Localization Optimization Methods For Wireless Sensor Networks

Wireless sensor network (WSN) has received increasing attention in recent years, which has wide applications in the aspects of industry, military, and environment. As one of the key enabling technologies and research hotspots, the nodes localization is very important significance due to its direct correlation with theoretical study and practical application. This paper focuses on the localization optimization methods for WSN, aims to obtain accurate location information, lower energy consumption and comprehensive performance optimization of positioning system.Different coordinate calculation schemes such as maximum likelihood estimator, Taylor based least-squares method, weighted least squares are applied to localization refinement in order to improve accuracy, and their performances are analyzed and compared. Aiming at self-position deviation of anchors in actual localization process, some adaptive coordinate refinement methods such as total least squares, constrained total least squares are designed. The estimation lower bound of localization error, named Cramer-Rao lower bound is extended so as to the location error characters in wireless sensor networks.A novel LDV-Hop algorithm is put forward based on the DV-Hop so as to the need for low cost and low power in WSNs. It can adjust the quantity of communication packets according to the hops in the course of localization. Location information is obtained only in limited ranges, and thus error information caused by collision is prevented in MAC layer. Simulation results based on the network simulator Ns-2 show that the LDV-Hop algorithm can achieve better localization accuracy and reduce the number of communication messages. The algorithm is also adaptive to irregular distribution of sensor nodes. A novel ODLA algorithm is put forward so as to the problems that a few new comer or moved nodes, instead of whole network, need to be located or relocated in WSNs. It only computes coordinates of demandable nodes, and thus avoids being concerned with all nodes in networks. In this way, the energy consumption is reduced greatly and localization speed is improved. Simulation results show the localization accuracy and required bounds of neighbors for localization in two coordinate calculation methods. The ODLA algorithm is also adaptive to irregular distribution of sensor nodes.In order to solve localization error problem caused by radio irregularity in wireless sensor networks, related principles and methods of graph theory are used to carry on modeling for the irregular radio range and directional communication. In the case of asymmetric communication, solution schemes are put forward for locating nodes through the analysis to communication path between nodes in directed network. The localization performances are compared in different scenes and conditions. Simulation results show that localization accuracy of the corrected network under regular network topology can improve 10% to that of without correction. While in irregular network topology, the localization accuracy can be improved when strong connected network comes into being.There are a few of boundary nodes and some isolated nodes in wireless sensor networks. If these nodes are judged and treated availably, the location error of whole network will be reduced. In order to solve localization error problem of boundary nodes and isolated nodes, related principles and methods of graph theory are used to judge these misbehavior nodes. Through the analysis of density around misbehavior nodes, range and direction of received anchors, solution schemes are put forward for locating boundary nodes and isolated nodes. The localization performance is compared in different scenes. Simulation results show that localization accuracy of misbehavior nodes with correction under regular network topology can improve 17% to that of without correction. While in irregular network topology, the localization accuracy of isolated nodes can be improved 10% to that of without correction. A novel relative localization algorithm ESGF is presented based on extended subgraph so as to the problem of invalid nodes caused by message collision and energy restriction of wireless sensor networks. The whole network is divided into several subgraphs. The sensor nodes in the local subgraph area are located relatively. Then the located subgraphs are combined to generate extended subgraphs. Meanwhile, new boundary nodes are supplemented to assist to locate other nodes. Compared with the algorithm of clustering-based SPA, ESGF can reduce the amount of communication packets and invalid nodes. It also increases the location coverage rate and is more adaptive to irregular distribution of sensor nodes.In actual application, nodes of wireless sensor networks are often distributed in three-dimensional space. A novel three-dimensional localization algorithm is put forward for the sake of reducing location error and improving algorithm applicability. It uses the method of sampling in three-dimensional space and range constraint, combined with weighted filtration to successful sample points, to acquire three-dimensional coordinates of nodes. According to different node function, the algorithm can be executed in hop-based or range-based mode. Three sampling schemes are adopted to analyze and compare the localization algorithm. Irregular space distribution of sensor nodes is introduced to prove localization performance.In summary, by researching on location refinement, local localization, on demand localization, badness conditions processing and three-dimensional sampling algorithm, this paper presents the system solution of localization optimization methods for wireless sensor networks.