchmark it must be remembered that the purpose of the REM is to detect the primary users in order to not interfere with them and to detect any spectral holes availa
31. ble In order to decide if a primary user is transmitting or not a detection threshold is used If the power reading is above the threshold the channel is flagged as used or if below the channel is flagged as empty Then in order to correctly evaluate the benchmark the following three metrics must be evaluated As the interpolated grid differs from the original there can be grid points where the interpolated power level is above the detection threshold and the real power level is below it That event is called a false alarm In that case a site that could be used for transmission is wasted because the primarv user is falselv detected Another more dangerous event is not detecting the primarv user Complimentarv as before some point of the grid could be below the threshold being above the threshold in the real power distribution This possibilitv is reallv dangerous because it leads to primarv user is being interfered Finallv other useful figure is the RMSE between the svnthetic distribution and the interpolated one because it gives a degree of closeness between the two figures For evaluating each interpolation method the three parameters described above as figures of merit were used and the results are reported in section 4 4 of this document Implementation This functional block is tasked with reading the data samples from different files creating a grid resulting from the interpolation of the initial scattered samples and
32. celona Metropolitan Area Major constituent Creating a GUI software Sheet 4 of 7 Short description Elaborate the requirements for the GUI program and select an adequate technology for the implementation Then implement it and validate the software produced Planned start date 20 10 2014 Real start date 27 10 2014 Planned end date 19 11 2014 Real end date 1 12 2014 16 UNIVERSITAT POLIT CNICA DE CATALUNYA BARCELONATECH D sen Start event End event Internal task T1 Requirements Deliverables Dates GUI software 1 12 2014 Internal task T2 Technology assessment Internal task T3 Implementation Internal task T2 Validation Table 4 Work package 4 Project Construction of a Radio Environment Map REM of TV WP ref 5 White Spaces for Barcelona Metropolitan Area Major constituent Integration and validation software Sheet 5 of 7 Short description Integrate all the nodes of the REM and validate its correct operation Planned start date 20 11 2014 Real start date 3 11 2014 Planned end date 19 12 2014 Real end date 18 1 2015 Start event End event Internal task T1 Integration of the nodes and interconnections Deliverables Dates Integrated 19 12 2015 Internal task T2 Validation of the REM PCM prototype Table 5 Work package 5 17 UNIVERSITAT POLIT CNICA DE CATALUNYA BARCELONATECH 6 D e
33. chitecture 2 Select and validate an interpolation algorithm to be used in the REM manager in order to approximate the REM data at areas not covered by the measurements 3 Create a GUI tool to easily access the data stored in the REM 4 Integrate all the REM components into a single system 1 4 Previous works This project uses the measurements taken in the following projects Pedro Delicado Alc ntara medidas para la construcci n de un REM Radio Environment Map de TV WS en la mitad norte del ria Metropolitana de Barcelona 17 9 2014 Gustavo Herraiz Montes medidas para la construcci n de un REM Radio Environment Map de TV WS en la mitad sur del ria Metropolitana de Barcelona to be presented For the development of this project the following 3 party software tools have been considered Matlab for the data processing and interpolation e GDAL for converting the interpolated data to GeoJsons for the REM e node js for the development of the back end REM software And the following libraries and frameworks 13 UNIVERSITAT POLIT CNICA 22 telecom DE CATALUNYA BARCELONATECH e geojson utils a set of utilities for working with geojson objects in javascript e Express js a flexible web application development framwork e Node sqlite3 a library that provides bindings for using an sqlite database from node js e Google Maps Javascript API v3 a framework to manipulate and create map
34. d had to be expressed as a polygon for correct representation at the GUI The polygon was generated in UTM coordinates and then translated al to elliptic using the WGS84 ellipse for usage in the GUI That polygon corresponds to the first column of the output CSV file and containing the geometry of the object to represent The interpolated values of every channel are then concatenated to this string and form the rest of columns of the CSV file that is temporarily stored as a cell of strings where each string corresponds to a CSV row and a grid point The obtained interpolated data has also to be processed First the grid points that fall outside of the convex hull delimited by the data cannot be interpolated because they are too isolated thus the interpolator outputs NaN or not defined for the interpolated 29 UNIVERSITAT POLIT CNICA 2 telecom DE CATALUNYA BARCELONATECH values in this points The CSV rows containing NaN values are then pruned from the cell array Next a function to save the results to files was created and it is used to print the cell array containing the CSV rows values to a CSV file and generate a companion VRT file A VRT file is used as a description for a CSV file It consists of XML encoded data that defines the contents of column of the CSV file The VRT descriptor file is needed to convert the CSV file to GeoJson function save_all_data sample_data geometry_data Finally after all the process the GD
