The road to immortal sensor nodes

Transcription

1 Uiversity of Wollogog Research Olie Uiversity of Wollogog i Dubai - Papers Uiversity of Wollogog i Dubai 008 The road to immortal sesor odes ohamed K. Watfa Uiversity of Wollogog i Dubai Haitham Al-Hassaieh America Uiversity of Beirut Samir Salme America Uiversity of Beirut Publicatio Details Watfa,., Al-Hassaieh, H. & Salme, S. 008, 'The road to immortal sesor odes', Iteratioal Coferece o Itelliget Sesors, Sesor ets ad Iformatio Processig, IEEE, IEEExplore, pp Research Olie is the ope access istitutioal repository for the Uiversity of Wollogog. For further iformatio cotact the UOW Library: research-pubs@uow.edu.au

2 The Road to Immortal Sesor odes ohamed K.Watfa Computer Sciece Departmet Uiversity of Wollogog Dubai, UAE Haitham Al-Hassaieh ad Samir Salme Computer Sciece Departmet America Uiversity of Beirut Beirut, Lebao {hza05, Abstract Oe major limitatio i WS is the lifetime of the ode s battery. The aim of this paper is to overcome the power costrait i WS by wirelessly chargig the odes usig a ewly discovered techique for sigle hop wireless eergy trasfer called Witricity. I this paper, we preset ew techiques for multi-hop wireless eergy trasfer. We specify the hardware a sesor ode eeds i order to wirelessly trasmit ad receive eergy. Fially, we preset a chargig protocol for a WS with flat topology ad aother for a WS with clustered topology. Our results show that multi-hop wireless eergy trasfer ca be doe with a efficiecy as high as 0% over 8 hops. The i this paper is the first that addresses multihop wireless eergy trasfer i sesor ets ad is the buildig block of our ad other future research i this area. Keywords Witricity; multihop wireless eergy trasfer; sesor odes. I. ITRODUCTIO Sesor ets is a growig field i computer sciece. The techological advaces i power electroics ad radio trasmissio allow us to build small ad iexpesive sesor odes which together ca form a wireless ifrastructure less et. Dispersig the odes i a certai area, allows us to accurately moitor the area. The odes ca sese a pheomeo or detect a evet ad sed data to a base statio. Sesor ets ca be used i a variety of areas. They ca be used for surveillace i warehouses, factories, ad compaies. They ca be used i the military to detect eemy ifiltratio ad i the medical field to moitor patiets coditios. They ca be also used to moitor evirometal pheomea as well like volcaoes, earthquakes, or polar meltdow. Oe of the mai cocers i wireless sesor ets is eergy costraits. A sesor ode has a limited amout of eergy stored i its battery ad oce this eergy is cosumed, the ode dies. As the odes i the et start to die, the coverage ad coectivity i the et gradually dimiish util the et fails. I most applicatios, like moitorig of volcaic areas, tropical rai forests, ad polar ice caps, it is impractical to retrieve the odes ad replace their batteries. oreover, i some applicatios sesor odes failure is ot acceptable. I medical ad military applicatios, odes failure might be fatal; resultig i the death of a patiet or a eemy ifiltratio goig uoticed. ay papers have bee published that itroduce ew eergy efficiet protocols. However, all these protocols oly exted the life of the et. However, today a breakthrough i techological developmet called Witricity [] allows us to wirelessly trasfer eergy betwee odes at mid rages (- meters). With this techology we ca overcome the power costrait i WS by costatly chargig the odes. I this paper, we use the cocept of Witricity i WS to efficietly charge the odes pavig the road to immortality. Itegratig Witricity ito WS requires cosiderig multi-hop wireless eergy trasfer sice most odes are several hops away from the base statio. We preset three ew techiques for multi-hop wireless eergy trasfer; the store ad forward techique, the direct flow techique ad the hybrid techique. As the ames suggest the store ad forward stores the eergy at every hop whereas the direct flow does t ad the hybrid is a combiatio of the two. For each of these techiques we derived the efficiecy equatio. The results for simulatig these techiques show that eergy ca be trasferred over 8 hops with a efficiecy of 0% -for sigle hop the efficiecy was 60% accordig to []. I order for a sesor ode to trasmit ad receive eergy, some hardware compoets must be added to the regular sesor ode to make it