A Human Detection Method for Residential Smart Energy Systems Based on Zigbee RSSI Changes

Transcription

1 B. Mrazovac et al.: A Huma Detectio Method for Residetial Smart Eergy Systems Based o Zigbee RSSI Chages 819 A Huma Detectio Method for Residetial Smart Eergy Systems Based o Zigbee RSSI Chages Boja Mrazovac Mila Z. Bjelica Draga Kukolj Seior Member IEEE Braislav M. Todorović Member IEEE ad Draga Samardžija Member IEEE Abstract I this article the device-free huma presece detectio method based o radio sigal stregth variatios is proposed. The method exploits the kow fact that huma body iterferes with radio sigals by causig fadig ad shadowig effects. Itroduced irregularities i the radio propagatio patter idicate possible presece of a huma. The proposed method is icorporated ito the existig platform for itelliget residetial eergy maagemet. As opposed to covetioal solutios which utilize a complex set of sesors for huma detectio the proposed approach achieves the same oly by aalyzig ad quatifyig radio sigal stregth variatios icorporated i messages exchaged betwee 2.4 GHz radio trasceivers. Oe of the key beefits of the proposed solutio is the itegratio of the detectio algorithm ito the smart power outlets ad smart light switches. Such a approach improves iteractios i smart home systems eables itelliget power cosumptio maagemet ad low istallatio cost 1. Idex Terms Object detectio RF sigals sesorless cotrol wireless sesor etworks. I. INTRODUCTION Radio sigal propagatio characteristics are sigificatly versatile across differet eviromets especially idoors where the sigal ca be absorbed reflected scattered or diffracted by may objects i its propagatio path as explaied by Ababeh [1] ad Zhou et al. [2]. At microwave frequecies absorptio by molecular resoace is a major factor affectig the radio propagatio (Youssef et al. [3]). Therefore the presece of a huma subject withi the wireless etwork rage results i sigificat sigal stregth variatios at the receiver iput whereas the degree of variatios is correlated with the level of huma motio. Due to the ature of radio sigals a approach for huma presece detectio would be to use the impact of a huma body o radio sigal s propagatio characteristics. Wireless sesor etworks play a importat role i smart home evirometal challeges. Some of the smart home systems such as [4]-[7] combie huma presece detectio 1 This work was partially supported by the Miistry of Educatio ad Sciece of the Republic of Serbia uder the Grat TR Boja Mrazovac Mila Z. Bjelica ad Draga Kukolj are with the Faculty of Techical Scieces Uiversity of Novi Sad Serbia ( {boja.mrazovac mila.bjelica draga.kukolj}@rt-rk.com). Braislav M. Todorović ad Draga Samardžija are with the RT-RK Istitute for Computer Based Systems Novi Sad Serbia ( {braislav.todorovic draga.samardzija}@rt-rk.com). Cotributed Paper Mauscript received 07/12/12 Curret versio published 09/25/12 Electroic versio published 09/25/ /12/$ IEEE sesors with power outlets i order to provide the iteractivity ad eergy savigs i smart homes. The authors of this paper exteded the existig smart eergy setup to detect irregularities i the radio propagatio patters for the purpose of huma presece detectio. The proposed method was motivated by a eed to decrease the istallatio complexity ad developmet costs of a system which icorporates specific huma detectio sesors. As opposed to covetioal smart home solutios the proposed presece detectio is performed without additioal sesors oly by aalyzig ad quatifyig Received Sigal Stregth Idicator s (RSSI) variatios at the iputs of 2.4 GHz (Zigbee) radio trasceivers embedded i wireless odes. RSSI is the most commoly used sigal characteristic as it is much easier to estimate by the receiver tha Time of Arrival (ToA) or Agle of Arrival (AoA) which require specialized hardware. Wireless smart power outlets ad light switches from the preistalled residetial electrical istallatios used i the proposed solutio are described i the previous work [7]. The firmware of the power outlets ad light switches is exteded for providig the RSSI values o demad which eables them to be used as ivisible sesors i may applicatios for smart homes from surveillace to eergy savig. By icorporatig the proposed algorithm smart outlets ad light switches detect huma presece without decreasig the quality of the regular commuicatio used for eergy-moitorig related operatios. The paper is structured as follows. I Sectio II a overview of device-free methods for huma presece detectio is give. The proposed method is explaied i Sectio III. I Sectio IV the ecosystem for smart home (ESH) which applies the proposed method is described. Experimetal results are give i Sectio V. The paper cocludes with Sectio VI. II. RELATED WORK The use of the shadowig effect caused by a huma subject movig withi the lie-of-sight path betwee the trasmitter ad the receiver for huma presece detectio is proposed by Woyach et al. [8]. That approach maily based o RSSI variatios moitorig at the receiver has bee exteded by Puccielli et al. [9] for outdoor people coutig mechaism. The feasibility of itrusio detectio based o the sigal stregth variatios i wireless sesor etworks (WSN) has bee ivestigated by Lee et al. [10]. The authors have succeeded to characterize the sigal stregth variatios ad to traslate them ito sufficiet iformatio that correspods to huma activity. Sigificat sigal stregth variatios based o