35. e REM Ifleft T 4 From 49 to 60 Seid Fames all gt Secondary Users From 62 to 63 devices From 65 to 66 Selecta Device id Se for deletetion lt Torrelles de um Llobregat qe EI Channel Availability Grid gt j Delete tegen Ge LT Cursor Information ke 7 a A Af IZ Olivella po REM management Controls ya 2 lat 41 42316 Ing 16980 Bandwidth MHz 8 Insert the devicel Pwr 95 dBm Dad del mapa 2015 Google basado en BON IGN Espa a Temesi condicions inforneu d un enoral mapa _ Zeck Figure 16 Elements of the GUI The different sections are explained below e Map Controls Contains the elements that control which information is displayed on the map REM management controls Allow the user to change the parameters that the REM uses to insert cognitive devices and to delete devices from the REM Selected Position Cursor Allows the user to select one location from the map This position controls the information shown at the Available Channels and Cursor Information elements 48 UNIVERSITAT POLIT CNICA 22 telecom DE CATALUNYA BARCELONATECH e Available Channels List the UHF channel blocks available for cognitive use at the cursor location e Cursor Information Displays the primary user power level at the cursor s position and the position of the cursor Also lets the user specify a bandwidth for creating a new simulated cognitive device at
36. e components of the REM implementation the nodes involved and the interactions between them Planned start date 6 10 2014 Real start date 9 10 2014 Planned end date 15 10 2014 Real end date 15 10 2014 Start event End event Internal task T1 Definition of the nodes involved Internal task T2 Definition of the interconnections between nodes Dates 15 10 2014 Deliverables REM architecture interim working document Table 2 Work package 2 15 UNIVERSITAT POLIT CNICA DE CATALUNYA BARCELONATECH 6 D en Project Construction of a Radio Environment Map REM of TV WP ref 3 White Spaces for Barcelona Metropolitan Area Major constituent Interpolation Methodologv Research and Sheet 3 of 7 software Short description Review and comparison selection and implementation of a interpolation method Planned start date 10 10 2014 Real start date 12 10 2014 Planned end date 31 10 2014 Real end date 3 11 2014 Start event End event Internal task T1 Review and comparison of the various methods Deliverables Dates Implemented 3 11 2014 Internal task T2 Selection of the method to implement imerpolation method program Internal task T3 Software implementation Internal task T2 Validation of the GUI software Table 3 Work package 3 Project Construction of a Radio Environment Map REM of TV WP ref 4 White Spaces for Bar
37. environment and can change dynamically its transmission or reception characteristics in order to be adapt to it This adaptability allows it to hop between frequency bands if necessary a feature needed for the opportunistic usage of the spectrum as it allows for a secondary user to clean the channel if it senses that a primary user is transmitting The mechanism that allows a cognitive terminal to be aware of its surroundings and detect if it s interfering a primary user is called spectrum awareness lt can be implemented in different ways One is having a geo localization device in the terminal and use the obtained location to consult an external agent what spectrum zones are available for use at the current location In this technique the devices consult to a REM or Radio Environment Map A REM is a system that combines up to date information about the radio characteristics of the environment measurements taken by the terminals local policies known positions or usage patterns of the primary and secondary users etc and uses them to maintain a real time map of the radio scene of a certain area and to transmit to the cognitive terminals where they can transmit and in what frequencies A REM allows for an easy and economical implementation of a cognitive network because it allows the cognitive terminals to be relatively simple since the REM takes the most complex part of the spectrum awareness task Nowadays the right conditions have appeared to
38. handled through a REST API and with JSON as interchange format The paths and available interactions are listed below lapi used to give commands to the REM different http verbs trigger different actions All requests must have set the headers content application json because all communication is done via JSON files POST tries to add a device to the network the request must carry a JSON with the structure lat xx Ing yy bandwidth zz Being lat Ing the coordinates of the device in decimal degrees and bandwidth expressed in MHz If the request is valid the response will be either the inserted device if the device can be inserted correctly in the network or a JSON containing an error field with a string identifying the error cause Error causes can be that the device is outside the REM grid or that the REM cannot allocate the selected bandwidth at the given location The device object returned will have the same fields as in this example device id 27 Ing 2 143904 lat 41 415651 tx pow 19 2697929297 19877 coverage radius 1045 3093215671413 channels CHANNEL21 The tx_pow field indicates the maximum transmission power for the device at the selected location in dBm Coverage radius is an estimation of the coverage of the device and the channels field is a string with the name of the channels assigned for transmission In case of error the response w
39. he UHF spectrum vacant However this part of the spectrum is still allocated for broadcasting purposes In addition an added difficulty for using this free band is that the vacant channels are not the same at all locations even for relatively small geographical areas With the aim of facilitating the use of the free spectrum without creating interference for the licenced users we propose the construction of a REM for TV WS as the first step for the creation of a test bed for cognitive network A REM is a system that maintains an updated database of the behaviour of the radio medium at a given geographical area This system can be used to provide information about the usage of the radio spectrum at a selected area enabling connected devices to know when a certain channel it s not used by it s licensed users in order access it This type of access constitutes opportunistic usage of the spectrum and it is one of the main applications of cognitive radio technology This thesis covers the design and implementation of a REM manager that will construct the REM from previously taken field samples and a GUI component that facilitates using the REM 1 2 Background Nowadays one of the main problems that the telecommunications industry has to face is that almost all the spectrum useful for wireless radio communications is already allocated However the amount of wireless data bandwidth needed is always increasing This situation poses a very serious