a Immortal sesor ode. The ode requires a coil, capacitace, rechargeable battery, buffers ad other compoet discussed i a later sectio. To hadle the chargig i the WS, we itroduce a ew Chargig layer betwee the et layer ad the data-lik layer. This layer is resposible for routig the eergy i the et form a source ode to the destiatio ode while maitaiig good efficiecy eergy trasfer. This layer will implemet the chargig protocols for the flat et topology ad the clustered et topology. The overhead of these algorithms is low ad they allow the chargig of all the odes i the et while achievig good efficiecy of eergy trasfer betwee the odes. Thus our cotributio i this paper is maily two-folds. First, we preset a strategy for mutlihop wireless eergy trasfer, comparig three differet schemes through derivig their respective efficiecy equatios ad deducig the optimal scheme. Secod, we propose a efficiet utilizatio of this strategy for the cases of flat ad clustered topology through a itelliget commuicatio protocol lyig o top of the chargig algorithm proposed. The rest of the paper is orgaized as follows. Sectio II summarizes some related. Sectio III itroduces the multihop eergy trasfer aalysis. Sectio IV provides the /08/$ IEEE 5 ISSIP 008

3 hardware desig details. Sectio V discusses the chargig protocol. Simulatio results are provided i Sectio VI. We coclude this paper is Sectio VII. II. RELATED WORK The bulk of our research is based o the cocept of Witricity [] which was recetly discovered ad allows us to wirelessly trasfer eergy betwee objects at mid rages. The idea behid Witricity, preseted i [] ad [], is resoat electromagetic couplig; two objects i resoace at the same frequecy will ted to couple ad eergy will be trasferred betwee them. Efficiet eergy trasfer betwee the two objects requires strog couplig; i.e. the couplig coefficietκ must be larger tha all the eergy loss rates,. The figure of merit for wireless eergy trasfer is κ fom = Witricity is practically demostrated i []. Eergy is trasferred betwee oe coil capacitively coupled to a source ad aother coil capacitively coupled to a load at a resoat frequecy ω. The experimet was performed i the presece ad absece of a blockig wall betwee the two coils. The obvious limitatios of their was the distace by which the eergy could be trasferred ad its resultig efficiecy. The equatios for ω, κ ad, are give by: ω = () LC () ω κ = () L L 4 μω 0 μ π 0 r h ω ω = + + L σ 4πa ε 0 c π c I the above equatios, L, are the iductaces of the coils, C is the capacitace, is the mutual iductace betwee the coils, μ0 is the permeability of free space, r is the radius of the coil, c is the speed of light, a is cross sectioal radius of the wire, is the umber of turs, ad h is the height of the coil. The amplitudes of the magetic fields at the source object ad receivig device are respectively A S ad A. If the load d is cosumig eergy at a rate of, the relatio betwee the amplitudes of the fields becomes: κ Ad = AS (5) d + [] ad [] show that the efficiecy of wireless eergy trasfer is give by: (4) η Ad S AS d Ad Ad = + + = d fom d It is clear from this equatio that as fom efficiecy of eergy trasfer icreases. (6) icreases, the Oce we icorporate Witricity ito WS, we will examie two types of et topologies; a flat topology ad a clustered topology. Eergy efficiet clusterig algorithms are preseted i [], [4], ad [5]. The most importat of which is LEACH (Low Eergy Adaptive Clusterig Hierarchy) protocol [5]. LEACH divides the et operatio ito rouds. Durig each roud, clusters are formed with ew cluster-heads. Rotatig the cluster-heads allows balacig the eergy betwee the odes ad icreases the et lifetime. The cocept of clusterig will aid us to desig a efficiet protocol that utilizes the idea of eergy trasfer to balace the eergy betwee a fully charged ode ad a ode that is i eed of eergy. III. ULTIHOP WIRELESS EERGY TRASFER Sigle-hop wireless eergy trasfer was preseted i [] ad []. I this paper, we will preset three ew techiques for multi-hop wireless eergy trasfer ad derive the efficiecy of eergy trasfer i each of these techiques. The aim is to trasmit eergy from a base statio or a source ode to a ode hops away. A. Store ad Forward Techique I this techique, each ode o the path from the source to the destiatio ode receives the eergy ad stores it i its battery. It the forwards the eergy to the ext ode o the path. At every hop, two odes couple together idepedet of other odes i the et. Startig from equatios (5) ad (6), we ca derive the efficiecy of store ad forward multihop wireless eergy trasfer over hops. We showed that the efficiecy over hops is just the product of efficiecies at every hop. This is highly ituitive sice the eergy trasfer at each hop takes place at a differet time ad idepedetly of the other hops. The efficiecy is give by: η = i= [] arg [] i + ch lossi charg igi [] ch arg ig[ i] fom [ i] [ i] +ch arg loss[ i] (7) 54

