Genetic Algorithm for Wireless Sensor Network With Localization Based Techniques

Transcription

1 International Journal of Scientific and Researc Publications, Volume, Issue 9, September Genetic Algoritm for Wireless Sensor Network Wit Localization Based Tecniques * Kapil Uraiya, ** Dilip Kumar Gandi *M.Tec (EC), TIT, Bopal ** Department of EC, TIT, Bopal Abstract-In wireless sensor network nodes position estimation in space is known as localization. Node localization in wireless sensor network is important for many applications and to find te position wit Received Signal Strengt Indicator requires a number of ancor nodes. However te estimation of distance from signal strengt decay in not very accurate especially in time varying environmental conditions and te estimation of exact direction required igly directive antenna but, may still affected by multipat fading. A Genetic Algoritm for wireless sensor network localization is proposed in tis paper to solve te issue tat te positioning accuracy is low wit minimum ancor nodes. Hence in tis paper we are presenting a Genetic algoritm for optimization approac wic tries to find te optimal location by satisfying bot te criteria wit minimal error. Te simulation results also sows effectively outperform bot te tecniques. Index Terms- WSN, Localization, Optimization, Genetic Algoritm (GA). W I. INTRODUCTION SN Sensor network node location information is important for numerous reasons. In many cases te sensed data as no value witout te location information. Te location information can be used by routing and oter protocols, algoritms and services. Te straigtforward solution to te localization problem of equipping nodes wit GPS receivers is not a suitable option because GPS receivers require line of sigt to GPS satellites. Moreover GPS is costly and power ungry. Terefore for te randomly deployed sensor networks various localization algoritms as been introduced were only a small number of sensor nodes are equipped wit GPS receivers and oter sensor nodes derive teir locations by using te localizations tecniques [1]. Toug localization is not a recent topic it still as issues and callenges to andle because some solutions are not ceap and some ave unexpected level of errors. WSN Localization tecniques are largely categorized into range-based and range-free localizations. Te range-based tecnique involves in deriving absolute distances or angles wereas te range free tecnique involves in deriving distances from non-ancor nodes to ancor nodes. Well known range-based localization tecniques are receive signal strengt indicator (RSSI), angle-ofarrival (AoA), time of arrival (ToA) or time difference of arrival (TDoA) etc. Ideally distance can be measured from transmit and receive signal strengts of radios. If transmit and receive signal strengts are pi and pj tan te distance can be measured as d ij = β pi/pj. Were β is known as pat loss exponent and can be calculated by measuring power at unit distance. Since te estimation of distance on te basis of received signal strengt (RSS) in not very accurate because of te fading caracteristics of te pat greatly varies wit time and weater also te angle of arrival (AOA) estimation eiter required a igly directional antenna or array antenna structure wit complex processing algoritm but still error cannot be neglected. Te oter problem is wit localization is it required iger numbers of ancor points to exactly estimate te position in tree dimensional space of node. Te metods discusses above are fall in te range based tecniques owever tere exist anoter approac wic uses only te connectivity information between unknown nodes and landmarks. Tese tecniques can be furter divided into two categories: local tecniques and op counting tecnique. In op counting tecnique node estimates te distances to its neigbor ancor nodes by te op counts and te op size for te closest ancor node and ten estimate its own position, wile te local tecnique node collects te position information of its neigbor ancor nodes to estimate its position. In tis paper we are focusing on te range based tecnique because of its accuracy and adaptability to any protocol and presented a combine RSS and AOA based optimization approac to accurately estimate te location of node. Tis rest of paper is organized as follow, section II presents a brief review of te related literature work wile te III and IV section explains te General localization metod using RSS and AOA tecniques respectively for location estimation. Te section V explains te genetic algoritm and section VI explains te proposed algoritm follower by te simulation results and conclusion wit future scope in VII and VIII respectively.