40. ill be error The device at 2 143904 41 415651 asks for too much bandwidth Containing a string related to the cause of the error 51 UNIVERSITAT POLIT CNICA 22 telecom DE CATALUNYA BARCELONATECH DELETE If a JSON with the field id is provided with the request the device with the same device id will be deleted from the network If no json is provided the backend just will delete all the devices on the network For deleting only a device the querry JSON device id 3 the response for a successful querry is message ok params is used to change the parameters that the spectrum broker uses to allocate the devices MinimumPCIR is the minimum CIR primary user needs in order to operate Alpha The path loss exponent used in the REM HeightB and heightM the heights used in the flat earth propagation model MinimumSCIR is the minimum CIR secondary user needs to operate GridSize is the separation between samples in the REM grid The route admits the following verbs GET returns a JSON with the values of all the parameters used example response minimumCIR 12 5 alpha 4 maxPow 25 minSNR 16 heightB 10 heightM 1 gridSize 1 POST updates the parameters to the selected values to update a parameter a JSON with the parameter name and the new value must be send 52 UNIVERSITAT POLIT CNICA 2 telecom
41. lling channel 11 Another drawback for a REM based approach is how dynamic is the radio medium If the medium changes fast REM information will become out dated rapidly This causes overhead in the network as synchronization for the new data is required or capacity issues in the designated channel for sharing REM information as explained in 12 23 BARCELONATECH UNIVERSITAT POLIT CNICA 22 telecom DE CATALUNYA 2 5 REM in TV WS The characteristics of REM enabled cognitive radio are suitable for TV WS The spectrum activity at the TV band is almost static such as the locations of the primary users as described in 8 However a static database is not enough since the Secondary User SU activity is dynamic and knowledge of SU activity is needed for cognitive radio In addition a REM based approach is also useful for keeping the cost of cognitive devices down a needed feature for CR applications in TV WS as described in 8 2 6 Current REM enabled CR standards IEEE 802 22 WRAN The first Cognitive Radio standard is the IEEE 802 22 13 aimed at providing broadband wireless access at rural areas using the TV WS spectrum due to its excellent propagation characteristics The standard defines 3 types of devices An incumbent database that acts as a global REM base stations BS where each one that creates its own wireless cell acts as local REM and as coordinator between the different devices in the cell and Custo
42. m 21 to 21 From 25 to 25 From 28 to 28 From 30 to 30 From 32 to 32 From 35 to 35 From 37 to 38 From 41 to 42 From 46 to 46 From 49 to 60 From 62 to 63 From 65 to 66 lat 41 47500 Ing 2 17829 Bandwidth MHz n Insert the device Pwr 73 48136 Dades del mapa 2015 Google basado en BCN IGN Espa a Termes i condiciodfB grinformeu d un error al mapa 38 BCN BARCELONATECH UNIVERSITAT POLIT CNICA Sch telecom DE CATALUNYA The GUI user can also click on a certain point to obtain information on the available channel blocks If the user clicks on a certain point inside the grid a red marker will appear on the selected location and on the left edge of the screen the geographical coordinates and the available channel blocks in that location will be represented This behaviour can be observed in figure 6 and figure 7 On addition the GUI can also be used to insert simulated cognitive devices in the network deleting them and changing the parameters the REM uses to insert devices or selecting different grid sizes for the Primary User information If the parameters of the REM are changed or the grid size changes all devices are updated and reintroduced in the REM This procedure must be done since the information used to insert the devices into the REM has changed 4 Load Grid Y Load Samples L Display Primary User Power Level Channel 21 v 0 5 kilometers Y Sant Andre de
43. mer Premises Equipment CPE that act as sensors for the REM and end nodes for the cognitive network Figure 4 shows the relations between the three entities mentioned before Transmit antenna characteristics Incumbent database sum Available n i channels i information 8 EIRP Spectrum manager Channel management Coexistence management n Sensing management n Transmit power control limits information channels control GEO SSF enl Occupied Sensing i Sensing Location Channel J reports information information Spectrum automaton a Sensing automaton i information channels control GEO SSF Se e Sensing Figure 4 Relation between the spectrum manager and other cognitive function entities Stevenson C Chouinard G Zhongding Lei Wendong Hu Shellhammer S J Caldwell W IEEE 802 22 The first cognitive radio wireless regional area network standard Communications Magazine IEEE vol 47 no 1 pp 130 138 January 2009 24 UNIVERSITAT POLIT CNICA ach woo DE CATALUNYA BARCELONATECH 3 Methodology project development 1 Rem Architecture The first aspect of the construction was to decide on architecture for the implementation of the REM In order to implement a functional REM and a GUI to comply with the project objectives three functional blocks were implemented The REM manager which is tasked
