The Energy Consumption Research of Sensor Network Nodes Based on Economic Theory

Transcription

1 Sensors & ransducers 2013 by IFSA he Energy Consumpton esearch of Sensor Network Nodes Based on Economc heory Xn LIU Economcs and Management School of Wuhan Unversty oom 707, No. 78 Dong Ge oad. Nannng, Guang X, Chna el.: , , fax: E-mal: eceved: 2 November 2013 /Accepted: 22 November 2013 /Publshed: 30 December 2013 Abstract: he core of ths paper s to mprove network energy effcency effectvely, to reduce energy consumpton of nodes, and to extend network lfetme, surroundng the subject of takng energy effectveness and proportonalty of wreless sensor network nto account n routng, adaptng neural networked mathematcal model as the analyss tool based on economc theory, ths paper premses on studyng system archtecture of sensor network, dssectng the hardware structure and desgn of sensor nodes, the man algorthm and protocol of each level, common used sensor networked energy consumpton technologes and strateges one by one, and analyzng the man cause of energy consumpton from hardware and protocol stack these two levels. Copyrght 2013 IFSA. Keywords: Wreless sensor network, Economc theory, Energy consumpton of node. 1. Introducton Sensor network s the one based on applcaton, compares wth tradtonal wreless communcaton network, t has the characterstcs of bg node scale, multhop self- organzaton, unattended operaton, and no communcaton nfrastructure, whle energy constrant from begnnng to end s the bottleneck of lmtng the developng and usng of sensor network technology. Sensor network s a very complcated network structure, whch has lots of smlartes wth human socety economy system, thus ths paper solved several problems of energy consumpton of node n sensor network wth the help of economc theory [1]. hs paper ams to the man factors whch nfluence energy consumpton, analyzed the massve structure, node structure, communcaton mode and network coverage of sensor network, ths paper gves the mathematcal model of energy consumpton of wreless transmsson based on economc theory. On the bass of the summary of key energy consumpton technology of sensor networks such as sngle node energy consumpton technology, data fuson technology and cross-layer energy consumpton technology, focused and compared several types of classcal meda access controls, protocols n sensor network; analyzed resolved problems n energy consumpton algorthms, protocols and strategc functons of wreless sensor network at present. 2. Energy Consumpton Analyss of Wreless Sensor and Sensor Node 2.1. Energy Consumpton Analyss of Wreless Sensor and Sensor Node All levels of wreless sensor network are the concrete realzaton based on network archtecture, because the self-characterstc of sensor network, ts Artcle number P_

2 concerned ssues are dfferent from the tradtonal wreless communcaton network. he layered archtecture of wreless sensor network s as Fg. 1. whch ncludes postonng unt, movng unt, communcaton unt and power supply unt, we can choose and confg. hese three unts accordng to the dfferent applcaton scenaros.its structure chart are showed as Fg. 3. Fg. 1. Layered archtecture of wreless sensor network. he concrete realzed functons s referred to the sensng, processng, analyzng, comparson and data transmsson facng to sensor nodes, specfc applcaton s the remote operaton and mantan sensor network facng to user termnal, the most mportant feature of sensor network s energy constrant [2]. herefore, the object of the protocol based on tradtonal wreless sensor network s to obtan hgh QOS, whle the object of the protocol based on sensor network s to reduce energy consumpton Wreless Sensor Network Structure Wreless sensor network structure s showed as Fg. 2, the entre network consttutes of sensng node, host node or aggregaton node, base staton controller, base staton transmttng. Fg. 3. Structure chart of sensor node Sensor Node Communcaton Mode he communcaton unt n sensor node s transcever module, whch can adapts actve or passve lght wave module, lookng from the applcaton of wreless sensor angle, data transmsson manly dvdes nto contnuous sendng type, event drven type, query drven type and mxed type [3]. Lookng from the node data communcaton of sensor drecton, can dvde nto Node-BS and BSC-Node two types Energy Consumpton Analyss Based on Hardware Nodes he energy consumpton of sensor node s hgher than the energy consumpton of sensng, processng, space and dormancy, as Fg. 4 shows: Fg. 2. Structure dagram of wreless sensor network Sensor Node Structure Sensor node manly conssts of sensor unt, processor unt, communcaton unt and power supply unt these four elementary unt, wthn the dashed box s the auxlary unt of sensor node, Fg. 4. Energy consumpton dstrbuton dagram of communcaton unt. otal energy consumpton formula of computer transcever module n wreless communcaton: PC N [ P ( on st ) Pout ( on )] N[ P ( on st )] (1) 286