2 International Journal of Scientific and Researc Publications, Volume, Issue 9, September II. RELATED WORK A survey on different localization tecniques available is presented by Guangjie Han et al [2], tey also reclassify te localization algoritms on te mobility state of landmarks and unknown nodes point of view wit a detailed analysis. Distributed Angle Estimation based approac is presented in [3]. In te literature two antenna ancor are used to transmit linear cirp waves simultaneously, and te angle of departure (AOD) of te emitted waves at eac receiving node is estimated via frequency measurement of te local received signal strengt indication (RSSI) signal. Estimation metod is also improved wit te adaption of multiple parallel arrays to provide te space diversity. Te oter advantage of te tecnique is relying only on radio transceivers and syncronization is needed. Zero-configuration indoor localization to estimate a relationsips between RSSI samples and te distance between nodes is presented in []. A localization approac specifically for te mine environments proposed in []. A. OVERVIEW OF LOCALIZATION ALGORITHMS Te various positioning algoritms tat are reported in te literature can be categorized as follows: Centralized scemes [6] in wic every node sends its neigborood table to a base station. Tis may cause congestion in a large-scale network. Multi-op progress scemes [7] in wic orizontal data flow is required. Tis is not feasible because of te first constraint. Also, most of tese algoritms are not scalable. Recursive Localization Algoritms [8] require tat all nodes can function as ancors and send teir own beacons after estimating teir position. In EMMON te arcitecture ancors are predetermined and are te only nodes tat can broadcast beacon signals (constraint A), terefore te implementation of tis sceme is not feasible. Range-based algoritms include distance estimation between communicating nodes by taking advantage of some caracteristic of te signals excanged suc as Time of Arrival (TOA), Time Difference of Arrival (TDoA), Angle of Arrival (AoA), Received Signal Strengt Indicator (RSSI) or Link Quality Indicator(LQI),Generally tey require te use of extra ardware suc as ultrasound ardware, sopisticated syncronization mecanisms or special antenna equipment [9]. III. LOCALIZATION BASED ALGORITHMS A. General Localization Metod Using Received Signal Strengt By definition, te received signal strengt is te voltage or power measured at te receiver end using signal strengt indicator (RSSI) circuit. Many algoritms take advantage of te power level at te receivers to infer teir distance from te sender. In a wireless sensor network, were nodes apply tis mecanism to self-localize, ancors include teir power level in te transmitted packet and receivers subtract it from te received power. Tis approac is very attractive in terms of device complexity and cost, but te acieved accuracy is its major drawback; accuracy decreases wen te distance increases. Following derivation are used for finding distance from signal strengt: = = ffi = Ten te following equation can be used for estimation of distance between ancor node and te receiver nodes: = =. (1) =. = 2,. 2 = Ones te node estimates te distance from different ancor nodes it utilizes te following algoritm to estimate its location =

3 International Journal of Scientific and Researc Publications, Volume, Issue 9, September = (,, ), =,, Writing te equalities = (,, ) ( ) + ( ) + ( ) =, angle between two peaks represents te relative angle between for te receiver s point of view. Te relation between relative angles and te coordinates is given as follows = 2 atan ( ( ) ( ) ) ( ( ) + ( ) ).. ().. (2) Now ones te angle is estimated from all ancor points te location vector can be calculated by minimizing te equation () = ( ) + ( ) + ( ),.. (3) Hence te location of node can be estimated by searcing te values of (,, )wic satisfies te equation (2) or minimizing te value of objective function (equation (3)). B. General Localization Metod Using Angle-Of- Arrival Range-based system refers to te Angle of Arrival of a signal to a receiver. Tis metod usually provides corresponding information to ToA and RSS by signifying te direction of adjacent sensors. Tere are two ways to obtain AoA measurements and bot require multiple antenna rudiments: te first one is to use a sensor array and array signal processing tecniques and te second one is to use te RSS ratio between two directional antennas located on te sensor. AOA (Angle-of-Arrival) measures local angle information to neigboring nodes, wic can eiter be used as corresponding to oter distance measurements (suc as RSSI), or be used to compute te locations of nodes wit te elp of connectivity information, wic can be acieved in any WSNs. At te moment two different tecniques are used for te estimation of AOA. AT first tecnique receiver utilizes te array antenna parameter and te received signal from eac elements of te array is ten processed to estimate te AOA utilizing te array antenna properties. Te second tecnique for measuring te source signal s AOA, utilizes te rotating, directional antennas, and te angle is estimated by observing te peaks. Te rotational = 2 atan ( ( ) ( ) ) ( ( ) + ( ) ),.. (6), = IV. GENETIC ALGORITHM Genetic algoritms (GA) are searc algoritms based on te mecanics of natural selection and natural genetics, wic maintains a invariable size population P of candidate solutions. During eac iteration step (generation) tree genetic operators (reproduction, crossover, and mutation) are performing to generate new populations (offspring), and te cromosomes of te new populations are evaluated via te value of te fitness wic is related to cost function. Based on tese genetic operators and te evaluations, te better new populations of candidate solution are formed. Wit te above description, a simple genetic algoritm is given as follow [1]: Te algoritm flowcart is sown as figure 1 and implementation steps are as follows: Step 1 Initialize te network, nodes obtain te information of teir neigbor nodes. Step 2. Create arbitrarily a population of binary string. Step 3. Use aritmetic crossover in formula as crossover operator.