44. method is based on the Voronoi decomposition of a set of points This decomposition generates a partition of the plane where the point set is located assigning each point a Voronoi cell The Voronoi cell of a given point is the area of the plane that is closer to it than any other point of the set given a certain definition of distance When a new point is introduced in the set or interpolated the Voronoi diagram changes as a cell must be assigned to this new point This new cell will have an intersection with the cells that formed the diagram before its introduction its neighbours Finally the value of the new introduced point is calculated as a weighted average of the values of its neighbours The weights used are proportion of each neighbouring cell area or stolen area that has gone into the new cell containing the interpolated point 3 Linear interpolation This method uses a decomposition of the sample space in triangles All the samples are joined in groups of three creating a surface formed by flat triangular shapes joined by their edges Each of the individual triangles defines a surface and the parameters of that surface can be obtained given the three points that form it To interpolate a point first the point is located into one of the triangles that form the sample space then the value interpolated value for the point its the one that adjusts to the surface equation defined by the points in the triangle where is located
45. n Project Construction of a radio Environment Map REM of TV WP ref 6 White Spaces for Barcelona Metropolitan Area Major constituent Create a mobile Demo application Sheet 6 of 7 Extra work not planned initially Short description Create a demo mobile application that will be used to show the REM working Planned start date 22 12 2014 Real Start Day 20 1 2015 Planned end date 9 1 2015 Real end day 25 1 2015 Start event End event Deliverables Dates Demo mobile 30 1 2015 application Table 6 Work package 6 Project Construction of a radio Environment Map REM of TV WP ref 7 White Spaces for Barcelona Metropolitan Area Major constituent Final Report writing Sheet 7 of 7 Short description Develop the final report of this project using an iterative methodology Planned start date 11 12 2014 Real start date 19 1 2015 Planned end date 30 1 2015 Real end date 6 2 2015 Start event End event Internal task T1 Writing the final report Deliverables Dates Final Report 30 1 2015 Internal task T2 Revising the final report Table 7 Work package 7 18 D tecon UNIVERSITAT POLIT CNICA DE CATALUNYA BARCELONATECH Gantt Diagram 1 6 2015 Ht WBS Name Start Finish sep2014 oct2014 nov 2014 ene2015 Ffeb2015 1 V State of the art review and planning sep15 oct10 Project start puy 14 Bibliographic resea
46. nels length possible e getMaxTxPow insertion point block The objective of this function is to find the maximum Tx power for the device that will not degrade the CIR for the primary users below a certain threshold It assigns the entire available interference margin to the device if needed 32 UNIVERSITAT POLIT CNICA 22 telecom DE CATALUNYA BARCELONATECH The maximum CIR that primary users can affront is a parameter of the REM S CIR N 1 Let SEIRP be the EIRP of the secondary user at the direction of the primary user antenna and let L be the path loss obtained using a flat earth model with exponent alpha is I sEIRPAL hj hm de Then the sEIRP for a given CIR ratio constraint can be found as S E SEIRP UK And if hvpothetic secondarv users with isotropic antennas are used the max transmission power can be equated to sEIRP This formula is used in the pseudo code below to get the max transmission power achievable without breaking the interference constraints for a single device TxPow MaxTxPow for every channel in the block for every cell in the grid if cell is used by primary obtain the distance between the insertion point and the cell find tentativeTxPow for the selected distance if tentativeTxPow lt TxPow then TxPow tentativeTxPow end end end end return TxPow 33 UNIVERSITAT POLIT CNICA 2 telecom DE CATALUNYA BARCELONATECH getCoverage p
47. note that the minimum 250 meters grid resolution is due to performance constraints on the GUI side The test machine was a 2009 Thinkpad t400 with a core2duo Q 2 13 GHz running Ubuntu Linux and on Chromium browser Figure 9 and Figure 10 show the result of a grid size change In them we can observe that the parameters for the selected device has changed The device with device id 2 was using channel 21 with the 500m grid and when the grid size changed to 1km the zone where the device sits is marker ad occupied by the Primary user The device is then introduced in the channel 23 where that cell is marked as free in the new grid 40 UNIVERSITAT POLIT CNICA 22 telecom DE CATALUNYA BARCELONATECH 4 2 REM Manager The two sub components of the REM Manager the back end and the Matlab script perform correctly within the specifications However it is interesting to disclose the performance constrains of the system For the Matlab script the total run time for generating the three used grids was 3 minutes on the same PC used for the GUI testing It might seem like a bottleneck for the system at first glance but as the Primary User activity is almost static the task of creating new GeoJson files has only to be performed seldom The performance of the back end is also not a concern for the grid sizes and algorithms used The median time for a device insertion is around 20 ms and its almost constant for every grid size This migh
48. oviding reliable communications when needed and using efficiently the available spectrum 2 The adaptability of cognitive radio derives from the concept of software radio and software defined radio Software Radio is a radio where all the RF processing including RF channel access is defined on a general computing platform using software commonly using high performance DSPs or FPGAs to process it Software defined radio is an architecture where not all the processing is necessarily implemented by software but at least is configurable by software by switching between different ASICs and other hardware modules 2 The fact that all processing is configurable by software provides an incredible grade of flexibility to the system enabling it to switch between different parts of the spectrum and using different modulations and wireless standards Cognitive radio ads a layer of environmental awareness to a software radio or software defined radio architecture enabling it to change it s transmission or reception parameters autonomously in order to adapt to the current state of the local radio channel The process of sampling the channel reviewing the acquired information and taking a decision based on the obtained and prior knowledge is called the cognitive cycle A simplified version of the cognitive cycle is represented in figure 2 Radio environment Outside world Action RI transmitted stimuli signa S