2 820 IEEE Trasactios o Cosumer Electroics Vol. 58 No. 3 August 2012 a huma s movemet are cofirmed experimetally ad the preseted idea is exteded for a idoor automated people coutig mechaism [11]. Itruder detectio method [12] eabled by exploitig RSSI measuremets cofirms the hypothesis that irregularities i the RSSI sigature ca be used as huma presece idicatio. Through distributed processig of RSSI samples odes deployed i a idoor eviromet ca also detect huma presece ad possibly help i localizatio ad trackig of movig idividuals as show by Kaltiokallio et al. [13]. The use of RSSI variatios due to radio irregularity for security threats detectio alogside a roadway demostrates the ability of applyig passive WSN for the outdoor surveillace (Puzo et al. [14]). Patwari ad Wilso [15] explaied how multi-path fadig ca be used i device-free localizatio systems. I such systems a huma positio ca be iferred by measurig the absorptio reflectio scatterig ad diffractio of electromagetic waves itersected by a huma body. Device-free huma localizatio i idoor eviromets usig RF sesor etworks is the topic of several researches [16]-[19]. The phrase RF sesor etwork comes from the fact that the wireless etwork itself is the sesor usig RF sigals to probe the eviromet. Device-free localizatio is a method i which a huma does ot eed to carry ay form of wireless trasceiver to be detected. Compared to existig work this paper focuses o a devicefree huma presece detectio algorithm desiged for the residetial smart eergy system based of smart power outlets ad light switches (smart odes). I order to make the system reliable ad cost-effective solutio-specific huma detectio algorithm ad the smart odes firmware have bee implemeted as described i the followig sectios. III. HUMAN PRESENCE DETECTION METHOD The proposed algorithm for huma presece detectio eables the ESH to be always aware of huma presece or motio. ESH is based o wireless 2.4GHz (IEEE ) smart outlets ad light switches which are cotrolled by the home cotroller device. Mai advatages of the preseted solutio are: (1) the reductio of required umber of physical devices ad the elimiatio of additioal sesors solutios (such as presece sesors or RFID tags); (2) the RSSI moitorig algorithm applied solely to the smart electrical istallatios is used to icrease the iteractio with the eviromet. By aalyzig RSSI variatios of the messages exchaged betwee odes the system ca geerate a fuctioal status depedet o people eterig or leavig the room. The proposed method is maily based o the shadowig effect betwee statioary wireless odes which lie-of-sight is obstructed by a huma body as illustrated i Fig. 1. RSSI ofte fluctuates i differet eviromets with higher or smaller variatios aroud its mea value. However RSSI variatios over a period of time sigificatly icrease i the presece of a huma. Whe obody is preset i a room the iitial RSSI variatios eable the defiitio of the iterval of iitial sigal stregth variatios (ISV). ISV is set durig the system iitializatio by aalyzig the exchaged messages ad RSSI samples for each commuicatio lik that create evirometal radio image. Durig the iitializatio there are o humas i the room ad the RSSI is affected oly by evirometal ad device s properties. Fig. 1. Huma presece detectio cocept based o wireless smart outlets ad light switches utilizatio supported by the aalysis of RSSI variatios withi the exchaged commuicatio messages. The RSSI variatios aalyzed durig iitial coditios are used to defie the high ad low ISV thresholds. The thresholds are defied by usig a set of RSSI samples take withi a predefied time iterval. The duratio of the iterval is empirically defied ad verified by a umber of experimets to be oe miute. At least two wireless odes with the associated cotrol uit are used for the system realizatio. The home cotrol uit (HC) is implemeted as a software module for maagig the cotrol message flow ad moitorig the RSSI variatios betwee the odes. As the fial result the HC is able to cotrol ay appliace coected to smart outlets through o/off switchig. The umber of odes used for this research was 4 ad the pollig time of each ode was 100ms. This way the system was able to detect a huma which rus at the fastest ruig speed without uecessary frequet pollig that icreases the system processig load. Smart outlets smart light switches ad the HC icorporate specific software for iitial sigal stregth measuremets Po i order to defie thresholds Th1 ad Th2. The iitial sigal stregth measuremets for each ode ad thresholds settig for the time period t are calculated durig the start-up phase. Takig ito accout the atteuatio factors of distace Pd ad spatial obstacles Pa the sigal stregth variatio ΔRSSI ca be calculated as: RSSI P( t 1) P( t) P( t) Po Pd ( t) Pa ( t) P( t 1) Po Pd ( t 1) Pa ( t 1) RSSI presece RSSI Th 2 RSSI opresece Th1 RSSI Th 2 Th1. (1)