4 ch arg loss[ i] is the rate of eergy lost due to chargig at ode i, is the rate of eergy cosumed by chargig ch arg ig[ i] the battery at ode I, ad is the eergy loss rate at ode i. i [] I order to aalyze the efficiecy, we simplified this equatio by assumig that the odes are idetical. This is a reasoable assumptio sice the odes ca be maufactured with the same parameters ad thus the loss rates will approximately be the same. We will also icorporate the values of ch arg loss [ i ] ito. The simplified equatio becomes equatio (6) to the power : η = ch arg ig ch arg ig fom B. Direct Flow Techique This techique takes advatage of the fact that a sigle device ca couple with multiple devices at the same time. I this techique, each ode couples with previous ad followig odes o the path from the source ode to the destiatio ode. Oce this ode receives the eergy from the previous ode, it directly trasmits it to the followig ode without storig it i its battery. We do t have chargig ad dischargig losses except at the last ode. I this techique, we assume that the loss rates at the itermediate odes will ot be doubled. I the store ad forward techique, of the itermediate odes is cosidered twice; oce while receivig eergy ad oce while trasmittig eergy. I this case, the eergy is beig received ad trasmitted simultaeously, ad thus should be cosidered oly oce. We also make the assumptio that equatio (5) is still valid with beig replaced by the summatio of eergy loss rate at subsequet odes sice the eergy lost at a ode is beig used at later odes. The figure of merit i this case is the geometric mea of all the couplig coefficiets over the geometric mea of all the loss rates. It is give by: fom = + j= κ [ j] [0] [ j] j= I this case the efficiecy is more complex ad it is give by: (8) (9) η = + i i [ j] + charg ig+ [ m] i [] j=+ i m= j+ ( κ [ j] ) ch arg ig j= + fom [0] [ j] j= [ j] charg ig [ m] [0] j= m= j+ [ ] ch arg ig ch arg ig + fom [0] [ j] j= = (0) Agai, we simplified this equatio by assumig that the odes are idetical. The resultig efficiecy equatio becomes: ch arg ig j + ch arg ig fom j= () η = ch arg ig + j ( i) ch arg ig i= fom j=+ i ch arg ig Oe importat thig to otice about this equatio is that as the figure of merit icreases, the efficiecy will always icrease. C. Hybrid Techique The hybrid techique uses a combiatio of both the store ad forward techique ad the virtual circuit techique. I this techique, eergy is trasferred usig the direct flow techique for a small umber of hops ad the it is stored at the th ode. It is the forwarded to the ext th ode agai usig direct flow techique. So if we wat to trasfer eergy over hops. We ca divide the eergy trasfer over k direct flow trasmissios, each of hops with =k*. The figure of merit is give by (9) with beig replaced by. The efficiecy of the eergy trasfer will be the product of the efficiecies of the k direct flow trasmissios. For simplicity, we will preset the efficiecy of the hybrid techique for the case where all the odes are idetical. Thus, for eergy trasmissio over hops with hops direct flow trasmissio, the efficiecy is give by: ch arg ig j + ch arg ig fom j= η = ch arg ig + j ( i) ch arg ig i= fom j=+ i ch arg ig () For =, the hybrid techique is idetical to the direct flow techique. For =, the hybrid techique is idetical to the store ad forward techique. The questio remais as how to choose the value of. We will show i sectio 6 that for every value of, there is a choice of that will result i maximizig the efficiecy of eergy trasfer. There is o closed form solutio for this value, but we ca fid it umerically. 55

