Research On Node Localization Methods For Wireless Sensor Networks

A wireless sensor network (WSN) consists of a large number of micro sensor nodes through communicating with each other to form a multi-hop self-organizing network system in the surveillance area. The purpose of a WSN is to monitor various environmental characteristics of the deployment area, such as temperature, humidity, light, voice, chemical concentration and so on. The WSN fuses the logical information world and the objective physical world, and changes the way that people interact with nature. The WSN is widely applied in the areas of environmental monitoring, traffic management, medical health, military defense and many other areas.In many applications of WSNs, node localization is the base of target tracking and the corresponding operation in the surveillance area. Sensor nodes collect and send the information needed to the server. The physical information which is sent by the sensor nodes is insignificant without the location information. Even sensor nodes sometimes only need to send the location information. Therefore, node localization is one of the key technologies of WSNs.This dissertation makes a deep analysis of the WSNs' own characteristics. This dissertation also improves and replenishes the deficiency of existing researches. Finally, the research results which have theoretic meaning and practical application value are obtained. The specific contents of this dissertation are as follows:(1) A range-free method based on the placement of anchor nodes is proposed, and the corresponding four corners DV-HOP localization algorithm is designed. In the original DV-HOP localization algorithm, anchor nodes need twice broadcasts in the while network. One broadcast is used to obtain minimum hop counts between two anchor nodes. The other is used to broadcast the correction value of each hop. Twice broadcasts cause a lot of traffic. The traffic causes a lot of energy consumption and decreases the life of WSNs. The four corners DV-HOP localization algorithm utilizes the placement of anchor nodes to divide the whole area into some small regions. The broadcast of anchor nodes is limited near a small region, and unknown nodes only use the four anchor nodes in the small region to compute their location. The simulation results show that the four corners DV-HOP algorithm decreases the communication traffic and the localization error.(2) To address the problem that existing range-free localization methods encounter large localization error, a range-free localization method based on proximity is proposed and the corresponding PNN-MAP localization algorithm is designed. Proximity is used to denote the distance relation between neighbor nodes. First, a linear function is derived based on the geometric feature. The input is the number of neighbor nodes, and the output is the value of proximity. Then a correction value of proximity is calculated by the distances of anchor nodes and proximity of neighbor nodes. The product of the correction value and proximity between neighbor nodes is the estimated distance. Finally, the estimated locations of unknown nodes are calculated based on the estimated distances and MDS-MAP algorithm. The simulation results show the PNN-MAP localization algorithm achieves better results than the current algorithms in both the distance estimated error and localization error.(3) According to the high localization error of range-free localization methods in the irregular area, a range-free method based on a filter parameter which is suitable for irregular areas is proposed and the corresponding DPLA algorithm is designed. First, a distance estimated algorithm is designed. To make this algorithm also suitable for the irregular area, a filter parameter is derived to filter the anchor nodes' information which is influenced by the coverage hole when we calculate the correction value. After obtaining the estimated distances, we divide two kinds of situations to calculate the unknown nodes' locations. If the area is regular, unknown nodes use the estimated distances to all anchor nodes and perform the maximum likelihood estimation method to calculate their estimated locations. If the area is irregular, unknown nodes use the distances and corresponding coordinates of four anchor nodes which are closest to it, and perform the maximum likelihood estimation method to calculate their estimated locations. The simulation results show that the distance estimated error and localization error of the DPLA localization algorithm are lower than the same type of distributed localization algorithms in both the regular and irregular areas.(4) According to the problem that how to obtain more accurate location results with the same distances between neighbor nodes, a two-stage localization method based on greedy idea is proposed and the corresponding GIL localization algorithm is designed. The localization problem is considered as a combinatorial optimization. First, we design an objective function to represent the quantitative measure of the accuracy of estimated locations. Then, we make a deep analysis on the distance relationship between two successive location results and design a neighborhood function based on the assumption that the next location results are correct. To obtain more accurate location results, the proposed algorithm is divided into two phases. In the first phase, a set of location results is generated based on the greedy iterative optimization. In the second phase, some unknown nodes will be elevated to anchor nodes, and the first phase is executed again to alleviate the flip ambiguity. The second phase is repeated until there is no unknown node that can be elevated to an anchor node. Finally, the experimental results show that the GIL localization algorithm achieves more accurate result and takes less time than the existing localization algorithms after obtaining the same distances of neighbor nodes in both regular and irregular areas.