Performance Study of Positioning Structures for Underwater Sensor Networks

Size: px

Start display at page:

Download "Performance Study of Positioning Structures for Underwater Sensor Networks"

Amice Fowler
5 years ago
Views:

1 PROCEEDINGS OF THE nd WORKSHOP ON POSITIONING, NAVIGATION AND COMMUNICATION (WPNC 05) & 1s ULTRA-WIDEBAND EXPERT TALK (UET'05) Performance Sudy of Posiioning Srucures for Underwaer Sensor Neworks Jose Eseban GARCIA, Kai ZIMMERMANN, Klaus JOBMANN Insiue of Communicaions Engineering (IANT), Universiy of Hannover Appelsr. 9A, 0167 Hannover, Germany Absrac - An impac sudy of he mos relevan facors which influence he performance of general posiioning srucures in underwaer sensor neworks is conduced wihin his work. A major conribuion of his work is he performance evaluaion of four differen posiioning procedures and he influences of he waer physical properies in he posiioning performance achievemens. Among ohers we invesigae he performance dependence on emperaure, saliniy, and deph. The four approaches seleced o perform his sudy are he ad-hoc posiioning srucure (AHPOS), he robus posiioning scheme (RPS), he n-hop mulilaeraion (N-HOP) algorihm and a new proposed approach known as he accurae underwaer posiioning (APOS). 1 Inroducion Nowadays he need for communicaion beween wo erminals wihouh any supporing insfrasrucure is growing up. Ad hoc neworks are able o suppor such communicaion requiremen due o he inexisence of cenral conrolling unis. Addiionally ad-hoc neworks are coninuously gaining imporance due o heir cheapness and efficiency of deploymen. The underwaer environmen offers many possibiliies for applicaions based on ad hoc neworks boh civilian as well as in miliary sysems [1]. Auonomous acing sensor neworks inspec underwaer pipelines and cables in submarine gas and oil neworks and posiioning assisance is needed for supporing underwaer surveying operaions. However here are some imporan facors o be considered e.g. coss of deploymen, node s failure due o baery depleion, or he reliabiliy of he nework iself which srongly impac he posiioning srucure performance and which should be carefully analized before deploymen. This work offers a deailed impac sudy of he mos imporan physical condiions which affec he performance of posiioning srucures in underwaer acousic sensor neworks. The performance comparison and discussion is based on our simulaion resuls. Finally we conclude his work wih our major findings regarding he deploymen of posiioning sysems for underwaer sensor neworks. Underwaer Posiioning Many ocean observing and surveillance applicaions require accurae posiion knowledge of he sensor nodes which form he nework. The nodes communicae o each oher by message exchange. Based on hese message broadcass, he neighbour nodes may use his informaion exchange in order o perform a posiion esimaion. However one major problem is he inhomogeneous propagaion characerisics of acousic signals in he underwaer channel e.g. he underwaer propagaion velociy is no consan and heavily depends on emperaure, deph, saliniy and pressure. Therefore he channel differs from node o node. 69

PROCEEDINGS OF THE nd WORKSHOP ON POSITIONING, NAVIGATION AND COMMUNICATION (WPNC 05) & 1s ULTRA-WIDEBAND EXPERT TALK (UET'05) These variaions may criically affec o he posiion esimaion because range