3 B. Mrazovac et al.: A Huma Detectio Method for Residetial Smart Eergy Systems Based o Zigbee RSSI Chages 821 The received power (Pr) is directly affected by the trasmitted power (Pt). Accordig to Friis free space trasmissio equatio the sigal power at the receiver iput is decreased quadratically with the distace from the trasmitter. Therefore the distace atteuatio factor Pd ca be approximated with: Pr 2 1 d Pt Gt Gr 4 where Gt ad Gr deote gais of the trasmitter s ad the receiver s atea λ is the wavelegth ad d deotes the distace betwee the trasmitter ad the receiver. The obstacles atteuatio factor Pa ca be calculated as sigal losses itroduced by walls whe traversig the direct propagatio path: N Pa k i li i 1 where ki deotes the umber of peetrated walls of type i ad li deotes the atteuatio due to the wall of type i. Durig the start-up phase the ISV high ad low thresholds are determied as: Th1 ( RSSI Th 2 ( RSSI mi max RSSI RSSI / RSSI / RSSI mi max 2 ) 100 % ) 100 % where ΔRSSI mi ad ΔRSSI max deote miimal ad maximal values from the set of samples ad RSSI deotes the mea value of the ΔRSSI. Withi the ISV the ΔRSSI ca vary but the presece of a huma subject will ot be reported. Whe the huma subject eters ito the sesig area ΔRSSI variatios are stimulated to exceed the predefied ISV thresholds resultig i the detected presece. ΔRSSI varies sigificatly across differet eviromets makig the defiitio of uiversal thresholds difficult. Therefore the adaptive ISV thresholds defiitio is ecessary durig the iitial phase. Thresholds are recalculated periodically after the layout of objects (such as furiture) i the room is chaged. The sigal stregth deviatio is calculated over a set of oe miute slidig widow of RSSI samples by: ( RSSI i RSSI ) RSSI RSSI i. (5) i i Durig the iitial phase the sigal stregth deviatio is very low. Whe a huma eters the room ad obstructs oe of the propagatio paths the RSSI deviatio o that radio lik is icreased. After the huma leaves the room RSSI variatios are decreased to miimal values. If the huma moves a object (such as a chair) before leavig a room the sigal deviatio becomes slightly icreased. The additioal software mechaism is implemeted to moitor the dyamic of the sigal stregth deviatio over a period of evirometal updates (PEU). If the sigal deviatio is costat durig the PEU iterval regardless of the iitial value the room layout chage is recogized ad ew ISV bouds are automatically recalculated based o the last samplig miute. i 1 (2) (3) (4) IV. SYSTEM DESIGN FOR HUMAN PRESENCE DETECTION The proposed method for huma detectio is applied to the smart eergy residetial system [7] show i Fig. 2. Fig. 2. ESH smart outlets smart light switch ad the home cotroller. The ESH is composed of the home cotroller ad a umber or smart power outlets ad light switches. The commuicatio cotrol periodic pollig mechaism ad the RSSI data sets aalysis are implemeted withi the core modules of the home cotroller. HC (illustrated i Fig. 3) is made as the software module based o POSIX/C ope stadards which provide scalability. The software is platform idepedet ad ca be easily ported to various POSIX-based target cotrollers. Fig. 3. Home cotroller software desig overview. The device hadler is coected to device drivers uits providig a commuicatio mechaism with the odes via the cetral processig uit. The commuicatio module is i charge of sedig cotrol messages as resposes to detectio evets. The RSSI aalyzer is used for periodic pollig of the smart odes i order to retrieve the curret RSSI values. The list of odes addresses is accessible by the RSSI aalyzer. The odes are polled every 100ms whereas the received values are stored i the local database. After the ode receives the pollig commad from the HC its RSSI vector is set as a broadcast message. The message cotais a set of RSSI samples toward all other odes. Whe oe ode is polled other odes are i the listeig mode. Therefore there is o iterferece or superpositio betwee their radio sigals. The broadcasted message is received by the cotroller ad other odes eablig them to update their RSSI vectors with the values of sigal stregth from that lik. After the message is received the values are stored i the

