Analysis of Energy Consumption of Virtual MIMO Wireless Sensor Network

Transcription

1 JOURAL OF ETORKS, VOL. 7, O., DECEMBER 0 0 Analysis of Energy Consumpion of Virual MIMO ireless Sensor ework Yongxian Song,. School of Elecronic and Informaion Engineering, Jiangsu Universiy, Zhenjiang, China soyox@6.com Rongiao Zhang and Zhuo Shen. School of Elecronic Engineering, Huaihai Insiue of Technology, Lianyungang, China Asrac In order o solve he prolem ha he exising virual muliple-inpu muliple-oupu (Virual MIMO) rouing algorihm isn' suiale o isomorphism wireless sensor nework, virual MIMO clusering (VMC) algorihm which is applicale o small and medium scale isomorphism S is proposed. By comining he energy-efficien virual MIMO communicaion echnology wih he mehod ha cluser heads are seleced randomly and cyclically, energy load of nework is alanced and life of S is exended. e uild energy es plaform of wireless sensor nework wih microconroller MSP430F35 and wireless radio ransceiver chip CC40. The relaion eween ransmiing power and he RSSI is researched y he experimenal plaform in greenhouse, he pah loss facor is solved, and he energy model of virual MIMO clusering nework is creaed. Then, we focus on he effec of he nework size, node densiy and pah loss facor on he virual MIMO S energy-saving performance. To achieve he opimizaion ojecive ha he longes life of he nework, we adop he geneic algorihm o opimize he raio of cluser head which is a key parameer of S. The simulaion resuls show ha he VMC has more energy-efficien and longer lifeime han LEACH. hen he parameers of nework srucure are appropriae, he lifeime can e exended several imes. Index Terms wireless sensor neworks; clusering algorihm; virual muliple-inpu-muliple-oupu (MIMO); energy efficiency; load alancing I. ITRODUCTIO ireless sensor nework (S) is a comprehensive inelligence informaion sysem which is made up of a numer of low-cos and energy-consrained sensor nodes y way of self-organizaion. In view of he sensor nodes are usually deployed randomly, he nodes will die when he aery power runs ou ecause of small aeries which are difficul o e replaced. So energy is a major facor which resrics he lifeime of S. To he exen possile o prolong he lifeime of S, i is very imporan o design of low-power srucure of he nodes and research of energy-efficien rouing algorihm. A presen, he domesic and foreign research personnel has oained preliminary resuls in wireless sensor nework sudy, u, a lo of prolems have no een solved. Especially, he energy consumpion of he wireless sensor Manuscrip received July.0, 0; revised Aug., 0; acceped Sep. 0, 0. nework prolems, which seriously resric working performance and service life for S, and have impac on pracical and efficiency of he wireless sensor nework. In order o improve energy efficiency of S, solve he prolem ha resric S popularizing due o he nodes energy are limied, so i is necessary o research high energy efficiency sensor nework nodes srucure and rouing algorihm, so as o prolong he S lifeime, promoe he rapid spread of S y he exising energy saving echnology and communicaion echnology. In recen years, muli-inpu-muli-oupu (MIMO) sysems have een exensively researched due o heir ailiy o reak hrough he channel capaciy limiaion which radiional single-inpu-single-oupu () wireless communicaion sysem has [], and compared wih sysems, MIMO sysems require less energy consumpion when ransmission disances are larger han a given hreshold []. MIMO ransmission sraegy and nodes srucure of a heerogeneous S were proposed in [3], and he resuls show ha he approach is eer in energy consumpion and delay of he nework han. However, as echnique of MIMO requires complexiy ransmier circui and powerful signal processing capailiy, he realizaion of muli-anenna echnique is impracical o sensor nodes whose physical size