49. ow Just returns the approximate coverage radius considering the interference level at the grid cell where the device resides and a flat earth path loss model Device js gt this module does not implement any method just defines the Device object and some static properties used by the spectrum broker module Device Height hm in flat earth model Base Station Height hb in flat earth model MaxTx Defines the max transmission power for the device in dBm SCIR devfines the minimum secondary user CIR api js gt Defines the route used by API commands and the functions called by each HTTP verb The functionality and the structure of API calls is described at the appendices params js gt Defines the route and the commands used to modify the REM parameters The functionality is defined in the appendices All this operations also update the web connected client GUI in real time using Sockets io so changes in the cognitive devices list or in the REM parameters are immediately updated 5 Front end The front end webpage interface is based in the Google Maps Javascript API This API allows the user to paint overlay objects to the map The web consists on a HTML with a full page map view and several boxes containing information input and controls The implemented Javascript code around 600 lines in length performs functions such as updating the view contents displaying the map data overlays and communicating wi
50. pectrum holes Noise floor statistics Radio Traffic statistics t scene ES sill analvsis mio y control and Interference spectrum temperature management es estimation and Quantized predictive I channel capacity modeling Transmitter Receiver Figure 2 Simplified version of the cognitive cycle Haykin Simon Cognitive radio brain empowered wireless communications Selected Areas in Communications IEEE Journal on vol 23 no 2 pp 201 220 Feb 2005 21 BARCELONATECH UNIVERSITAT POLIT CNICA 22 telecom DE CATALUNYA 2 2 Cognitive Radio use cases One of the challenges that modern wireless communications have to face is the lack of unassigned spectrum and the lack of utilization of some spectral bands The present regulatory framework for the usage of the electromagnetic radio spectrum is very restrictive enforcing that only the assigned users can operate in a certain spectral band The effect of this strict policies is that some parts of the spectrum are not utilized spectral blanks continuously or that in some geographical locations are completely non used 4 A more local example can be found in 5 where we can find that the average duty cycle for the zone between 75MHz and 3GHz is only 22 57 Figure 3 Therefore there is much room for improvement in the utilization of the radio spectrum TABLE II AVERAGE DUTY CYCLE STATISTICS Frequency ES range MHz Average duty cycle 58 65
51. rch sep 15 oct3 em 1 2 Project planning sep15 oct10 2 W REM architecture definition oct oct17 i i 21 Define involved nodes oct6 oct13 E 23 Define interconnections between nodes oct13 oct17 O 3 V Interpolation methodology oct 13 nov10 i m 3 1 Review and comparison oct13 oct 24 32 Selection oct27 oct 27 i h 3 3 Software implementation oct27 nov I 3 4 Validation nov10 nov10 4 Y Creating a GUI oct 27 dei e 4 1 Requirements oct27 oct27 h 42 Technology assessment oct28 nov 4 3 Software Implementation nov10 nov28 i i 44 Validation of the GUI software dict dici i l i 5 V Integration and validation nov 10 ene 28 54 Integration of the different nodes ene 15 i i 5 2 Validation of the REM ene 28 E 6 Creation of a mobile demo extra task ene 26 i m Final Report i 7 1 Writting Final report ene16 Feb i me 7 2 Revising final report ene28 feb6 ES 19 Figure 1 Gantt Diagram UNIVERSITAT POLIT CNICA 22 telecom DE CATALUNYA BARCELONATECH 1 7 Deviations and incidences The first design contemplated creating the REM as a web application using a Google service called Google Maps Engine as back end provider for the REM Latter this year Google announced that it would close the service soon so a new and dedicated back end program had to be created for the REM This unexpected development increased the length of the integration tasks and also caused delays the development of
52. ry users activity is mostly static information because of the static nature of TV WS and can be stored simply as a file with a particular format in this case a GeoJson JSON is a particular data format that is widely used as a data interchange format in web applications GeoJson is a standard json format used for geospatial applications The second type of information the cognitive device list is not static so it is better stored in a database The database chose in this case is a sqlite3 database It can be accessed using standard SQL commands and it is auto contained making the deployment of the application easier 26 UNIVERSITAT POLIT CNICA q telecom DE CATALUNYA BARCELONATECH GUI The function of the GUI block is to provide an easy way to consult the REM information The GUI is developed as a webpage which uses Google Maps API to provide geographical data and Javascript functions to show the primary and secondary user activity as overlays to the map The GUI also offers administration features The GUI allows introducing simulated devices to the REM at locations selected by the user to delete devices from the REM and to change relevant parameters used by the REM such as the grid size for the primary user activity data or the information about the operational parameters of the cognitive devices that form the network Last of all the GUI is synchronized in real time with the back end so any change to the REM will