4 822 IEEE Trasactios o Cosumer Electroics Vol. 58 No. 3 August 2012 local database whereas the RSSI aalyzer is suspeded to wait aother 100ms for the subsequet ode pollig. The procedure is repeated util all odes provide their RSSI values correspodig to all liks. After the RSSI matrix is completed ΔRSSI ad the sigal stregth deviatio are calculated by (1) ad (5) respectively. After the thresholds are defied the pollig procedure cotiues obtaiig ew samples which are compared with the ISV thresholds. If resultig samples exceed the ISV thresholds the presece of a huma is recogized. Otherwise o huma is detected. The cocept of behavioral patters proposed by Bjelica et al. [20] eables the ESH to be used for various setups related to the eviromet. Behavioral patters are give i the form of XML scripts which defie timely actios ad reactios to exteral evets such as huma detectio with the primary goal to achieve desired iteractive eviromet. The script iterpreter module (depicted as a block i Fig. 3) is resposible for the script executio ad iterpretatio. The proposed algorithm ca be easily itegrated ito wireless home power maagemet systems ad other similar solutios by extedig their origial purpose with the detectio capability. Smart power outlets ad light switches fit ito existig electrical istallatios o the wall. Smart power outlet provides power to electrical devices with stadard flat twopole AC power plug (CEE 7/16) which is desiged for voltages up to 250V ad currets up to 2.5A. Besides simple o/off switchig smart sockets are able to pass ay percetage of power to the cosumig electrical device (e.g. light dimmer). IEEE trasceiver (2.4GHz) is used as the commuicatio module. Smart outlets are powered from Vac (±10%) 50Hz curret electric power supply. It is a iexpesive ad the safest way to provide full compatibility with the regulatory requiremets. With a average curret of 35mA ad operatioal voltage of 3.3V for a outlet ad 2.4V for a switch the power supply cosumptio is 0.12W per outlet ad 0.08W per switch. The smart outlets ad light switches icorporate specific firmware which is implemeted to eable: (1) the access to the cosumptio overview o demad (2) switchig the plugged devices o or off ad (3) the probe of the eviromet by broadcastig RF messages to the eighborig odes. The firmware modules are illustrated i Fig. 4. Fig. 4. Smart power outlet ad light switch firmware desig overview. The module for power measuremet periodically provides its curret RSSI samples to the HC. The power cosumptio o daily weekly ad mothly basis are calculated by the HC ad stored ito its local database. The module for RSSI broadcast which is the extesio of the origial firmware waits for a evet from the HC which actuates the message broadcastig. The same module receives broadcasted messages from other odes durig the pollig cycle. Commad iterpreter executes commads received from the HC such as dimmig cotrol switchig the plugged device o or of etc. V. EXPERIMENTAL RESULTS The system described i Sectio IV was istalled i two buildigs with walls built of differet materials: (a) a combiatio of cocrete ad gypsum walls with fiberglass isolatio (b) alumium walls with plastic covers ad fiberglass isolatio. Four odes were used for each experimet three outlets ad oe light switch placed at a elevatio of 40 cm (outlets) ad 120 cm (switch) above the floor. The testig room built of cocrete ad gypsum walls (further referred to as R1) was cm whereas the room built of alumium ad plastic walls (further referred to as R2) was cm large. Layouts of test rooms ad the positios of a huma subject (show as poits P1-P5 i R1 apropos P1-P6 i R2) ad odes positios (show as squares N1-N4) are illustrated i the Fig. 5. Fig. 5. Layouts of the test rooms. Left the room R1 is made as a combiatio of cocrete ad gypsum walls. Right the room R2 is made of alumium-plastic walls. Poits (marked with P) show huma s positios whereas odes are marked with symbols N1-N4. The test sceario for R1 was the followig: o humas were preset i the room for a period of two miutes therefore o detectio was reported. Oce a huma subject etered the room he walked aroud the room passig the positios P1- P5 show i the Fig. 5 - left. After a miute of walkig the huma subject was stadig i each positio for a miute without movemets. The sceario was defied to ivestigate the hypothesis whether is possible to detect ad distiguish betwee huma motio huma presece ad huma absece i the room. A example of RSSI variatios patter i R1 is show i the Fig. 6 for the liks N3 N1 N3 N2 ad N3 N4. Maily because of the huma body s absorptio ad reflectio effects the variatios i RSSI patters were quite satisfactory for accurate huma presece detectio.