5 IV. HARDWARE OF THE IORTAL SESOR ODE Each sesor ode will be formed of a basic ode ad additioal hardware for wireless eergy trasfer. The sesor ode must have a coil capacitively coupled to a rechargeable battery. The rechargeable battery will act as load whe the battery is beig charged by the coil ad as a source whe the battery is dischargig i the coil. Sice multiple odes might be trasferrig eergy at the same time, they must use differet resoat frequecies as ot to iterfere with each other. Thus each ode will have a specific resoat frequecy which it will use to get charged. The chargig odes will have to tue there capacitors to get the same resoat frequecy usig equatio (). The actual buildig of the immortal sesor ode is the subject of further research. However, we preset here a prelimiary model of the immortal sesor ode show i Figure. Figure. The hardware model of a immortal sesor ode V. CHARGIG PROTOCOL We will examie multi-hop wireless eergy trasfer i WS uder two topologies; a flat topology ad a clustered topology. We will suggest a protocol for each topology. However before goig ito the protocol, a questio is raised as to whether we ca charge every ode i the et. Thus we itroduce the cocept of charge coverage. If eergy ca be delivered to the ode, the the ode is said to be charge covered. O the other had if eergy caot be delivered to a ode tha the ode is ot charge covered. Oce the et is deployed, the efficiecy of eergy trasfer will deped o mai parameters; the distace betwee the odes ad the total umber of hops of eergy trasfer. If the total umber of hops is larger tha 8, the efficiecy becomes sigificatly low. oreover, from equatio () icreasig the distace betwee the odes drastically reduces the couplig coefficiet ad cosequetly the efficiecy. Therefore, we will defie a ode a charge covered if it satisfies the followig two coditios. - The ode has at least oe eighbor who is at a distace ot greater tha.5m. - There is a ode capable of chargig this ode that is o more tha 8 hops away. The choice of.5m ad 8 hops is partially arbitrary. Oe ca choose smaller values as.m ad 6hops. However, oe caot choose higher values such as m ad 0 hops sice the eergy trasfer i that case will be totally iefficiet. Requirig the odes to be at least.5m away from each other implies that the et must be dese. Esurig that the et is charge covered will be the topic of a later research. I this paper, we will assume that all the odes are charge covered. A. Chargig Algorithm for a Flat Topology I case of flat topology, all odes are at the same hierarchy. Whe a ode s battery capacity goes below the threshold of 0%, the ode eeds to be charged ad it switches to the chargig mode. I the chargig mode, the ode stops sedig ad receivig data packets ad starts sedig chargig cotrol packets. The ode will sed a RTC (Request to charge message). The message cotais the ID of the ode issuig the request the ID ca be IP address or AC address- ad a couter as show i Fig.. The ode issuig the request will sed the RTC to all its eighbors. The RTC will the be flooded i the et. At every hop, the couter i the RTC is icremeted ad the ode ID is appeded to the RTC message. Whe a ode receives the RTC message, it will test the couter; if this couter is above 8, the RTC message will be dropped. I order to reduce the overhead of floodig, each ode floods a specific RTC oly oce ad it does ot sed the packet back to the eighbor from which the packet arrived. Oce a RTC message reaches a ode capable of chargig the requestig ode, it will sed back a ATC (Accept to charge). This ode will be called the servicig ode. The ATC message will iclude the ID of this ode, the umber of hops, ad the IDs of the odes o the path. Whe a ode receives the first ATC message, it waits for 0sec; if aother ATC message arrives the ode chooses the oe with the least umber of hops. We will show i the followig sectio that the optimal techique to use for total umber of hops -6 is the direct flow techique. For the total umber of hops 7-8, the optimal techique is a Hybrid with =4. Figure. The differet cotrol packets for flat topology chargig protocol. All packets start with the type ad the ID of the ode sedig the message. (a) Request to charge message (b) Accept to charge message (c) Cofirm to charge message (d) Start to charge message (e) o to charge message The ode the seds a CTC (Cofirm to charge) messages to all the odes o the eergy route. The CTC message 56