53. ry users for all sample sizes linear g natural neighbor r nearest neighborib linear g natural neighbor ir nearest neighbor b sn E 15 O 4 ee 8 A E U Aw g 2 N o 30 9 10 N z SN a Ka p 20 XN E N we Ti NN 5 KN L au di 10 SR E 1 ease a ES 0 50 100 150 20 40 60 80 100 120 140 Samples Samples Figure 14 Alarm Probability versus sample size Figure 15 Primary Not Detected versus sample size 43 UNIVERSITAT POLIT CNICA 1 telecom DE CATALUNYA BARCELONATECH 5 Budget 5 1 Labor costs The personnel cost table below is based on the following figures A junior engineer perceiving 13 04 Euros per hour e The employee costs a 30 more to the employer than the perceived remuneration due to social security and other taxes elevating the cost per hour to 17 4 euros Task Dedicated Cost Hours Euros WP 1 Planning and research 37 643 8 WP 2 Rem architecture definition 15 261 WP 3 Interpolation methodology 47 817 8 WP 4 Creating a GUI 55 957 WP 5 Integration and validation 122 2122 8 WP 6 Creation of a mobile demo 10 174 WP 7 Final Report 75 1305 Total 361 6281 4 Table 8 Labour costs 5 2 Software costs The only non free software used in the development of this project was Matlab for unix systems The cost is described below Matlab license for education Yearly cost 500 Euros Used for 143 days Cost 196
54. s for browsers All the software components except Matlab are open source geojson utils is distributed under the BSD license and the rest of open source components are distributed under the MIT license This work has been developed in the framework of the research project Redes m viles eficientes para la ampliaci n de servicios a nuevos sectores profesionales RAMSES founded by Comisi n Interministerial de Ciencia y Tecnolog a C I C V T project referenceTEC2013 41698 R 14 UNIVERSITAT POLIT CNICA DE CATALUNYA BARCELONATECH 6 D en 1 5 Work plan Project Construction of a Radio Environment Map REM of TV WP ref 1 White Spaces for Barcelona Metropolitan Area Major constituent Planning and Research Sheet 1 of 7 Short description Do an initial bibliographical research about the topic in the literature and define the project scope goals timeplan work packages and interaction procedures with the advisor Real start date 15 9 2014 Real end date 10 10 2014 Start event End event Internal task T1 Bibliographic research Deliverables Dates Project 9 10 2014 Internal task T2 Project Planning Proposal Ang Workplan Table 1 Work package 1 Project Construction of a Radio Environment Map REM of TV WP ref 2 White Spaces for Barcelona Metropolitan Area Major constituent REM architecture definition design Sheet 2 of 7 Short description Define th
55. spectrum broker module Now the spectrum broker uses a pair of basic functions to introduce cognitive devices in the network and does not take into account the interference between the devices Also it does not optimise the assignment of the spectrum for the cognitive users In order to maximise the overall capacity of the network this factors should be taken into account when allocating the devices on a certain channel or assigning their maximum EIRP 2 Implementing and validating an authentication scheme for accessing the REM and implementing privacv for the communications between the users and the REM manager At its current state all API operations are performed on plaintext and without authentication implementing this features would increase significantiv the overall safetv of the users because the REM manager works with private data such as the cognitive users geographical positions also it would protect the REM manager from being tampered with bv unauthorized users 45 UNIVERSITAT POLIT CNICA 2 telecom DE CATALUNYA BARCELONATECH Bibliography 1 Mitola J Maguire G Q Jr Cognitive radio making software radios more personal Personal Communications IEEE vol 6 no 4 pp 13 18 Aug 1999 DOI 10 1109 98 788210 2 Haykin Simon Cognitive radio brain empowered wireless communications Selected Areas in Communications IEEE Journal on vol 23 no 2 pp 201 220 Feb 2005 DOI 10 1109 JSAC 2004 839380
56. t seem counter intuitive as for a greater number of cells the difficulty of introducing a new device increments A possible explanation for this phenomenon is that what really impacts the run time of this operation is the introduction of the device in the database so the bottleneck for this system is I O performance The run time of the most computationally intensive operation updating the whole cognitive network after a parameter change has a run time that seems to be dependent of the number of devices This also correlates with a system limited by the I O operations as more devices on the network means more update operations for the database which are performed sequentially For reference the run time of a update for 20 cognitive devices is close to 500 ms 4 3 Demo App REM Demo App Welcome to the REM test app The demo application is an android app which uses the device geo location features and a bandwidth parameter chosen by the user to insert a new simulated cognitive device through the REM manager API Figure 11 shows the interface of the test app for Android 5 0 1 Please enter the IP address of the REM server 1192 168 1 32 3000 Enter the selected bandwidth in MHz 8 Once the REM manager responds to the query the application shows the received information or error for the newly introduced device shown on Figure 12 Location acquired SUBMIT Figure 11 app interface 41 UNIVERSITAT
57. th the back end The operation of the scripts follows the loading of the page sequentially 1 The browser loads the HTML and the Javascript files 2 When the page and its contents are loaded the load finished event is triggered which calls the function initialize 3 The function initialize loads the Google Maps window and the geoJsons from the backend creating a map view of the Primary User information 4 Queries the backend API for the device list using a GET request to the route lapi and params this loads the Secondary User information and the configured REM operational parameters 5 Opens a socket io connection with the backend 34 UNIVERSITAT POLIT CNICA D telecom DE CATALUNYA BARCELONATECH Once this process has ended the webpage will display the availability grid for a certain channel with markers showing the devices added to the REM and their approximated coverage radius if some of the channels used by them are selected If some change is made in the REM either by the GUI user someone using the API or another GUI user connected to the REM the socket connection will make the view update in real time That applies also if the REM parameters are changed The frontend offers the user the following functionality e Visualization of the data contained in the REM the primary user power levels grid by selected channel and the cognitive devices and channel availability for a selected location