5 B. Mrazovac et al.: A Huma Detectio Method for Residetial Smart Eergy Systems Based o Zigbee RSSI Chages 823 N3 liks Iit Walk P1 P2 P3 P4 P5 No humas RSSI samples 8bit value N3->N1 N3->N2 N3->N4 Fig. 6. RSSI samples observed i the first experimetal room (R1). The coclusios from the first experimet are as follows: (1) RSSI for wireless odes which commuicate far from the huma varies slightly or has a costat value; (2) whe the huma is positioed closer to a ode without obstructig the lie-of-sight RSSI varies sigificatly (the larger variatio is the cosequece of sigal reflectios); (3) whe the huma obstructs the lie-of-sight the RSSI is atteuated ad the lower ISV threshold is exceeded. It is sufficiet to exceed the ISV oly o oe lik ad the detectio would be reported. The secod experimet was performed i a buildig whose walls were built of alumium ad plastic slices isolated with fiberglass. The layout of R2 is illustrated i Fig. 5 - right. The squares idicate odes (coutig from N1 to N4) whereas the poits (P1 to P6) idicate huma subject s positios. The test sceario was slightly differet from the sceario explaied i the previous experimet. Iitially o humas were preset i the room for a period of two miutes ad o detectio was reported. Oce a huma etered the room he was stadig i each positio P1-P6 for a miute without movemets. After the RSSI samples were collected the huma walked for a miute passig the positios P1-P6. A example of the collected RSSI samples that correspod to liks N3 N1 N3 N2 ad N3 N4 is show i the Fig. 7. N3 liks Iit 210 8bit value P1 P2 P3 P4 P5 P6 Walk No humas RSSI samples N3->N1 N3->N2 N3->N4 Fig. 7. RSSI samples observed i the secod experimetal room (R2). Startig from the iitial samples it ca be oticed that the sigal stregth varies eve if o huma is preset. The wall structure i R2 formed a Faraday s cage ad the sigal was reflected by the walls. Therefore the sigal stregth deviatio is higher tha the deviatio measured i the previous experimet but is adequate to detect ad differetiate betwee huma presece ad huma motio from the empty room. At the ed of the experimet the huma subject was walkig aroud the room causig higher RSSI variatios. After oe miute of walkig the huma subject left the room. The room was empty agai as it was at the begiig of the experimet but oly for a miute. The coclusio from the secod experimet is slightly differet from the previous oe. Although the iitial radio map was disturbed i this eviromet because the wall reflectios were iterfered with the reflectios caused by the huma body huma presece ad motio were successfully detected for most of the positios. Huma absece was also recogized. The experimets cofirmed that certai positios i large rooms such as the R2 ca be out of the detectio rage (P6). The detectio accuracy of the proposed method was aroud 84% i total. For stadard room dimesios such as the R1 the detectio accuracy was aroud 98% i total. The detectio accuracy for both experimets is show i Fig. 8. Correct detectios [%] Huma positios Empty Walk P1 P1 P2 P2 P2 P3 P3 P3 P4 P4 P4 P5 P5 P5 P6 P6 Fig. 8. Huma presece detectio accuracy for both experimetal scearios preseted for each subject s positio ad trasitio. As show i Fig. 8 the detectio accuracy is high more tha 99% particularly for huma walkig empty room detectio ad trasitios from oe positio to aother. Each received RSSI sample is compared with the predefied ISV thresholds i order to distiguish the detectio. The presece of a subject i the positios which are slightly away from the direct RF liks such as P1 i both experimets is recogized with approximately 5% of false detectios. Huma presece i the positio P5 i R2 is detected with approximately 10% of false detectios. The positio P6 i R2 which was the farthest positio from all odes iduced very low sigal stregth variatios. The algorithm did ot