and energy are limied. ih he increasing mauriy of MIMO echnique, virual MIMO echnique which developed on he asis of MIMO has een inensively sudied [4-8]. Virual MIMO echnique is a diversiy ransmission echnique ha is applicale o S, The sudy shows ha alhough addiional circui energy consumpion and raining overhead have een aken ino accoun, he applicaion of virual MIMO echnique will e more energy efficiency han or muli-op ransmission [4-5]. The S ransmission scheme ased on virual MIMO was proposed in [6], and proved ha he scheme can reduce he oal nework energy consumpion. However, he scheme only applied virual muli-inpu-single-oupu (MISO) echnique o he daacenric clusering sag, and increased he energy consumpion for choosing cooperaive cluser and coding complexiy. According o he Isomorphism characerisics of S, Comined a alanced energy load LEACH [9] algorihm wih virual MIMO echnology and Virual MIMO doi:0.4304/jnw

2 0 JOURAL OF ETORKS, VOL. 7, O., DECEMBER 0 Clusering Algorihm (VMC) is proposed. In order o fully reduce nework energy consumpion, virual MIMO echnique is applied o all nework nodes y VMC algorihm, and he energy load is evenly disriued o each sensor node y roaion elecion of virual MIMO cluser head nodes, so ha he energy consumpion of cluser head nodes can deplee rapidly and he nework lifeime can e effecively prolonged. e analyzed he energy consumpion of differen nework size, node densiy and pah loss facor, and he nework parameers o opimize he nework lifeime are oained y Geneic Algorihm (GA), Simulaion and experimen show ha compared wih LEACH VMC proocol can effecively reduce energy consumpion and prolong neworks lifeime. II. VIRTUAL MIMO S MODEL A. Virual MIMO Technique In recen years, wih furher sudy of MIMO wireless communicaion sysem, a Virual MIMO communicaion model has formed []. amely, In a Virual MIMO communicaion sysem, each moile erminal no only has a single anenna, u also has one or more parners for cooperaion ransmission, and no only ransmi is own informaion u also ransmi is parners' informaion. As shown in Figure, he Virual MIMO ransceivers are composed of cooperaive moile erminals, which form a virual muli-anenna srucure. Each erminal use oh heir own and is parners channel space o ransmi informaion joinly, so more spaial diversiy gain is oained and single-anenna moile erminals spaial diversiy is implemened. I is proved ha he virual MIMO communicaion sysem can increase channel capaciy, improve nework service qualiy and sysem performance. B. ework Srucure Due o he resricions of qualiy, volume and power consumpion, i is very difficul o implemen mulianenna srucure for S nodes. However, a single anenna node can ge he muli-anenna spaial diversiy gain y using virual MIMO. Assuming ha each sending node has a se of daa o e sen o he receiving node, firs, In he local communicaion process, he sender nodes roadcas is own informaion o is parners y TDMA, so ha he sending erminal have all parners' daa; Second, In he long range communicaion process, a single node is seen as a anenna of virual anenna array, he encoded daa is ransmied o he receiving erminal in parallel y he sending erminal. Based on he srucure of virual MIMO communicaion, VMC algorihm of he homogeneous S nework is proposed. Every M nodes form a virual MIMO ransceiver, and nework erminals are divided ino several clusers, each cluser has a virual MIMO cluser head which consis of M nodes. Finally, he local clusers erminal informaion which has colleced is ransmied o S (Sink ode) y VCH (VCH, virual MIMO Cluser Head) In order o faciliae he research and discussion, we make he following assumpions:. Each sensor node has a dual work which is and virual MIMO, and according o communicaion requiremens, he wo communicaion mehods can e convered each oher.. Virual MIMO sysems are coded wih Space- Time Block Code (STBC) which has lower decoding complexiy han Space-Time Trellis Code (STTC). 