58. the GUI since parts of the back end had to be integrated before making a final version of it and also to adapt parts of the Matlab script used to build the REM data The former circumstances pushed back the delivery of the GUI to the start of December and also lengthened the integration tasks to mid January Moreover an extension task was added to the initially devised tasks This new task was to create a simple android app to showcase the REM API functionality The app used the integrated geo localtion capabilities of the smart phone to insert a simulated device at the REM The app sent a request to the REM manager with the position of the device and the required bandwidth and then received a response from the REM manager with the simulated device characteristics 20 UNIVERSITAT POLIT CNICA 22 telecom DE CATALUNYA BARCELONATECH 2 State of the art of the technology used or applied in this thesis 2 1 Cognitive Radio an introduction The key technology of this thesis is cognitive radio The term cognitive radio was first used in an article Cognitive Radio An integrated agent architecture for software radio architecture 1 and is defined as a radio that is self aware in terms of its operational characteristics It extracts information of its physical environment and uses it to adapt to the current conditions and can exchange and interpret radio environment information with other entities with the objective of pr
59. the cursor s position e Channel availability grid It is a grid overlay that is used to display the information about the primary users behaviour in the area The cell coloring can show two types of information o If the Display Primary User Power Level checkbox in the map controls is not checked The color code will show if primary user activity is detected red or if such activity is not detected and the channel is available for cognitive use o If the after mentioned checkbox is checked the color code will represent the total power detected at each channel e Secondary users Green or grey markers are used to represent the cognitive devices stored in the REM A device will have only one of the two representations if the following conditions are applied o If the device in question is using the UHF channel represented in the GUI then a green marker with a circle that marks its approximate coverage range will represent the device o f the device is not using the selected channel a grey marker wvill represent the device Device tooltip It is a short description of the device main parameters that appears when the user moves the mouse pointer over the device s marker Use cases Changing the displayed UHF channel on the map To change the displayed channel the user has to select the desired channel in the dropdown selection menu shown in fig 15 4 Load Gnd sg Load Samples Display Primary User Power Le El Ghana Dr
60. the event the client side script implements the following code socket on device added function msg devices push new Device msg This snippet of code sets a new callback on the socket connection opened before When a device added event arrives a new device will be created from the contents of the event and added to the client devices list The connection can be also used to send events from the browser client socket emit event name msg hello This would generate a event name event at the backend with content msg hello 36 DE CATALUNYA BARCELONATECH e UNIVERSITAT POLIT CNICA ks telecom 4 Results Despite the setback suffered when Google announced that Google Maps Engine would be closed all objectives and requirements of the project were completed even the extension objective of creating a Demo app for the REM 4 1 GUI The resulting GUI allows the user to consult REM information easily In figure 6 a sample screenshot of the GUI is represented 0 3192168 1 32 3000 Ullastrell Ca n Amat Les Fi ar a 1 Y Load Grid Load Samples Display Primary User Power Level Channel 21 v 1 kilometer L Ci ll Minimum Primary ampsentelles User CIR 12 5 Minimum Secondary User Alpha CH WI d Max Tx power Alella Parc Blocks available From 21 to 21 From 23 to 23 From 25 to 25 From 28 to 30 From
61. the user hovers the mouse over a device marker a tooltip containing information about the device in question will appear as shown in figure 8 37 UNIVERSITAT POLIT CNICA DE CATALUNYA BARCELONATECH D telecom T x a 7 em C 58 Yi Load Grid l Load Samples Display Primary User Power Level Channel 39 y et kilometers v acarisses Min 95 dBm Max 40 dBm Polinv 7 EH Vilanova La Conreria Sant t del Vall s Mateu Cellecs Argent 7 dii ai A m s Coll Premi Viladecavall 12 5 Minimum Secondary User Ke F 15 a Florida CIR hi A a osta WE Be SS El Alpha l 4 y Max Tx power f 25 ae lontgat abadell N 150 Vallromanes Ullastrell Martorelles Alella El Masnou Base Station Height dj 10 Mobile devices Height 1 Submit Select a device id to remove it from the REM Tf left void removes all devices Sant Climent de Llobregat Select a Device id for deletetion Delete Figure 7 sample GUI heatmap visualization Montnegre el Corredor deMunt a g Sant Andreu nys d de Llavaner E 4 AE E man C 37 a Blocks available From 21 to 21 From 25 to 25 From 28 to 28 From 30 to 30 From 32 to 32 From 37 to 38 From 40 to 41 From 46 to 46 From 49 to 57 From 59 to 60 From 62 to 63 From 65 to 66 lat 41 49212
62. tive radio The best solution from the technical perspective would be a cognitive network based in REM architecture A REM implementation would be optimal for the TV WS situation since the radio scene is mostly 12 UNIVERSITAT POLIT CNICA 2 telecom DE CATALUNYA BARCELONATECH static and if a channel is vacant is highly probable that it will remain vacant for a great period of time This allows for a simple REM implementation since the primary user activity is mostly static and predictable and the refresh ratio of the REM information is very low Finally it is not worthless to mention that the implementation of a REM will provide the GRCM research group with a tool for accessing the information about the radio scene easily and in real time as well as a key component to implement a test bed for cognitive devices 1 3 Requirements and Objectives This project corresponds to an early prototype feasibility test Even thought this project does not have many concrete requirements since it does not have to adhere to a standard or implement a market product still there are two main requirements One is to complete a valid implementation of a REM system with a functional GUI at the end of the project and the other is to validate the interpolation algorithm used in the construction of the REM manager a key functionality in the REM implementation From the requirements the following set of goals can be devised 1 Propose a suitable REM ar