detect a huma stadig i the positio P6 which was marked as the blid spot. The solutio to improve the detectio accuracy i larger rooms is to istall additioal odes for a better radio coverage. VI. CONCLUSION The proposed device-free RF based huma detectio method is icorporated i the etwork of smart power outlets ad light switches. The outlets ad light switches are exteded to detect huma presece i order to icrease user awareess ad to improve huma-computer iteractio i smart homes. For certai applicatios i smart homes the proposed method is eve more reliable ad robust tha other covetioal presece sesor techologies. I cotrary to PIR sesor which is maily depedat o ambiet temperature the RF based detectio accuracy is ot disturbed with the typical temperature chages. Whe the subject is statioary for a period of time the PIR sesor is uable to detect huma presece. The RF based huma detectio successfully detects the presece eve i such demadig cases. The authors believe that this idea will ecourage other maufacturers to apply the preseted method to their smart outlets i order to improve the user awareess i smart homes. R1 R2

6 824 IEEE Trasactios o Cosumer Electroics Vol. 58 No. 3 August 2012 REFERENCES [1] N. Ababeh Radio irregularity problem i wireless sesor etworks: New experimetal results Proc. of IEEE Saroff Symp. pp [2] G. Zhou T. He S. Krishamurthy ad J. A. Stakovic Impact of radio irregularity o wireless sesor etworks Proc. of the 2d Itl. Cof. o Mobile Sys Apps ad Services (MobiSys '04) pp [3] M. Youssef M. Mah ad A. Agrawala Challeges:device-free passive localizatio for wireless eviromets Proc. of the 13th aual ACM Itl Cof. o Mobile Computig ad Networkig pp [4] V. Sudramoorthy G. Cooper N. Lige ad Q. Liu Domesticatig Eergy-Moitorig Systems: Challeges ad Desig Cocers IEEE Pervasive Computig vol. 10 o. 1 pp [5] G. Sog F. Dig W. Zhag ad A. Sog A wireless power outlet system for smart homes IEEE Tras. Cosumer Electroics vol. 54 o. 4 pp [6] D. M. Ha ad J. H. Lim Desig ad implemetatio of smart home eergy maagemet systems based o zigbee IEEE Trasactios O Cosumer Electroics vol. 56 o. 3 pp [7] B. Mrazovac M. Z. Bjelica I. Papp ad N. Teslic Towards ubiquitous smart outlets for safety ad eergetic efficiecy of home electric appliaces Proc. of Itl. Cof. o Cosumer Electroics (ICCE 11) Berli (Germay) pp [8] K. Woyach D. Puccielli ad M. Haeggi Sesorless sesig i wireless etworks: Implemetatio ad measuremets Proc. of WiMee'06 Bosto (USA) pp [9] D. Puccielli A. Foerster A. Puiatti ad S. Giordao Radio-based trail usage moitorig with low-ed motes Proc. of the 7th IEEE Iteratioal Workshop o Sesor Networks ad Systems for Pervasive Computig (PerSes 11) Seattle (USA) pp [10] Pius W.Q. Lee Wisto K.G. Seah H.P. Ta ad Z.X. Yao Wireless sesig without sesors - a experimetal study of motio/itrusio detectio usig RF irregularity Jour. Measuremet Sci. ad Tech. Spec. Issue o Wireless Sesor Networks: Desigig for real-world deploymet ad deploymet experieces vol. 21 o [11] W.C. Li Wisto K.G. Seah ad W. Li Exploitig radio irregularity i the iteret of thigs for automated people coutig Proc. of the 22 d IEEE Symposium o Persoal Idoor Mobile ad Radio Commuicatios (PIMRC 11) Toroto (Caada) [12] S. Hussai R. Peters ad D. L. Silver Usig received sigal stregth variatio for surveillace i residetial areas Proc. of 9th ACM/IEEE Iteratioal Coferece o Iformatio Processig i Sesor Networks (IPSN '10) vol pp [13] O. Kaltiokallio M. Bocca ad L. Eriksso Distributed RSSI processig for itrusio detectio i idoor eviromets Proc. of 9th ACM/IEEE Itl. Cof. o Iformatio Processig i Sesor Networks (IPSN 10) pp [14] Eri L. Puzo Rya P. Breer T. Owes Walker ad Christopher R. Aderso The Matrix: a roadside wireless security system Proc. of Virgiia Tech Symp. o Wireless Persoal Comm. USA [15] N. Patwari ad J. Wilso RF sesor etworks for device-free