6 cotais the ID of the requestig ode, the ID of the iteded ode, the IDs of the two odes cosecutive to the itermediate ode from which it will receive ad sed eergy, the frequecy of eergy trasfer, ad a flag tell the ode whether to store the eergy i its battery or to directly trasfer it to the ext ode as show i Fig. 0. Whe the odes receive the CTC, they tue their capacitor to the specified frequecy, switch to chargig mode, ad sed a STC (Start to charge) message which cotais the ID of the ode to the requestig ode. Whe the requestig ode receives the STC from all the odes, they couple together ad the wireless eergy trasfer starts. Oce the battery of the chargig ode reaches a certai threshold (60%), it stops chargig ad seds a TC (o to charge) message to all the odes ivolved. The odes the switch back to regular mode of operatio. If a ode is ear to the base statio, the ode ca always charge from the base statio usig the same protocol. The oly differece i this case is that the ode seds the RTC directly to the base statio. of the servicig ode ad type which meas that the eergy trasfer will be over two hops; i.e. through the cluster head as show i Fig. b. Followig the same steps as before, the requestig ode seds a CTC message ad whe the servicig ode ad the cluster head reply with the STC message, the ode gets charged; Fig c. I case there are o odes i the cluster that are able to service the request, the cluster head charges itself first ad the charges the requestig ode (Store ad Forward). The cluster head uses the algorithm preseted for the flat et to charge itself. The differece is that cluster head seds the RTC to the other cluster heads ad ot all the odes i the et. Oce the cluster head is charged, it ca trasfer eergy to the requestig ode. B. Chargig Protocol for Clustered Topology. I this protocol, the et is divided ito clusters usig oe of the algorithms i [], [4], or [5]. We assume that the odes i the same cluster kow the distace betwee each other. The idea is for the odes of the same cluster to use a direct flow techique ad outside the cluster to use hybrid techique. Whe a ode eeds to charge its battery it will sed a RTC message to the cluster head. If the cluster head is able to charge the ode it does; else it seds the RTC to all the odes i the cluster as show i Fig. 6a. The odes capable of servicig the request will sed a ATC (Accept to charge) message. The ATC will cotai the ID of the ode servicig the request ad the distace betwee the servicig ode ad the requestig ode. At the cluster head, if there are some ATC messages with distaces o more tha.5m, all the messages havig distaces greater tha.5m are dropped. The the ode which is closest to the requestig ode is chose to service the request. The cluster head seds a ATC to the requestig ode cotaiig the ID of the servicig ode ad type which meas that the eergy trasfer will be over hop as show i Fig. b. The requestig ode seds the servicig ode a CTC message cotaiig its ID ad the resoat frequecy of eergy trasfer. Whe the servicig ode receives the CTC, it tues the capacitor to the specified frequecy, seds a STC (Start to charge) message to the requestig ode, ad switches to the chargig mode (Fig. c). Whe the requestig ode receives the STC, it switches to the chargig mode. The odes the couple together ad the wireless eergy trasfer starts as show i Fig. d. Oce the battery of the chargig ode reaches the threshold (50%), it stops chargig, switches to the wake-up mode ad seds a TC (o more chargig) message to the other ode. Whe requestig ode receives the TC message, it switches back to the wake-up mode.i case at the cluster head there were o messages received with distace less tha.5m. The the cluster head chooses the ode that is earest to it. I this case the eergy trasfer will take place over two hops through the cluster head usig the direct flow techique. The cluster head seds a ATC cotaiig the ID Figure. The chargig protocol for the clustered topology showig two cases (a) Requestig ode seds RTC to cluster head which forwards RTC to all other odes i the cluster. (b) Servicig ode seds ATC to cluster head which seds it to the requestig ode. (c) Requestig ode seds CTC to ivolved ode. The ivolved odes reply by a STC. (d) Eergy is trasferred through the cluster head i the first case ad directly i the secod case. VI. SIULATIO RESULTS To be able to simulate the efficiecy of the chargig techiques, we first exted the equatio () of the couplig coefficiet betwee two coils: π ( rr ) ω μ0 () ω κ = = D LL 8r 8r μ0 rr l l a a 57

