Research On Localization Algorithms For Nodes Of Mobile Wireless Sensor Networks

Localization is one of the key technologies in wireless sensor networks. It is meaningless to send the data measured by sensor nodes without adding the position information of nodes. So far, most researches are based on static wireless sensor networks, however, when applied to mobile wireless sensor networks, their localization performances are not very good because of the mobility of nodes. Considering many specific needs, such as mobile target tracking and localization, wide range of environmental monitoring and so on, mobile wireless sensor networks has some irreplaceable advantages. Therefore, it has an important theoretical significance and application value to design localization algorithms for the mobile wireless sensor networks.This dissertation is based on monte carlo algorithm and find some methods to improve the localization performance for mobile wireless sensor networks. The main works and innovation of this dissertation are as follows:1. A sampling optimized-based monte carlo boxed localization algorithm is proposed, called SOMCB. In the existing monte carlo based algorithms, when a sample does not meet the filtering condition, it will be filtered and another new sample will be obtained. In order to get a certain number of samples which meet the filtering condition, the sampling and filtering periods are done repeatly, so the efficiency of sampling is not high. To solve this problem, we propose a sampling optimized-based monte caro boxed (SOMCB)localization algorithm, it introduces the received strength of signal of node to further restrict the sampling box. When obtained some orginal samples, we choose the weight of samples as objective function, and apply differential evolution method to them to make them move towards the real position of node to be localized. The simulation results show that, under the default parameter of networks, the number of candidate samples of SOMCB reduced significantly, meanwhile, the area of sampling box is reduced by about20%, the localization accuracy of node is reduced by about25%when compared with MCB.2. A tempory anchor-based localization algorithm is propsed, called MTB. In monte carlo based mothods, when filtering samples, nodes use the information of anchors heard. If the anchor density is low, the number of anchors can be measured by a node will be reduced and some samples which are far from the real position of node are posibly not filtered, finally, the localization accuracy will become worse. To solve this problem, we propose a tempory anchor based localization algorithm. It introduces common nodes and let them be tempory anchors, also, we propose a new concept-credit to weigh the performace of tempory anchors and modify the filtering equation. The simulation results show that, in low anchor density,such as Sd=0.125, the localization error is reduced by about33%compared with WMCL, reduced50%by average compared with MCL, MCB, MSL. Meanwhile, the performance of MTB if better than other four algorithms when the parameters of networks vary.