3. The nodes have compleed locaion and seleced heir collaoraion nodes efore he neworks are formed, and a virual MIMO ransceiver consiss of M neighor nodes. 4. M nodes of virual MIMO ransceiver can provide synchronous ransmission when daa is sen and received. 5. S ha has muliple sending and receiving anennas is no limied o aspecs of any energy and volume, which can implemen he MIMO communicaions. C. VMC Algorihm VMC scheme is ased on cluser-heads-randomselecing echnology hrough revolving pu forword y LEACH. I also comines is echnology wih ha of virual MIMO, and akes he characerisics of isomorphism S ino consideraion. Thus, i can e proposed as follows.. Iniializaion. S roadcas AOS (Adverisemen of sink) which is iniializaion informaion o all nodes in S region, and he node operaing parameers are iniialized.. The selecion of VCH. Assuming ha CH is he proporion of cluser-heads (he raio of he numer of VCH required and he oal numer of nodes), so he R round Cluser-Head selecion hreshold is CH T = ρ ρch * ( rmod(/ ρch )). () M here, mod is he seeking modulo. Random numer ha is generaed y each node is eween 0 and. If p T, his node is chosen as he cluser head node, and send informaion of cooperaion o is parners nodes o consiue he VCH. 3. Cluser. Informaion AOCH (adverisemen of cluser head) ha VCH have ecome cluser head is roadcas o all nodes in he region, and he ID informaion of VCH has also een included. VS (virual MIMO sensor nodes) are virual MIMO erminals excep VCH, which judge he signal srengh of AOCH ha is roadcased y differen VCH, and send REG (regisraion) informaion o he VCH ha have greaes signal srengh, hen, join in heir domain. 4. Time-slo allocaion. TDMA slo is produced y VCH, which is sen o is cluser memers, and each VS is assigned a ime-slo. 5. Transmission in cluser. Collaoraion nodes of VS exchange daa y he way of. According o ρ

3 JOURAL OF ETORKS, VOL. 7, O., DECEMBER 0 03 he TDMA ime-slo allocaed y cluser head, daa of M nodes are sen o VCH y virual MIMO mode. Then VS reurn o he sleep sae o save energy unil he sep is acivaed again. 6. Convergence. VCH ransmi he daa of all he cluser nodes o S, hen, reurn o Sep. III. EERGY COSUMPTIO AALYSIS OF VIRTUAL MIMO AD A. Energy Consumpion Model of odes The energy consumpion ha sen per i daa y node, which can e descried y he following expressed [5]: ( + ) PT on + ( Pc + Pde ecor) Ton + PsynTr E = α L () Psyn here, P de ecor, and Pc are he energy consumpion of frequency synhesizer, deecors and oher pars of he circui respecively. The ime required o send L is of daa is given y Ton = L/ R, and he i-rae R is given y R = B, where is he numer of is ransmied i per second per Hz andwidh and B is he modulaion andwidh. T r is he ransiional period ha node conver from he sleeping mode o working mode. α is raio ha deermined y he modulaion and. The ransmiing power P can e expressed as a funcion of he communicaion disance d and pah loss facor n []: (4 π ) P dn= ER dm (3) n (, ) GG rλ l f here G and Gr are he anenna gain of sending erminal and receiving erminal respecively. λ is he carrier wavelengh, M l is he link compensaion f coefficien of aenuaion, is he receiver noise figure. E is per i energy consumpion for he receiving erminal in a cerain i error rae, while he value is decided y he modulaion. For local communicaions, BPSK(muliple phase shif keying) modulaion is adoped and = ; for MIMO communicaions, if BPSK modulaion is adoped, compared wih he, he energy efficiency have no advanage, so MQAM (M-order quadraure ampliude modulaion) modulaion is adoped and = [4]. Assuming ha Channel model is fla Rayleigh fading channel, he per i energy consumpion of and MIMO have he following relaions []: E MIMO E M M (4) P 0 / M P ( ) M (5) M e can derive from () - (5) ha he energy consumpion for each i under and MIMO ransmission model is as follows: M (4 π ) E d n = + M + 0 n (, ) ( α ) d / M P GG rλ f ( P + P ) R + c de ecor, siso PsynT L r M MIMO P E ( d, n) = ( + α )( ) M M (4 π ) P n s ynt d M ( f + Pc + Pde ecor ) R, mimo + G G λ L r B. Calculaion of he ework Energy Consumpion In he previous secion, a single node energy consumpion of wo differen modes is analyzed. In his secion, he overall energy is considered from sending fixed daa view poin. Assuming ha each node send L is daa, he energy consumpion of virual MIMO and are as follows: M M MIMO V MIMO k =, i + k i= i= E ( D, d, n) L E ( D, n) L E ( d, n) (8) E ( Dn, ) = LE ( Dn, ) (9) Virual MIMO nodes need o complee he local d communicaion ha disance is k, and exchange he informaion of M collaoraive nodes of sender efore sending daa. Therefore, when he energy consumpion of virual MIMO communicaion is calculaed, we consider he energy consumpion no only he long range communicaion which disance is D, u also local communicaion eween collaoraion nodes. IV. DESIG OF LO EERGY COSUMPTIO S ODES I GREEHOUSE Low power hardware design is he key o save energy in wireless sensor nework sysem, he energy consumpion characerisics of chip is a key research ojec. In general, he supply volage of nodes are essenially consans, he curren of nodes are key facors in power consumpion sudy. Therefore, considering energy saving and compaiiliy, he design of nodes is pu forward ased on MSP430 and CC40. ireless sensor nework sysems are composed of sensor nodes and sink nodes. Sensor nodes are mainly composed of he processor module, wireless communicaion module, r (6) (7)

4 04 JOURAL OF ETORKS, VOL. 7, O., DECEMBER 0 power supply module, sensor module and locaion seing swich ha se is physical locaion informaion. Sink nodes are mainly composed of a processor module, wireless communicaion module, a coninuous power supply module and serial communicaion module. A The Sensor ode Module Design Figure. Sensor node srucure char Sensor module is responsile for monioring regional informaion acquisiion and daa conversion, according o applicaion requiremens, emperaure sensor, humidiy sensor, a ligh sensor and a caron dioxide concenraion sensor ec can e chosen. Processor module is responsile for conrolling sensor nodes operaion, sorage and processing daa which colleced y he nodes and forwarded y oher nodes. ireless communicaion module is responsile for wireless communicaion, exchanging conrol informaion and ransceiver acquisiion daa eween his node and oher nodes. Posiion seing swich is used o se a sensor node specific physical locaion in greenhouses. Energy supply module can provide energy which he work need for sensor node, in he paper, we adop micro aery. Sensor node is shown in Figure. B Sink ode Module Design Figure.Sink node srucure char Sink nodes mainly complee he sensor nodes daa gahering and fusion wihin communicaion nework, and realize ascending and descending communicaion proocol conversion. I released monioring ask of managemen nodes, and he daa colleced is forwarded o he exernal nework hrough a serial por. I is no only an enhanced sensor node, u also special gaeway device which hasn monioring funcion and only has wireless communicaion inerface. Is srucure is shown in Figure.. I composes of a power supply module, sorage module, processor module, node communicaion module and serial inerface communicaion module and so on. V. EXPERIMET AALYSIS AD SIMULATIO A. The Disance Relaionship Experimen of Transmi Power and Communicaion On he asis of ransmission and recepion signals ha heir srenghs are known, he pah loss exponen of signals can e calculaed y channel ransmission model during he propagaion [4]. In his aricle, when he virual MIMO clusering algorihm is implemened, sensor nodes can judge