localizatio ad trackig Proc. of the IEEE vol. 98 o. 11 pp [16] M. Moussa ad M. Youssef Smart services for smart eviromets: Device-free passive detectio i real eviromets Proc. of IEEE PerCom-09 pp [17] J. Wilso ad N. Patwari Radio tomographic imagig with wireless etworks IEEE Tras. o Mobile Comput. vol. 9 pp [18] D. Zhag J. Ma Q. Che ad L. Ni A RF-based system for trackig trasceiver-free objects Proc. of IEEE PerCom 07 pp [19] G. Deak K. Curra ad J. Codell History Aware Device-free Passive (DfP) Localisatio Image Processig & Commuicatios Joural vol. 16 o. 3-4 pp [20] M. Z. Bjelica B. Mrazovac N. Teslic Evaluatio of the available scriptig laguages for home automatio etworks: Real world case study Proc. of TELSIKS 11 Nis (Serbia) BIOGRAPHIES Boja Mrazovac has received B.Sc. ad M.Sc. degrees from the Faculty of Techical Scieces Uiversity of Novi Sad Serbia. He is curretly employed as a software desig egieer at RT-RK Istitute for computer based systems ad he is also workig towards his PhD degree at the departmet for Computer Egieerig ad Computer Commuicatios. He is curretly erolled i several research projects icludig home automatio ad eergy aware systems. His expertise also covers the rage of embedded platforms ad software with the emphasis o video ad radio sigal processig algorithms. Mila Z. Bjelica has received B.Sc. ad M.Sc. degrees from the Faculty of Techical Scieces Uiversity of Novi Sad Serbia. He is curretly employed as a software desig egieer at RT-RK Istitute for computer based systems ad as the teachig assistat at the Uiversity of Novi Sad. He is also workig towards his PhD degree at the departmet for Computer Egieerig ad Computer Commuicatios. His research iterests ad expertise iclude ovel immersive cotext-aware systems usability assessmet ad home automatio systems frequetly applied to the differet cosumer electroics devices. Draga Kukolj (M 97-SM 06) received his Diploma degree i cotrol egieerig i 1982 M.Sc. degree i computer egieerig i 1988 ad Ph.D. degree i cotrol egieerig i 1993 all from the Uiversity of Novi Sad Serbia. He is curretly a Professor of computer-based systems with Departmet of Computig ad Cotrol Faculty of Egieerig Uiversity of Novi Sad. His mai research iterests iclude soft computig data miig techiques ad computer-based systems itegratio with applicatios i video processig ad digital sigal processig. He has published over 100 papers i referred jourals ad coferece proceedigs. He is the coordiator of Itellectual Property Cetre of Faculty of Egieerig. Braislav M. Todorović (M 09) received his Diploma ad M.Sc. degrees from the Faculty of Electrical Egieerig Uiversity of Belgrade Serbia ad Ph.D. degree from the Faculty of Techical Scieces Uiversity of Novi Sad Serbia i ad 1997 respectively. He is a seior research fellow at the RT-RK Istitute for computer based systems ad a professor at the Military Academy Uiversity of Defese Belgrade. His research iterests iclude radio telecommuicatios ad digital sigal processig. He has authored or coauthored about 100 peerreviewed joural ad coferece papers ad two books. He is a Fellow of the Istitute of Naotechology (UK). Draga Samardžija (M 02) received his B.S. degree i electrical egieerig ad computer sciece from the Uiversity of Novi Sad Serbia i 1996 ad M.S. ad Ph.D. degrees i electrical egieerig from Wireless Iformatio Network Laboratory (WINLAB) Rutgers Uiversity USA i 2000 ad 2004 respectively. Sice 2000 he has bee with Bell Laboratories Alcatel- Lucet where he is ivolved i research i the field of the ext geeratio wireless systems. He is also teachig at the Uiversity of Novi Sad. He authored over 50 joural ad coferece publicatios. As a member of the BLAST project team he is a recipiet of the 2002 Bell Labs Presidet's Award. He received the Cetral Bell Labs Teamwork Award for the HSDPA Demostratio Team 2003.