Zigbee wireless sensor network localization evaluation scheme with weighted centroid method

Transcription

1 Zgbee wreless sesor etwork localzato evaluato scheme wth weghted cetrod method Loesy Thammavog 1, Khamphog Khogsomboo 1, Thaadol Tegthog 2 ad Sathapor Promwog 2,* 1 Departmet of Electrocs ad Telecommucato Egeerg, Faculty of Egeerg, Natoal Uversty of Laos Vetae, LAOS 2 Departmet of Telecommucatos Egeerg, Faculty of Egeerg, Kg Mogkut s Isttute of Techology Ladkrabag, Bagkok, Thalad Abstract. Usg wreless commucato system, approprate ad correct door localzato wth Zgbee sesor etwork ad could provde terestg servces ad applcatos. I ths study the Zgbee trasmsso model wth postog method by usg the relatve-spa expoetally weghted cetrod method for the door localzato. The expermetal results ad aalyze results are evaluated a dstace error. The ZgBee trasmsso model measuremet cossts of 121 postos wth dstace betwee postos to postos s 0.3 meter. The expermetal setup at every posto operated at frequecy bad from 2.3 GHz to 2.5 GHz. The accuracy of estmated posto s cosdered the term of dstace error wth the cumulatve dstrbuto fucto (CDF) of dstace error s show. The result presets optmal value for REWL s 0.2 ad mea of dstace error s 0.65 m. 1 INTRODUCTION The Zgbee wreless etworks are wdely used several applcatos for door ad outdoor localzatos. The formato of localzato s a ma crtero for the capablty of wreless sesor etwork. The wreless etwork caot work properly wthout posto formato. Moreover, the accurate locato estmato s helpful wreless sesor etworks (WSNs) servces such as routg, formato processg [2], [3], ad [5]. Thus, WSNs, localzato has become a very popular research oretato. Based o the requremet for measurg the receved sgal stregth dcator (I) betwee referece odes ad target odes the postog process, WSNs localzato algorthm ca be splt to the rage-based localzato algorthm ad the rage-free localzato algorthm. A good localzato, algorthm should be mmue to ambet evrometal mpacts as well as explct dstaces ad should calculate a posto as soo as possble. A promsg algorthm of localzato s the REWL algorthm wth Zgbee techology. The Zgbee s short-rage wreless techology, lowcost ad low power whch satsfes IEEE stadard for both physcal layer (PHY) ad medum access cotrol (MAC) [8]. The advatage of the Zgbee are tme-savg ad hgh relablty, dstace ragg from 10 meter to 100 meter, sutable for costructo, dustral cotrol, work evromet, Data rate ragg from 125 kbps to 250 kbps ad medcal There are may techques wth postog method. It s avalable ether based o the hardware arragemet, or reled o smulato algorthm. I ths research, the performace of WSNs depeds o the desty of odes ad the qualty of data sgals. The precso of posto could be calculated from the co-ordatos of ay four referece odes. I ths paper we cosder algorthm for door localzato resoluto. We evaluato the Weghted Cetrod method based o I for WSNs. The estmato of posto s cosdered term of dstace error. The followg part of ths paper s arrage as follows. Secto II expla of evaluato of door localzato. Next, the measuremet setup ad results of ths paper are explaed Secto III ad IV. Evetually, summary the coclusos are gve secto V. 2 EVAULTION METHOD OF INDOOR LOCALIZATION 2.1. Receve Sgal Stregth Idcator (I) The I (Receve Sgal Stregth Idcator) s a method to measure the stregth of rado waves becomes dfferet depedg o the space betwee a referece ode ad a target ode. The stregth of receved sgals ca compare by usg probablty dstrbuto based o a statstcal method ad measures the locato. To use ths method, t should measure stregths of sample sgals at predefed varous locatos advace. The Zgbee used sesor etworks provdes a REWL algorthm wth Is. * Correspodg author: sathapor.pr@kmtl.ac.th The Authors, publshed by EDP Sceces. Ths s a ope access artcle dstrbuted uder the terms of the Creatve Commos Attrbuto Lcese 4.0 (