disance wih cluser head nodes according o he received signal srengh, and he ransmier power ha communicae he cluser heads is deermined, hen realize he nework layer dynamic power managemen. In order o research he relaionship of ransmission power and RSSI, we done he RSSI deecion experimens on he experimenal plaform ha was designed. Figure 3 The relaionship diagram of ransmi disance and RSSI in differen ransmi power In greenhouse canopy, a pairs of sensor nodes ha are wih ransceiver CC40, sink nodes change ransmission power y sending commands, he frequency of each power levels is ha send a daa frame in one second, and send 50 imes in differen disance. The RSSI of daa frame is couned in sink nodes, and ge a se of daa. The relaionship diagram of ransmi disance and RSSI in differen ransmi power is shown in Figure.3. I is shown ha RSSI decreased significanly when he communicaion disance is increased even if he same ransmi power, and indicae ha he large ransmiing power is eneficial o improve he signal srengh, increase he nework conneciviy. Bu in he same disance, he reducion of ransmission power would make RSSI value decreased oo, while RSSI decreases o a cerain value (aou 47dBm), he communicaion is even inerruped, he sensor nodes can no receive he daa ha sink nodes ransmi. So in he nework communicaion, he RSSI mus e in aove - 47dBm o mainain he nework conneciviy degree. Figure 4 is he ransmission power when RSSI is - 47dBm under differen communicaion disance, he

5 JOURAL OF ETORKS, VOL. 7, O., DECEMBER 0 05 ransmission power can e seen as power o mainain nework minimum conneciviy. wo adjacen sensor nodes form a virual MIMO erminal and S is MIMO ransceiver of wo anennas. Oher experimenal parameers are shown in Tale. TABLE EXPERIMETAL PARAMETERS Figure 4 he minimum ransmission power under Differen ransmission disance Based on he analysis aove, he relaion of he receiver RSSI and he communicaion disance is oained in a ransi power. e ake RSSI deecion resuls as example when he ransmi power is 5dBm, RSSI fiing curve is shown in Figure 5 when n is.6, a his ime, he value of RSSI is he mos close o he acual measuremen value. The node ransmied power of and MIMO can e oained y aking equaion (4) and (5) ino (3). Figure 6 is he relaions of nodes disance and ransmission power when n =.6, a his ime,he ransmission power of is he mos close o he noise of he greenhouse environmen. Apparenly, in he same communicaion disance, he node ransmission power MIMO is far less han ha of, u we only consider he energy consumpion of he wireless ransmiing module, don consider oher circui module power consumpion. Figure 5 The relaion fiing curve of communicaion disance and he RSSI B. The Energy Consumpion Analysis Virual MIMO and To analyze he energy efficiency of and virual MIMO, he experimenal environmen is esalished as follows. Assuming ha sensor nodes are randomly disriued in he nework domain, nodes densiy is ρ, nodes disriue in he fashion of wo-dimensional Poisson, The communicaion disance eween collaoraion nodes is he mean of d k, which can e calculaed y node densiy []. (0) E [ d ] = d = k k 4 ρ Due o STBC encoding, he complexiy of MIMO receiver increases linearly wih he numer of anennas. Therefore, in order o simplify he model, we assume ha he virual MIMO sysem is a sysem, which every Figure 6 Transmission power and he disance relaionship of nodes The energy consumpion of he node mainly composes of wo pars, namely, one is energy consumpion of RF modules, and oher is energy consumpion of oher circui modules. Figure 7 is energy consumpion of RF modules which each i daa is ransmied in difference disance, oviously, he node energy consumpion of MIMO is less han ha of. Bu a node includes processors, AD converer and oher circui modules, so he circui module power consumpion should e aken ino consideraion.