2 PROCEEDINGS OF THE nd WORKSHOP ON POSITIONING, NAVIGATION AND COMMUNICATION (WPNC 05) & 1s ULTRA-WIDEBAND EXPERT TALK (UET'05) These variaions may criically affec o he posiion esimaion because range measuremens are based on a good characerisaion of he underwaer acousic velociy. If hese velociy propagaion is no accuraely esimaed errors in he posiion esimaion may appear. For example a higher range deviaion leads o a less exacly posiion esimaion. Posiion esimaion can be performed according o differen algorihms and procedures. We idenify a hree main phases srucure common o each mehod. The firs phase is responsible for range deerminaion, he second esimaes he posiion. Finally he hird phase includes a refinemen procedure which inends o increase he final sysem accuracy. The combinaion of hese hree phases leads o a number of localisaion algorihms [],[],[] which are in fac he four ones ha are included in our performance sudy. Each algorihm has he abiliy of esimae he posiion of he sensor nodes in ad-hoc neworks bu according o differen algorihms. For a more deailed algorihm descripion refer o [],[],[]. Simulaion Environmen We have developed he Underwaer Posiioning (UWPOS) simulaion ool [5] which is inegraed in he OMNeT++ simulaion plaform [6]. The code is wrien using he C++ programming language. Moreover he UWPOS ool inegraes mos well known ad hoc disribued posiioning algorihms. Based on he basic package we have modified and enhanced he Posiif sofware o allow underwaer acousic propagaion and acousic signal ransmission for disance esimaion by implemening hree differen acousic propagaion models [7] well-esablished and recognised wihin he ocean research communiy. In his secion he differen blocks which compose he archiecure of he Underwaer Posiioning simulaion sofware are presened..1 The OMNeT++ Plaform The OMNeT++ simulaion environmen [6] is a discree even simulaion environmen. Is primary applicaion area is he simulaion of communicaion neworks and he simulaion of complex sysems. One major advanage of OMNeT++ is is free license for academic use. Basically a OMNeT++ complee model consiss of hierarchically conneced modules which allows he user o define and reflec he logical srucure of he sysem under consideraion while designing or analyzing differen approaches or models. The modules communicae hrough message passing which can conain differen daa ypes and which are sen and received hrough gaes and connecions. Each module can have heir own parameers which are used o define and specify he module behaviour. In fac modules a he lowes hierarchy level encapsulae he concre behaviour. These modules are called simple modules and are programmed using C++ code. The Figure 1 shows he OMNeT++ plaform archiecure and he ineracion of he differen basic componens. We idenify he execuing model under consideraion, he model componen library which conains he behaviour of he differen elemens. Also he SIM block which is responsible for managing he overall SIMulaion sysem, he ENVIR is he environmen conroller, and he CMDENV/TKENV block which defines he simulaion run eiher by command or graphically. Fig. 1 OMNeT++ Plaform Archiecure. 70

3 PROCEEDINGS OF THE nd WORKSHOP ON POSITIONING, NAVIGATION AND COMMUNICATION (WPNC 05) & 1s ULTRA-WIDEBAND EXPERT TALK (UET'05). The Posiif Package The Posiif esing framework [8] has been developed a TUDelf by K.Langendoen and N. Reijers as par of he Consensus projec [9],[10]. The Posiif package implemens he Robus posiioning (RPS) [], he Ad-hoc posiioning (AHPOS) [], and he N-hop mulilaeraion (N-HOP) [] posiioning srucures as par of is basic package. We also include he accurae underwaer posiioning (APOS) algorihm, a new posiioning srucure which has been adaped o he underwaer environmen. The communicaion wihin he nework layer suppors only localized broadcas [10]; messages are delivered direcly o a neighbour if his fails ino communicaion range which is modelled as a fixed circle from he sending node. A he sar of each simulaion a random or grid opology is generaed and a percenage of he nodes are configured wih known posiion which are known as anchor nodes. Afer he simulaion running a number of saisics are produced which allow for a performance evaluaion of he differen approaches. We use hese saisics for our performance sudy in order o asses and evaluae he achievemens of he differen approaches adaped in he underwaer environmen. These saisics are he overall average posiion error and he number of correcly achieved posiioned nodes.. The UWPOS Sofware The UWPOS sofware ool [5] is based on he Posiif package [8] and has been enhanced wih underwaer acousic propagaion faciliies by implemening hree differen propagaion mehods from he Ocean Acousics Library [7], which is a library mainained by he U.S. Office of Naval Research conaining modelling sofware of general purpose for use wihin he ocean research communiy. The incorporaion of underwaer acousic propagaion faciliies allows for a deailed simulaion sudy of he performance of he differen posiioning srucuures under realisic condiions. We are hen able o sudy he dependencies and performance achievemens according o he main physical enviromenal variables in he underwaer medium, such as emperaure, saliniy and deph. Among ohers, he UWPOS sofware allows o define a very precise underwaer sound speed profile (SSP) which is he dependency of he sound velociy according wih he above referred physical variables. The underwaer acousic sound velociy can be calculaed according o he accurae formula proposed by Chen and Millero, known as he UNESCO formula [11], as follows: c = c0 + c1p + cp + cp + AS + BS + CS m/s where: S = saliniy(ppm); = emp( C) pressure P = [ 1,00505(1 + 5,8 sin Φ) z +,6 z + 10,196] 9 5 c0 = 10,88 + 5,0711 5,8085 +,0 1,7800 +,16 7 c1 = 0, ,898 8, ,61 6, c =,160 1,7107 +,597,55 + 1, c = 9,779 +,850,6 A = A0 + A1 P + A P + A P A0 = 1,89 1,6 + 7,16 +,006,1 A1 = 9,7 1,580 6, ,0507, A =, ,101 1, , A = 1, ,69, B = 1,9, + (7,67 + 1,795 ) P C = 7,986 P + 1,77 71

