Study On Node Localization Algorithm For Dynamic Wireless Sensor Networks Based On Monte Carlo

Node location information is the foundation of system function whichare event location report, geographic routing, target tracking, network managementand so on in wireless sensor network. Therefore, node localization problem is a keyproblem in practical process of wireless sensor network. Mobile node localization indynamic wireless sensor networks is a hot problem in the field. Monte-Carlolocalization Boxed algorithm is a commonly used method of dynamic localization, notonly it can not be influenced by mobility, but can take advantage of mobility toimprove localization accuracy, reduce the localization energy consumption. However,MCB algorithm still has problems, such as sample degeneration and weak filteringconditions.Due to the problem of sample degeneration in MCB mobile localizationalgorithm, a new localization method named Improved Monte-Carlo localizationBoxed is proposed. Based on the MCB algorithm, through analyzing the localizationresults of current time、distance between nodes and the information of node relativeposition to obtain the sampling probability of different regional of sample box for nexttime, the sample points can fall in the area where the posterior probability is larger asmuch as possible, therefore the problem of low accuracy caused by sampledegeneration in the original algorithm is solved effectively. The simulation resultsshow that, under the same conditions, the average localization accuracy is improvedby about14%, and the average energy consumption for localization is reduced byabout27%by comparing with the MCB algorithm.Due to the problem of weak sample filtration condition of MCB mobilelocalization algorithm, a new Monte-Carlo localization Boxed algorithm based specialanchor node is proposed. Based on the MCB algorithm, the special anchor nodesbesides one-hop and two-hop are introduced to enhance sample filtration condition,therefore the problem of low accuracy caused by weak filtration condition in theoriginal algorithm is solved effectively. The simulation results show that, under thesame conditions, the localization accuracy is improved by about15%, and the energyconsumption for localization is reduced by about40%by comparing with the MCBalgorithm.