63. tton Sry Ue a Alpha x im 4 N so Max Tx power __ Mex Tx power LEZ Delete devices from the network a an Height 7 M cria on s E Mobi doue To delete a device from the network write e Height Ke device id property of the device to be deleted at the EI a input above the delete button can be obtained from m So Select a device i 90 Select a device id s bc device tooltip and click the button to remove Som void removes al Pl void removes all d e k To delete all cognitive devices click the delete E Select a Device id W A Select a Device id for deleteion button leaving the input field emptv for deletetion Deletel Figure 22 REM parameters Figure 23 delete devices mo Blocks available From 21 to 21 From 25 to 25 From 28 to 30 From 32 to 32 From 37 to 43 riga uie Add a simulated device to the REM From 62 to 63 Fons n order to add a new device click on a point inside the grid to select a eA position A red marker will appear at the selected position The available channel blocks for cognitive will appear Input a bandwidth in MHz at the lat 41 41544 P 1ng2 13753 Bandwidth MHz box above the Insert the device then click on the button Pwr 95 dBm Figure 24 Insert Device 50 UNIVERSITAT POLIT CNICA 22 telecom DE CATALUNYA BARCELONATECH API documentation The interaction with the REM manager is
64. urements from the metropolitan area of Barcelona are used for REM construction which had been posteriorly processed to adapt them to the REM structure The access to the REM is carried over a REST API which is also used to administrate it The GUI has been implemented as a webpage that can be used to access the REM data and to administrate the REM Finally an interface which permits to use the REM to manage simulated cognitive devices has been developed in order to showcase the capabilities of this technology UNIVERSITAT POLIT CNICA 22 telecom DE CATALUNYA BARCELONATECH Resum Aquest treball detalla la construcci d un REM Radio Enviroment Map de TV WS TV white spaces a partir de mesures de canal preses a l entorn de l rea metropolitana de Barcelona i de una GUI per permetre la seva consulta Un REM es una peca clau en la implementaci d una xarxa cognitiva ja que proporciona als dispositius cognitius coneixement del medi radio en el que es troben permetent als dispositius decidir quins canals han de fer servir i quan els han d utilitzar per operar de forma ptima Els TV WS son les zones espectrals que estan assignades a televisi per no s usen per a aquest servei des de la transici6 a l estandard DBTV que va augmentar considerablement la efici ncia espectral de les emissions de televisi deixant gran part de l espectre assignat sense utilitzar Aquest espectre es podria utilitzar de forma oportunista
65. variable geographical availability due to different national and local TV broadcasting scenarios The TV WS spectra in Europe is still allocated for TV use and a regulatory framework is on the way for its utilization with Cognitive radio technology 7 via spectrum sharing contracts between the spectrum holder and another party interested in using the spectrum The prospective uses for the TV WS include indoor LAN systems rural broadband and wireless sensor networks 8 2 4 REM based Cognitive Radio A Radio Environment Map REM characterizes the radio medium at a certain geographical area It contains information on the entire radio scene the Primary Users the Secondary Users the terrain the regulatory framework 9 A cognitive device can perform the spectrum awareness task of the cognitive cycle by using a REM to obtain information about the radio medium given its location Using a REM has an inherent advantages against using the device own sensors for detecting the primary user such as hidden node detection problem 10 and can reduce the cost of the Cognitive Devices since their spectrum sensing capabilities can be reduced or eliminated if the REM uses an independent sensor network The main drawback of REM based cognitive radio network is that it needs a dedicated channel for communication between the REM the cognitive devices and or the sensor network This limitation can be somehow limited by using the ISM band as signa
66. with processing the available samples and perform interpolation in order to construct the REM and managing the cognitive users e A storage database to store the REM data and device list e And a GUI to visualize the REM data This architecture excluding the interpolation tasks is implementable as a web application being the GUI a web page and using a back end that acts as REM manager responding to cognitive users requests As the interpolation tasks involved in REM construction was not manageable for a standard web application back end this functionality had to be segregated to a subdivision of the REM manager block implemented in Matlab that performed the interpolation of the primary user data and provided the results to the back end of the web application The final block diagram was the following Administers 4 Updates Figure 5 REM architecture Once decided the block structure and functions what was left to do is to define the implementations for the functional blocks 25 UNIVERSITAT POLIT CNICA 7 telecom DE CATALUNYA BARCELONATECH REM manager Matlab Matlab is used to process and to create interpolated grids from the available channel measurement data samples concerting the primary user activity Matlab provides an extensive set of mathematical functions and file system utilities More specifically it has a native implemented scattered sample interpolation library that

Construction of a Radio Environment Map of TV White Spaces for

Contents

Download Pdf Manuals

Related Search

Related Contents