6 06 JOURAL OF ETORKS, VOL. 7, O., DECEMBER 0 Figure 7 RF module energy consumpion of each i daa ransmied Figure 8 gives he oal energy consumpion of each i daa ransmied in differen ransmission disance, i can e seen from i ha he has more advanage in energy saving when he ransmission disance is less han 3m, u he MIMO has more advanage in energy-saving when he launch disance is more han 3m. And wih he increase of he ransmission disance, he MIMO has more ovious advanage in energy saving. To sudy he energy-saving raio (defined S I S O M I M O as ) of virual MIMO and, when n =.0, and V MIMO are compared under he differen nodes disriuion densiy in Figure 9. Figure 9 shows ha he energy-saving raio of virual MIMO and is geing higher wih R increasing when he node densiy disriuion is no changed. This is due o he virual MIMO have higher energy efficiency han in long ransmission disance. A he same ime, he node disriuion densiy also impacs he energy efficiency of virual MIMO, hen R is no changed, he greaer he node disriuion densiy, he higher energy efficiency of he virual MIMO. hen ρ = 0.00 and R > 40, he energy efficiency of virual MIMO is higher han, and when ρ = 0.05 and R >, he energy efficiency of virual MIMO is higher han. This is due o he energy consumpion of local communicaion in virual MIMO is aken ino accoun, Therefore, he smaller he node disriuion densiy, he longer disance eween he collaoraion nodes and he greaer he energy consumpion of he local communicaion. Figure 8 The oal energy consumpion of each i daa ransmied The aove is energy consumpion analysis of single node in and MIMO way, and he overall energy consumpion ha is considered from sending fixed daa angle is as follows. Based on he node disriuion model esalished in he previous secion, in his secion, we simulae he daa acquisiion process of a cluser, and compare he energy consumpion of virual MIMO and. Assuming ha he region's shape is a round which radius is R, and he cluser head is in he cener, while each sensor node sends L is daa o he cluser head. On he asis of he analysis of energy consumpion in he previous secion, he oal energy consumpion can e calculaed as follows: = E () V MIMO V MIMO, i i = = E () i =, i Figure9. The relaionship eween he virual MIMO energy-saving raion and R Pah loss is he aenuaion of power densiy when he elecromagneic wave spread y space, i is an imporan facor for analyzing and designing of elecommunicaion sysems link udge. The propagaion environmen, he ransmi media, he ransmi disance, he anenna heigh and locaion have influenced on he pah loss. n represens he pah loss index, which is usually in he range of -4. Propagaion model is under an ideal freespace communicaion when n =, and i is in he case of earh model for he plane when n = 4. In Figure 9, he pah loss facor is, Figure 0 show he relaionship eween energy-saving raio and n. Energy consumpion of virual MIMO and will increase when he channel is no ideal, However, he igger n, he higher energy-saving raio, and he energy efficiency raio is more han 80% when n > 3. I is proved ha virual MIMO have eer energy-saving performance under higher pah loss.

7 JOURAL OF ETORKS, VOL. 7, O., DECEMBER 0 07 soluion of ρ CH of ρ, as is shown in Tale. TABLE was oained in several differen values OPTIMIZATIO RESULTS OF ρ CH D. Validaion of VMC Algorihm Figure.0. Relaionship eween he virual MIMO energy-saving raio and n C. Sysem Opimizaion The numer of clusers is a key issue wheher energy efficien of he clusering algorihm can e implemened under given nework condiions. For energy-consrained S, he inappropriae numer of clusers will increase he energy consumpion, and leading o rapid deah of S. If he clusers are oo lile, he numer of memers in a cluser will increase, so he cluser-heads ear oo heavy load of sending and receiving, Cluserheads energy consumpion o accelerae, and