2 2.2 Relatve-Spa Expoetally Weghted Localzato Algorthm (REWL) The REWL algorthm s based o Relatve Weghted Localzato (RWL) for wreless sesor etworks proposed [1] whch the REWL s better tha RWL. We show dfferece betwee two follow equatos 1 ad 2. Ths soluto show how to approach the locato p of the target ode by calculatg the RWL of the coordate. More precsely, the posto of the estmated target ode s gve by. ad y 1 y 1 y y 1 y (4) p 1 1 a (1) 2.3 Performace Evaluato The localzato error s expla as a dstace betwee the exact posto ad the approxmated locato of a target ode. Whle RWL performs solely average of the coordates of referece ode devces to localze bld folded devces, the REWL uses weghts to guaratee a mprovemet of localzato. establshg from the calculato of the RWL equato 1, the equato for determg the posto wth REWL s derved. Reveal the term as sum of oes ad the multplcato of wth oes, equato 1 s expaded to the REWL equato 2. The REWL s a I-based rage-free localzato algorthm. Ths algorthm s ecouraged by the WCL method [4], [6], [7], [8] The weghts are acqured by the respectve postog of the achor I value sde the spa of all the values measured by the target ode. I the estmato of the target posto, the REWL algorthm oblge the referece odes, whch preset hgher I values ad hece are probable to be closer to the target ode. Ths s earg by used the weghtg factor λ, as state by the expoetally movg average cocept. The estmated target ode posto s gve by 3 ZIGBEE WIRELESS SENSOR NETWORK MEASUREMENT I ths research we have desged a model to measuremet for locato detfer of the target ode door evromet. For more accurate localzato, the expermet wll be doe wthout obstructos. I measuremet s carry o by usg Zgbee the term of trasmsso of the sgal stregth at frequecy of 2.4 GHz. The model used for measuremet cossts of 121 postos wth space betwee oe posto ad ext posto s 0.3 meter. The model of measuremet s depcted fgure 1. For x ad y axes o the coordates (0, A), (11, A), (0, K) ad (11, K) are used as referece odes. The measuremet was doe at the room 305 of 3th floor, E-Buldg, Faculty of Egeerg, Kg Mogkut s Isttute of Techology Ladkrabag. p a 1 (2) Where s umber of referece ode rage, p s the REWL approxmated posto of target ode, s the hghest value the spa of the I values measured by the target odes, s the I value of the target ode. From equato 2 ca be soluto the equatos 3 ad 4 Fg. 1. The expermetal setup. x 1 x 1 x x 1 x (3) 4 EXPERIMENTAL RESULTS I ths secto we proposed algorthm of performace aalyss. By cosder smulate the REWL based o WSNs for. The accuracy calculates from referece ode to target ode follow equato 2 ad the localzato error calculates follow equato 5. 2

3 Fg. 2. The CDF of the dstace error wth rage wth dfferet weghtg factor at λ =0.10. Fg. 4. The CDF of the dstace error wth rage wth dfferet weghtg factor at λ =0.20. Fg. 3. The CDF of the dstace error wth rage wth dfferet weghtg factor at λ =0.15. The dstace error of each estmated dstace s show term of CDF. The dstace error x, y axes of REWL algorthm wth λ = 0.1 s show fgure 2 the mea of dstace error s 1.24 m. The dstace error x, y axes of REWL algorthm wth λ = 0.15 s show fgure 3 the mea of dstace error s 0.9 m. The dstace error x, y axes of REWL algorthm wth λ = 0.2 s show fgure 4 the mea of dstace error s 0.65 m. The dstace error x, y axes of REWL algorthm wth λ = 0.25 s show fgure 5 the mea of dstace error s 0.75 m. To ascerta the optmal λ, we smulated a full equpped sesor etwork of the eclosed 2 2 referece ode (4 odes). I the fgures 2, 3, 4, ad 5 are demostratg the CDF of the dstace error wth dfferet weght factor (λ). The smulato result shows several mmum dstace errors depedg o the dstace ad the weght factor (λ). The smallest average of dstace error exsts at λ = 0.20, yelds the best result as show fgure 4. Cosequetly, before beg the localzato process, a cocetrated aalyss of the trasmsso rage, the dmesos of the etwork ad dstace are requred to get the smallest dstace error. Fg. 5. The CDF of the dstace error wth rage wth dfferet weghtg factor at λ =0.25. But practce, terfereces ad obstacles as well as ucerta electroc trasfer elemets wth the trascevers. Afterward, dstaces are estmated by usg REWL algorthm. Cosequetly, the dstace error of each posto s evaluated by mesh cotour ad show fgures 6, 7, 8 ad 9. 5 CONCLUSION Ths paper how to evaluate the Zgbee trasmsso model for door localzato based o I measuremets collected by usg REWL algorthm for WSNs. It provdes the approxmated posto error of the target ode dssmlar test a pots sde a specto power trasmsso loss. Ths research we are summarzg the basc theoretcal ad practcal facts mpact the measured data exceedgly. Therefore, calculatg dstace usg smple model wll result faulty dstace wth at least offset errors ad scalg errors for cocerg aalyss of the dstace error of REWL. 3