3 where P and P are the power dsspaton of transmsson and sendng respectvely, P out represents the output power of transcever, on and on are the onlne duraton of transmsson and sendng, st and st are the startng tme of transmsson and sendng respectvely, N and N are the converson tmes of transmsson and sendng wthn unt nterval, whch s determned by task management strategy and MAC, n the newest wreless transmsson transcever wth l. we energy consumpton at present, P s about 20 dbm, P out approaches to 0 dbm. Energy consumpton module of sngle node wreless communcaton shows as Fg. 5, n the condton of sngle hop, no data fuson and only smply receve and transmt, here m=n, the energy consumpton E of each bt n data transmsson s : b Eb Etx Erx, (2) E E E, (3) tx te td where Etx, E rx s the energy consumpton of recevng and sendng one bt of data, E tx s the energy consumpton of node sendng module successfully sends a data, E rx s he energy consumpton of node recevng module successfully receves a data, E tα s the energy consumpton of successfully sendng a data for one meter, s the physcal dstance from node sendng module to another node recevng module. s the energyconsumpton path, whch s a constant, and connected to geography and transmsson condton. Fg. 5. Energy consumpton module of wreless communcaton n sensor node. 3. Wreless Sensor Data Fuson Module Based on Economc heory and Artfcal Neural Network 2.1. Data Fuson he study of applyng data fuson nto wreless sensor network s properly extensve, especally the routng algorthm and protocol n network layer datacentered, data fuson s an nterdscplnary comprehensve theory and method, t can use several technologes and methods of other felds for reference [5] Economc heory he dstngushng features of the applcaton of economc theory and wreless sensor are: followng the hgh qualty and favorable prncple of resources, encouragng the nodes wth resources can put more better resources to network, and ganng proft from provdng servces; allowng the tasks of users to fnd approprate resources accordng to QoS requrement, whch not only meet the demands of users, but also reduce the wastng of resources as much as possble, f the prce of resources are determned accordng to the nternal performances, then the matchng resources are nearly the optmal soluton found n resource management, through economc ways to process a unform processng to all the resources, the transacton goods can nclude computng power, nternal memory, storage capacty, network bandwdth, data, devces and nstruments and so on Artfcal Neural Network Artfcal neural network has the features of dstrbuted and parallel processng, non-lnear mappng, adaptve learnng, strong robustness and fault tolerance, whch makes ts neural computaton got wdely values n data fuson feld. he basc prncples of artfcal neural network are smlarty to the processng of human bran syntheszes data or nformaton, n the routng wthout usng neural network fuson technology, and between the sensor nodes n network exsts much redundancy nformaton transmsson, whch ncreases the workng strength, processng tme and communcaton frequency of data. After ncreasng the startng energy consumpton, the energy consumpton E of data transmsson s: E E E b E E E l dec st sr b tx rx, (4) 2.4. outng Algorthm of Sensor Network outng algorthm of sensor network can dvde nto plane formula routng, herarchcal formula routng and locaton-based routng accordng to topologcal structure. Amng to the routng 287