energy consumpion is imalance. On he oher hand, if he cluser-heads are oo many, i will e excessive energy consumpion for forming clusers, and he nework lifeime will e shorened. Therefore, in his secion, he opimizaion prolem of he numer of cluser head is discussed o alance he node energy consumpion and prolong he lifeime of S. GA (geneic algorihm, GA) is an ideal algorihm o opimize he sysem parameers. Is principle is ha use he effecive par of pas searched informaion o do copying, crossover and muaion on populaion which a group of individual formed, so GA has srong gloal search ailiy. In order o find opimal raio of cluser head when he nework s lifeime is longes, he GA oolox ha he Sheffield Universiy inroduce is used o oain he opimal value of ρ CH when VMC is operaed in differen disriuion densiy of nodes. For example, nodes densiy is ρ and i is disriued randomly in a square area whose edge s lengh is 00m. The lifeime of nework is aken as he opimize arge, (he lifeime is defined as he period ha is from he eginning o half of nodes which have died) and ρch is searched eween % and 0%. Calculaion is ased on inary encoding. The numer of individuals of he populaion is 0, he lengh of each populaion is 0, and he maximum herediary generaion is 5. The individual of nex generaion is chosen y he random raversal sampling whose generaion gap is 0.9. The reorganizaion is wo-poin crossover whose proailiy is 0.7. The muaion proailiy of each elemen in he chromosome is 0.5. Afer 5 ime is calculaed y GA, he opimal LEACH is a classic clusering rouing algorihm, which can effecively improve he energy efficiency of nework. VMC scheme apply virual MIMO o S, and cluser head is seleced randomly and cyclically y LEACH, So VCM can furher improve energy efficiency of nework and prolong he nework lifeime. To compare he deah round of he firs node and nework lifeime in VMC algorihm wih hose in LEACH algorihm, he nodes whose densiy ρ is 0.0 are disriued randomly in a square area, whose edge s lengh is 00m, and hey have he same iniial energy, pah loss facor n is.0. hen all communicaion parameers which are shown in Tale and nework condiions are he same, he wo communicaion scheme are run under he differen values of ρch respecively, hen he deah rounds of he firs node and he nework lifeime are oained, as shown in Tale.3. TABLE 3 COMPARATIVE AALYSIS OF LEACH AD VMC ALGORITHM Tale 3 shows ha compared wih LEACH VMC can significanly improve he nework lifeime. The deah round of he firs node of VMC delay 4-6 imes han LEACH, and nework lifeime is aou -3 imes ha of he LEACH. V. COCLUSIO In his paper, we proposed he energy efficien VMC scheme which is applied o isomorphism S. In order o oain he mos opimal raio of he cluser head which have longes lifeime under cerain nodes densiy of he nework, and geneic algorihm is inroduced, he simulaion and experimen show ha he VMC algorihm can prolong he lifeime of nework compared wih he LEACH algorihm. Las, we analyze he evoluion of he energy consumpion raio of Virual MIMO and in differen cluser size, node densiy and pah loss facor. I is proved ha virual MIMO neworks have higher energy

8 08 JOURAL OF ETORKS, VOL. 7, O., DECEMBER 0 efficiency han nework in he appropriae nework parameers. ACKOLEDGMET This work was suppored y he aional Science Foundaion of China (6050), Graduae Suden s Research and Innovaion Plan of Jiangsu Higher Learning (Gran o. CXZZ_0574). The auhors would like o express heir graiude o he anonymous reviewers for heir invaluale commens on he original and revised versions of his manuscrip, a he same ime, graefully acknowledge useful discussions wih Rongiao Zhang and Zhuo Shen on he opic of his paper, and doing a lo of work for his paper. REFERECES [] Paulraj A, aar R, Gore D. Inroducion o space-ime wireless communicaions [M]. s. U.K.: Camridge Univ. Press, 003: 7-9. [] Shuguang