4 International Journal of Scientific and Researc Publications, Volume, Issue 9, September 201 Step. Use uniform mutation in formula as mutation operator. Step. If all nodes are localized, output te localization result and end te program. Oterwise turn to Step 2, positioning te next unknown node. Begin Step 3: Ones te node estimates te distance from all te ancor nodes it starts finding te angle of arrival from eac nodes by eiter using array antenna processing or by simple directional rotating antenna. Step : After calculating te information of distance and angles te node uses te genetic algoritm to find its coordinates suc tat in minimizes te objective function given in equation (7 Initialize te network Select number of Ancor Nodes = ( ) + ( ) + ( ), Optimal preservation routlette select offspring Aritmetic crossover and uniform mutation Yes GA Algoritm + 2 atan ( ( ) ( ) ) ( ( ) + ( ) ), No Weter maximum iteration.. (7) No weter all nodes localized Step : if te genetic algoritm finds a solution for te equation 8 it terminates and te returns te solution oterwise it gives te best fitted solution acieved witin te given iterations. Output Yes VI. SIMULATION ANALYSIS RESULTS End Fig 1: GA Algoritm Flow cart V. PROPOSED METHOD Te proposed system estimates te optimal location of node from te existing ancor nodes by using RSS and AOA and finding te optimal solution for bot at te same time. Te proposed algoritm can be described in following steps Step 1: let in te present topology of te network N- ancor nodes wit teir known location are present and all of tem are transmitting teir locations and te power and if tey are not transmitting tese information it is assumed tat te nodes already ave tese information. Step 2: know te node wants to locate estimates te signal strengt of te signal received form eac ancor nodes separately and uses te equation (1) to estimate te approximate distance from eac of te ancor nodes. Te evaluation of te proposed work is done by simulating it for different scenarios and configurations Scenario 1: Table 1: Configuration used for scenario 1 to assessment of te proposed algoritm. Lengt Number of Ancor Nodes 2 GA Population Size 6 Tecnique X y z % Error Time (Sec.) Original RSS AOA Proposed

5 International Journal of Scientific and Researc Publications, Volume, Issue 9, September 201 Scenario 2: Table 2: Configuration used for scenario 2 to assessment of te proposed algoritm. Lengt Number of Ancor Nodes 3 GA Population Size 6 % Time Tecnique x y z Error (Sec.) Original RSS AOA Proposed Tecniqu e Original RSS AOA Proposed X y z % Error Time (Sec.) Scenario 3: Table 3: Configuration used for scenario 3 to assessment of te proposed algoritm. Lengt Number of Ancor Nodes GA Population Size 6 Tecnique x y z % Error Time (Sec.) Original RSS AOA Proposed Scenario : Table : Configuration used for scenario to assessment of te proposed algoritm. Lengt Number of Ancor Nodes GA Population Size 6 Fig 2 : Proposed Algoritm (RSS+AOA) Wit RSS And AOA For Te Location Estimation For Ancor Node Number of individuals Score Histogram fun1 [ ] fun2 [2.8773e ] Score (range) Fig 3 : Proposed Algoritms Score Histogram Grap Between Number of Individuals and Score (range) for Ancor Node