4 Fg. 6. The dstace error of each posto at λ =0.10. Fg. 8. The dstace error of each posto at λ =0.20. Fg. 7. The dstace error of each posto at λ =0.15. Moreover, the REWL algorthm ad agast scalg error of dstace measuremet has bee clarfy. Localzato results usg Zgbee ode door evromet are descrbed. Eve though the postog algorthm does ot yet gve the sutable results very precsely, the preseted localzato algorthm mergg wth the Zgbee techology offer lots of advatages. It s very sgfcat advatages are ts mmum resource requremets, fast computato process ad low complexty whch advocate REWL as localzato algorthm Zgbee WSNs. Refereces 1. C. Lauredeau ad M. Barbeau, Relatve spa weghted localzato ofucooperatve odes wreless etworks, Proc. It. Cof. WASA, pp , August, 2009 Fg. 9. The dstace error of each posto at λ = F la. Akyldz, W. Su, Y. Sakarasubramaam, ad E. Cayrc, A survey o sesor etwork, IEEE Commucato Magaze, pp , August H. Lu, H. Darab, P. Baerjee, ad J. Lu, Survey of wreless door postog techques ad systems, Systems, Ma, ad Cyberetcs, Part C: Applcatos ad Revews, IEEE Trasactos o, vol. 37, o. 6, pp , November, J. Blumethal, R. Grossma, F. Golatowsk, ad D. Tmmerma, Weghted cetrod localzato Zgbee-based sesor etworks, Proceedgs of the IEEE Iteratoal Symposum o Itellget Sgal Processg (WISP '07), Madrd, Spa, pp.1-6, October, J. Blumethal, R. Grossma, F. Golatowsk, ad D. Tmmerma, Weghted cetrod localzato Zgbee-based sesor etworks, Proceedgs of the IEEE Iteratoal Symposum o Itellget Sgal Processg (WISP '07), Madrd, Spa, pp.1-6, October,

5 6. J. Blumethal, F. Rechebach ad D. Tmmerma: Posto Estmato Ad hoc Wreless Sesor Networks wth Low Complexty. I Jot 2 d Workshop o Postog, Navgato ad Commucato ad 1st Ultra-Wdebad Expert Talk, pp , March, Krstala, P., Wrawa & Hedratoro, G., Improved Rage-free Localzato Methods for Wreless Sesor Networks, Proc It l Cof. o Electrcal Egeerg ad Iformatcs (ICEEI), pp. I5-1, July, S. Schwarzer, M. Vossek, M Pchler, A. Seltzer, Precse Dstace Measuremet wth IEEE (Zgbee) Devces, Rado ad Wreless Symposum IEEE, pp , Jauary, Y Che, Q Pa, Y Lag, Z Hu, Adaptve Weghted Cetrod Target Localzato Algorthm Based o I WSN, Computer Scece ad Iformato Techology (ICCSIT), 3rd IEEE Iteratoal Coferece, pp , July,