4 PROCEEDINGS OF THE nd WORKSHOP ON POSITIONING, NAVIGATION AND COMMUNICATION (WPNC 05) & 1s ULTRA-WIDEBAND EXPERT TALK (UET'05) Performance Sudy In his secion we show an overview on he performance regarding he four sudied algorihms. The sandard scenario consiss of a he nework of 100 nodes which are placed in an area of 100 square kilomeres. Among he sensor nodes, here are five nodes wih previous posiion informaion known as anchors. The ransmission range is 000 meers in average. Nex we presen he simulaion resuls. Fig.. Average final error vs. Temperaure Fig.. Posiioned nodes vs. Temperaure In Figure and Figure he performance resuls according o he dependency on he waer emperaure are represened. Figure. shows he average final error achieved by each algorihm a differen waer emperaures from 0 C (found in aric regions) o 5 C (found in ropical and near he equaor areas). And Figure shows he number of correcly posiioned nodes a he end of simulaion. In his case he saliniy and deph were se o sandard underwaer condiions i.e. a a deph of 10m and a saliniy of,8pp (pars per housand). We obserwe ha he RPS algorihm []and our proposed APOS approach achieve a consan performance while he waer emperaure varies bu he performance achieved by RPOS is very limied. However he N-HOP [] algorihm is able o posiion a high percenage of he a a final error similar o APOS. A remark from hese resuls is ha he N- HOP [] is no as sable as our APOS. A general characerisic is ha he number of posiioned nodes decreases when he he emperaure increases. Fig.. Average final error vs. Saliniy Fig. 5. Posiioned nodes vs. Saliniy 7

5 PROCEEDINGS OF THE nd WORKSHOP ON POSITIONING, NAVIGATION AND COMMUNICATION (WPNC 05) & 1s ULTRA-WIDEBAND EXPERT TALK (UET'05) Following we invesigae he influence of he variaions in he waer saliniy. These influences are specially criical in suaries and in hose areas where rivers inerac wih he sea or ocean due o he mix of differen saliniy concenraions. In Figure and Figure 5 we observe how he saliniy effec is no as inense as wih emperaure variaions. Moreover we observe how he performance suffer minor variaions and is approximaly consan saring from a saliniy of abou 15pp. Regarding saliniy variaions again he N-HOP and our APOS approaches achieve similar good performance resuls. Nex Figure 6 and Figure 7 provide simulaion resuls of he invesigaion while varying waer deph. Fig. 6. Average final error vs. Range Fig. 7. Posiioned nodes vs. Range Waer deph should be considered wihin his sudy due o he naure of our nework dephs from 10 o 1000 meers are ypicall deph ranges. This range deph can be considered as shallow waer regions, more commonly known as coasal regions. However according o he simulaion resuls we observe how he effecs of he pressure in he performance achievemens of he differens algorihms is invarian due o he fac ha he acousic wave propagaion velociy is almos insensible o pressure variaion in hose ranges. Finally we include an ocean scenario o more accuraely esimae he achivable performance afer simulaion under realisic condiions. The environmen parameers and condiions refer o he equaor, 1 laiude of he alanic ocean. Under realisic condiions we observe ha he N-HOP and our proposed APOS posiioning srucures achieve bes performance resuls while he AHPOS approach seems o be he mos inesable one as Figure 8 and Figure 9 show. Fig. 8. Average final error vs. Range Fig. 9. Posiioned nodes vs. Range 7