7 Efficiecy Total umber of Hops Figure 4. The efficiecy of the multi-hop wireless eergy trasfer verses the total umber of hops. The plots correspod to the store ad forward techique, the direct flow techique, ad the hybrid techique for values of =, 4, & 6. The efficiecies are calculated for odes with coils of 4cm radii. Efficiecy % Total umber of hops Figure 5. The efficiecy of the hybrid multi-hop wireless eergy trasfer verses the total umber of hops ad the umber direct flow hops. I this equatio, D is the distace betwee the coils, r, are the radii of the coils, ad, are the umber of turs of each coils. It is clear here that icreasig the radii of the coils ad the umber of turs icreases the couplig coefficiet. O the other had, icreasig the distace betwee the coils reduces the couplig coefficiet. The couplig coefficiet is directly proportioal to the figure of merit ad the efficiecy follows the figure of merit. I [], the experimet is performed for coils with radii of 5cm each. Ufortuately, i a WS, we do ot have the luxury of usig large coils. For the values preseted i this sectio we chose the value of the radii of the coils to be 4cm ad the distace betwee the coils to be m. Usig equatios (4) ad (), we fid the values of the loss rates ad the couplig coefficiet ad the use a atlab code to calculate the efficiecy i equatios (8), (), ad () for Store ad Forward Direct Flow Hybrid (=) Hybrid (=4) Hybrid (=6) = differet values of. The efficiecy of eergy trasfer verses the total umber of hops is plotted i Fig. 4. The plot of Fig. 5 shows that the direct flow techique performs better tha the store ad forward techique for all umber of hops. The store ad forward techique reaches 0% at =5 hops whereas the direct flow techique reaches 0% at 8 hops. The performace of the Hybrid techique depeds o the choice of. For large umber of total hops, the hybrid techique seems to always perform better tha the other two techiques. For small values of, the hybrid techique performs better tha the store ad forward but ot better tha the direct flow techique. The efficiecy of the wireless eergy trasfer verses the total umber of hops ad the umber direct flow hops is plotted i Fig. 4. This graph shows that for each value of, there is a optimal value of that gives the highest efficiecy. For =7, the optimal is =4; for =0, the optimal is =5; for =, the optimal is =5. oreover, for each value of, the efficiecy of eergy trasfer with respect to is a cocave fuctio. Thus, the optimal value of is i the middle rage 4-6. For high ad low values of the efficiecy drops. At high values of (>6), oe ca see that for i the rage 4-6, the hybrid techique is still able to maitai a efficiecy over %. VII. COCLUSIOS AD FUTURE WORK I short, multi-hop wireless eergy trasfer is implemetable while keepig the efficiecy of wireless eergy trasfer acceptable. Thus, we ca use multi-hop wireless eergy trasfer i a WS to charge the odes ad icrease the efficiecy of eergy trasfer. The Chargig protocol that will be used to charge the odes will also icrease the efficiecy ad allow all odes i the et to be charged. Witricity is a ew research topic as ew ways to icrease the efficiecy of Witricity are developed; it willl become easier to implemet it i a WS ad the power costrait i WS will be overcome. Our curret research icludes developig ad testig real odes with the capability to sed ad receive eergy ad testig our speculatios ad theoretical aalysis preseted i this paper. Also, extesive simulatio results are beig developed ad the algorithms are beig refied ad would be icluded i a exteded versio of this paper. REFERECES [] Karalis A., Joaopoulos J.D., ad Soljačić. Efficiet wireless o- Aals of Physics (ov. 006). radiative mid-rage eergy trasfer [] Kurs A., Karalis A., offatt R., Joaopoulos J.D., Fisher P., Soljacˇic. Wireless Power Trasfer via Strogly Coupled agetic Resoaces Sciece 7 (007). [] Badyopadhyay S., Coyle E., A Eergy Efficiet Hierarchical Clusterig Algorithm for Wireless Sesor ets IEEE IFOCO (00). [4] Ghiasi S., Srivastava A., Yag X., Sarrafzadeh. Optimal Eergy Aware Clusterig i Sesor ets Sesors (July 00). [5] Heizelma W., Chadrakasa A., Balakrisha H. A Applicatiofor Wireless icrosesor ets Specific Protocol Architecture IEEE Trasactios o Wireless Commuicatios (OCTOBER 00). 58