4 algorthm of sensor network, two types of basc protocols were proposed: drect communcaton protocol and mnmum transmsson energy protocol, the basc block dagram of ts network topology structure based on SOMDF algorthm s showed as Fg. 6: samples, when sometme happens n surveyed area, the source node v of detecton event random chooses a neghborng node transmsson data package,after neghborng nodes recevng source node data, through every sensor node of package should detect the qualty of surroundng neghbor nodes, ts structure s shown as Fg. 8. duty cycle(t) error rate(t) swtch(t) resdual power(t) Fg. 8. Network structure sketch map of SOMDF algorthm. Fg. 6. he basc block dagram of network topology structure Mathematcal Model of SOM Network Defnng four nputs n SOMDF algorthm are; [ - (), - (), X t duty cycle t error rate t, swtch(), t resdual- power()] t (5) SOM s a knd of unsupervsed clusterng neural network wth crossrange assocatve ablty, ts man functon s mappng the n-dmensonal data to a lower dmensonal nput [7]. SOM model mtates the structure of neuron two dmensonal space lattce structurally and mtates clusterng, fuson functon, self-organzaton and self-learnng functon of bran s nformaton processng through the nteracton and competton between neuron n network, ts pe graph s showed as Fg. 7. Where duty-cycle (t) represents duty cycle, errorrate (t) represents the error rate of node transmsson. In SOMDF algorthm, adaptng mnmum Eucld dstance to fnd best matchng unt X ( ). Namely meets the followng unt X ( ) as the best matchng unt: Xt ( ( )) argmn Xt ( )- W,( j 1,2,3,4), (6) j x(t) (g) Suppose h s the topology feld of wnnng neuron centered on, d represents the lateral dstance between wnnng neuron and exctatory neuron j, the typcal functon of h chooses Gaussan functon; h 2 d exp(- ), (7) 2 2 Fg. 7. SOM pe graph Introducton of Sensor Node Neuron he basc dea of SOMDF algorthm s as follows: based on a certan number of sensor node he Gaussan functon h s shown n Fg. 9, parameter σ s the effectve breadth of topology feld. Although determnes the best matchng neuron (X) durng competton process, ts weght vectors can not fully express the vector X, n order to make weght vector express nput vector fully, then we must process self-organzaton learnng to weght, and request the synapse weght vector changes wth the nput vector X: 288

5 Fg. 9. Gaussan feld functon. wq ( ) wq ( -1) a( pq ( )-, (8) wq ( -1)) (1- a) wq ( -1) apq ( ) where area N * ( d) ncludes all the neuron subscrpts located n the wnnng neuron wth the center of *, radus of d.: N ( d) {j,d d}, (9) * 2.7. Analyss of Performance Algorthm Convergence he convergence of algorthm determnes the executonal and feasblty of algorthm, as shows n Fg. 10, wth the ncreasng of node number,the convergence speed of algorthm reduced, the prmary causes of the reducton of algorthm convergence speed are: he ncrease of the number of total nodes wll lead to the ncrease of number of neghbor node, and the curb between nodes s also ncreasng, whch leads to the update speed of weght s slow Data Fuson on Sensor Node Suppose node A, B processes data transmsson wth the way of sngle hop. hen the transport of node B can defne as: delay DelayA B node press DelayAB Data _ Fuson, (10) DelayABC Data _ Fuson E Bandwth 2( node Epress ), (11) Bandwth Suppose node A B C processes data transmsson between neghbor nodes wth the sngle hop way, suppose among node A BC... Snk processes data transmsson between neghbor nodes wth the sngle hop way, and defnes the hop count of node A... Snk s, then the transport delay of node Snk and transport delay DelayA DelayA BSC Snk of BSC can be defned as: node Epress DelayASnk Data _ Fuson, (12) Bandwth node Epress DelayAB SC Data _ Fuson ( 1) Bandwth (13) he data transmts to Snk through each node, the fuson only operated on Snk or BSC, defnton mn s the mnmum hot count of A... Snk, at ths moment the transport delay of Snk s: DelayA Bandwth Snk node DelayASnk mn, (14) where the relatonshp between network wdth and the average delay of data transmsson s showed as Fg Energy Consumpton Analyss 3.1. Performance Index of Wreless Sensor Network outng Algorthm learng effcency 0.1 learnng effcency 0.3 learng effcency 0.5 Fg. 10. he relatonshp between algorthm convergence speed and learnng effcency. It should be notced that when amng to the selfcharacterstc of sensor network to process routng algorthm desgn: (1)he energy consumpton s lmted and general wthout external energy supplement, whch requests routng protocol effcent uses energy; (2) the number of WSN node s very large, and node only can get local topologcal structure nformaton, routng protocol should choose compatble route on the bass of network local nformaton; (3) Sensor network has a strong 289