6 PROCEEDINGS OF THE nd WORKSHOP ON POSITIONING, NAVIGATION AND COMMUNICATION (WPNC 05) & 1s ULTRA-WIDEBAND EXPERT TALK (UET'05) Based on hese simulaion resuls a similar sysem performance is achieved by he N-HOP and our proposed APOS srucure. However we argue ha our APOS posiioning srucure is more suiable for underwaer acousic scenarios because during he firs phase of he algorihm i.e. he disance esimaion o neighbours, he compuaion is made based on he local geomery of he nodes insead on adding he succesive disance hop-disances from he nodes o he neares anchor as in he N-HOP approach. This effec is even more criical in underwaer neworks where he environmen condiions are coninuosly changing and where he disance esimaions can be srongly influenced leading o higher posiion error esimaions. Moreover his posiion error esimaion adds on every addiional hop. On he oher hand he local geomery of he nodes is supposed o remain sable in ime, only affeced by node baery depleion, node replacemen or by changing he nework opology hrough new nodes o he nework e.g. for increasing he deecion accuracy and he underwaer sensor nework performance. 5 Summary A impac sudy of he mos imporan physical variables which affec he acousic wave propagaion in he underwaer environmen i.e. saliniy, deph and emperaure, and heir influences on he performance of four differen posiioning algorihms has been conduced wihin his work. Afer evaluaion our proposed scheme seems o overcome he oher approaches. According o he obained simulaion resuls we argue ha hese physical variables have o be aken ino consideraion and should be inegraed ino an underwaer posiioning srucure in order o achieve a good sysem performance. 6 References [1] E.M. Sozer, M. Sojanovic, and J. Proakis, Underwaer Acousic Neworks, in IEEE Journal of Oceanic Engineering, vol. 5, pp. 7-8, January 000. [] C. Savarese, K. Langendoen, and J. Rabaey, Robus posiioning algorihms for disribued ad-hoc wireless sensor neworks, in UNISEX Technical Annual Conference, pp. 17-8, Monerey, USA, June 00. [] D. Niculescu, and B. Nah, Ad-hoc posiioning sysem, in IEEE Global Telecommunicaions Conference, vol. 5 pp , San Anonio, USA, November 001. [] A. Savvides, H. Park, and M Srivasava, The bis and flops of he N-hop mulilaeraion primiive for node localizaion problems, in Firs ACM In. Workshop on Wireless Sensor Neworks and Applicaions, pp , Alana, USA, 00. [5] J.E. Garcia, Ad Hoc Posiioning for Sensors in Underwaer Acousic Neworks, in IEEE OCEANS'0 / MTS TECHNO-OCEAN'0, Kobe, Japan, November 00. [6] A. Varga, The OMNET++ discree even simulaion sysem, in European Simulaion Muliconference, Prague, Czech Republic, 001. [7] Ocean Acousics Library hp:// (las visi January 005) [8] Posifi Tesing Framework hp:// (las visi January 005) [9] Consensus Projec Sie hp:// (las visi January 005) [10]K. Langendoen, and N. Reijers, Disribued Localizaion in Wireless Sensor Neworks: A Quaniaive Comparison, in Compuer Neworks, vol., pp , 00. [11]C-T. Chen, and F.J. Millero, Speed of sound in seawaer a high pressures, in Journal Acousical Sociey of America,. vol. 6(5) pp ,

Memorandum on Impulse Winding Tester

Memorandum on Impulse Winding Teser. Esimaion of Inducance by Impulse Response When he volage response is observed afer connecing an elecric charge sored up in he capaciy C o he coil L (including he inside