Cui, Goldsmih A J, Bahai A. Energyefficiency of MIMO and cooperaive MIMO echniques in sensor neworks [J]. IEEE Journal on Seleced Areas in Communicaions, 004, (6): [3] Zhao Baohua, Li Jing, Zhang ei, Qu Yugui. MIMOased energy-efficien wireless sensor neworks [J]. Aca Elecronica Sinica, 006, 34(8): [4] Jayaweera S K. Virual MIMO-ased cooperaive communicaion for energy-consrained wireless sensor neworks [J]. IEEE Trans. on ireless communicaions, 006, 5(5): [5] Bravos G, Kanaas A G. Energy efficiency comparison of MIMO-ased and mulihop sensor neworks [J]. EURASIP Journal on ireless Communicaions and eworking, 008. [6] Zhang Yu, Cai Yueming, Chen Xianming, Xie ei, Xu Youyun. A novel A novel cooperaive MIMO-ased ransmission scheme for wireless sensor neworks [J]. Chinese High Technology Leers, 008, 8(): [7] Sadek A K, Su F, Liu K J R. Mulinode cooperaive communicaions in wireless neworks [J]. IEEE Trans on Signal Processing, 007, 55(): [8] Jayaweera S K, Cheolu M L, Donapai R K. Signalprocessing-aided disriued compression in virual MIMO-Based wireless sensor neworks [J]. IEEE Trans. on Vehicular Technology, 007, 56(5): [9] Heinzelman B, Chandrakasan A P, Balakrishnan H. An applicaion-specific proocol archiecure for wireless microsensor neworks[j]. IEEE Trans. on ireless communicaions, 00, (4): [0] Foschini G J. Layered space-ime archiecure for wireless communicaion in a fading environmen when using muli-elemen anennas[j]. Bell Las Technical Jounal, 996(8): [] Proakis J G. Digial Communicaions[M]. 4h ed. ew York: McGrawHill,000: [] ang X F, Chen G R. Complex neworks: small-orld, Scale-Free, and Beyond[J]. IEEE Circuis and Sysems Magazine, 003, 3() : 6-0. [3] Baoqiang, Kan; Zhao, Lei; Fan, Jh; ang, Jy. Opimal design of virual MIMO for S performance improvemen[j]. SEAS Transacions on Communicaions, v0, n4, p9-35, April 0. [4] Purohi, eeesh; Verma, Ahinav; Agrawal, Himanshu. Performance analysis of wireless sensor nework wih virual MIMO[C]. Proceedings of he 00 Annual IEEE India Conference: Green Enery, Compuing and Communicaioon, IDICO 00. [5] Hai-Ying Zhou, Dan-Yan Luo, Yan Gao, De-Cheng Zuo. Modeling of ode Energy Consumpion for ireless Sensor eworks[j]. ireless Sensor ework, 0,3 : 8-3. [6] Chompunu Janarasorn, Chuima Prommak. Minimizing Energy Consumpion in ireless Sensor eworks using Binary Ineger Linear Programming[J]. Inernaional Journal of Compuer and Communicaion Engineering, 0, 6 :0-4. [7] Monica, Ajay K Sharma. Comparaive Sudy of Energy Consumpion for ireless Sensor eworks ased on Random and Grid Deploymen Sraegies[J]. Inernaional Journal of Compuer Applicaions, 00,6(): 8-35 Yongxian Song was orn in xuzhou, on April, 975. He received he B.S. degree in Applied Elecronic Technology from Huaihai Insiue of Technology, Lianyungang, China, in 997, and he M.S degree in Conrol Theory and Conrol Engineering from Jiangsu universiy, Zhenjiang, China, in 006. From 009 o now, He is sudying for Ph. D degree in Conrol Theory and Conrol Engineering from Jiangsu Universiy, Zhenjiang, China. Since 006, he has een a eacher in Huaihai Insiue of Technology, Lianyungang, China. His curren research ineress include signal processing, inelligen conrol, and indusrial conrol. Rongiao Zhang, male, docor, Professor and docoral uor, was orn in Haimen ciy, Jiangsu Province in 957. ow he is conrol heory and conrol engineering docoral sujec leader of Jiangsu Universiy, depuy direcor of he Insiue of auomaion. He is mainly engaged in wireless sensor nework and auomaic deecion echnology, inelligen insrumen and informaion processing echnology, faul diagnosis and sysem reliailiy and oher aspecs of he research work. Zhuo Shen was orn in 986, she received he B.S. degree from Jiangsu universiy, Zhenjiang, China, in 006, and he M.S degree in Conrol Theory and Conrol Engineering from Jiangsu Universiy, Zhenjiang, China, in 00.