6 applcaton correlaton, the routng protocol n dfferent applcaton may have a bg dfference, thus not a unversal routng protocol meet dfferent applcaton; (4) he routng mechansm of sensor network s also often connected wth data fuson technology, through reducng total communcaton of nodes to reduce energy consumpton Data fuson on Snk or BSC Data fuson on each node Bandwth(Mbps) Fg. 11. elatonshp between network wdth and the average delay of data transmsson opologcal Structure of Sensor Network Node and Node Energy Consumpton opologcal structure of sensor network node determnes the performances of routng protocol, the node dstrbuton can be frozen-n dstrbuton, also can be self-organzaton dstrbuton. he condton of node energy consumpton nfluences the set up procedure of routng n sensor network, accordng to the applcaton scenaros of sensor network, the transmsson mode of rendezvous pont or host node can be seral transmsson, event drven transmsson, queue drven transmsson and mxed type Communcaton Energy Consumpton Analyss of SOMDF Algorthm to educe Node Adaptng SOMDF algorthm can reduce the communcaton energy consumpton of nodes effectvely, data fuson technology s used on Snk node to compress and ntegrate the data transmtted n nodes. Suppose α s the proporton of workng tme wthn one frame, P work s energy exhausts of the work, P sleep s the energy exhausts of dormancy, P dle s the energy exhausts n dle condton, then the average energy exhaust P can be defned as: Consderng: PP (1- )P P, (15) work sleep dle P αp (1-α)P, (16) work sleep he wreless communcaton energy consumpton P frame wthn one frame can be : Pdec Pst Psr P frame ( P P ) W l, tx rx b (17 ) where P tx s the energy consumpton n sendng, Prx s the energy consumpton n recevng, Pdec s the energy consumpton of decodng one bt package n decode, Pst, Psr s the energy consumpton of sendng start and recevng start, l s the effectve data bt, W bt s the bt of one frame data. herefore, the average energy consumpton P of node s : Pdec Pst Psr P α ( Ptx Prx ) Wbt (1-α)P sleep l (18) So t can be seen that after adaptng data fuson technology n Snk, because elmnatng the redundancy control bt, therefore the effectve data bts n data package s permanent, thus makng the bt of one frame data reduce obvously, f mantan the other energy consumpton of communcaton n node nvarant, then the average energy consumpton of node reduced obvously, after adaptng data fuson technology the average energy consumpton can be reduced 10 % n Snk. 4. Concluson Sensor network as a burgeonng, whch wll change the nteractve mode between human and real physcal world technology, syntheszng sensor technology, embedded computng technology, dstrbuted nformaton processng technology and communcaton technology, and plays an mportant role n mltary, ndustry, medcal treatment, transportaton and envronmental protecton, whch has a extensve applcaton prospect and huge commercal value, wreless sensor network has been the research emphass and hotspot n I feld at home and abroad. hs paper takes mprove network energy effcency effectvely, reduce node energy consumpton, extend network lfetme as the core of research work, wth the help of methods of economc theory, artfcal ntellgence technology and data fuson, set about from the perspectve of mprovng routng algorthm n network layer, carred analyss and research to the energy consumpton ssue of 290

7 sensor node based on the both energy effectveness and balance of sensor network. hs paper analyses resolve ssues of energy consumpton algorthm, protocol, strategy of wreless sensor network at present, emphaszes the routng technology of wreless sensor network, and explans some classcal algorthms systematcally, detaled contrasts and explans the nscape, applcable scenaro, request and effect of nodes, desgned a nerve network fuson algorthm SOMDF of selforganzaton mappng module based on artfcal neural network, and detected ts performances. eferences [l]. J. erry, en Emergng echnologes that wll Change the World. MI Enter Prse echnology evew, Vol. 106, Issue 1, 2003, pp [2]. A. Demers, J. Gehrke,. ajaraman, he Cougar Project: A Work- In- Progress eport. USA, Cornell Unversty, [3]. Q. N. Ln, H. ongqang, Energy consumpton of wreless sensor network nodes, Journal of Wuhan Unversty of Scence and echnology Informaton and Management Engneerng, Vol. 6, 2009, pp [4]. Q. afan, Informaton fuson: neural network-fuzzy nference theory and applcaton, Natonal Defence Industry Press, 2002, pp [5]. G. Xaoln, Q. Depe, Grd computng model study based on Internet, Journal of X an Jaotong Unversty, Vol. 35, Issue 10, 2001, pp [6]. A. Wear, Innovaton strategy of knowledge economy, Xnhua Publshng House, 1998, pp [7]. J. Schumpeter, he theory of economc development, he Commercal Press, 1991, pp Copyrght, Internatonal Frequency Sensor Assocaton (IFSA). All rghts reserved. ( 291