6 International Journal of Scientific and Researc Publications, Volume, Issue 9, September VII. CONCLUSION AND FUTURE ASPECTS In tis paper, a GA localization algoritm wit Localization based algoritm wic are at te same time optimized by te genetic algoritm to find te optimal solution of te location of te sensor node using some ancor nodes. GA localization algoritm is used to estimate te position wit te measured by RSS + AoA. Te simulation results wit different scenario sows tat te present algoritm gives te igest accuracy wit a minimum error of 1% wit is twice better tan te closest competitor AOA. Te result also indicates tat only tree ancor node are sufficient to provide best estimation te furter increase in ancor node leads to increase in time but does not improves accuracy. Te simulation results sow tat te algoritm in tis paper can estimate te position of te unknown node wit less ancor nodes and improve te positioning accuracy efficiently because standard genetic algoritm is used. In summary, we can say tat sensor network localization continues to be an important researc callenge. Despite, many metods and systems to estimate te location of nodes in WSN are proposed and optimization tecnique can be developed but rigt now its depart for future proposed work. VIII. REFERENCES [1] Guangjie Han, Huiui Xu, Trung Q. Duong, Jinfang Jiang and Takairo Hara Localization algoritms of Wireless Sensor Networks: a survey, Telecommunication Systems Modelling, Analysis, Design and Management ISSN Telecommun System, DOI /s [2] Weile Zang, Qinye Yin, Hongyang Cen, Feifei Gao, Member, and Nirwan Ansari Distributed Angle Estimation for Localization inwireless Sensor Networks, IEEE TRANSACTIONS ON WIRELESS COMMUNICATIONS, VOL. 12, NO. 2, FEBRUARY [3] Alessandro Redondi, Marco Cirico, Luca Borsani, Matteo Cesana, Marco Tagliasacci An integrated system based on wireless sensor networks for patientmonitoring, localization and tracking, Ad Hoc Networks 11 (2013) [] Moammad Abdul Azim, Zeyar Aung, Weidong Micael Xiao and Vinod Kadkikar Localization in Wireless Sensor Networks by CrossEntropy Metod, Data & Network Analytics Researc Group (DNA). [8] Wang, C. and Xiao, L. (2007) 'Sensor localization under limited measurement capabilities', IEEE Networks, Vol. 21, No.3, pp [9] Zeng Yang, Censu Wu, Tao Cen,,,Yiyang Zao, Wei Gong, and Yunao Liu, Detecting Outlier Measurements Based on Grap Rigidity for Wireless Sensor Network Localization IEEE Transactions On Veicular Tecnology, Vol. 62, No. 1, January 2013 [10] Jianbin Fu And Silong Pan, Uwb-Over-Fiber Sensor Network For Accurate Localization Based On Optical Time- Division Multiplexing pans@ieee.org [11] Sun Kawakami and Tomoaki Otsuki, Localization using iterative angle of arrival metod saring snapsots of coerent subarrays Kawakami and Otsuki EURASIP Journal on Advances in Signal Processing 2011, 2011:6ttp://asp.eurasipjournals.com/content/2011/1/6 [12] Xinyu WANG, Ziwen SUN, Ziceng JI Genetic Algoritm for Wireless Sensor Network Localization Wit Level-based Reliability Sceme Journal of Computational Information Systems 9: 16 (2013) [13] MATLAB 7 R1 Documentation: ttps:// com. [1] M.L. Sun: Researc on GA based Self-localization Algoritm in Wireless Sensor Networks. Beijing, Cina: Cina University of Petroleum, [1] Teodoros Fronimos, Loukas Petrou, George Hassapis, A Sensor Positioning Algoritm for te EMMON Large Scale Wireless Sensor Network Arcitecture,2012 IEEE AUTHORS First Autor Kapil Uraiya, pursuing M.Tec. degree From TIT Group Of Institutes, Bopal (M.P),India kapiluraiya@gmail.com Second Autor Dilip Kumar Gandi, is working as Professor in te Department of Electronics & Communication, TIT Group of Institutes, Bopal (M.P), India dilipgin23@gmail.com [] Censu Wu, Zeng Yang, Yunao Liu, and Wei Xi, WILL: Wireless Indoor Localization witout Site Survey IEEE Transactions On Parallel And Distributed Systems, Vol. 2, No., April 2013 [6] Adel Youssef and Mustafa Youssef, A taxonomy of Localization scemes for WSN, Te 2007 International Conference on Wireless Networks ICWN'07 pp. -0, Las Vegas, Nevada, June 2-28, 2007 [7] G. Mao, B. Fidan, and B. Anderson, Wireless sensor network localization tecniques, in Computer Networks, vol. 1, no. 10, pp ,2007.