Research On Localization Algorithnms For Wireless Sensor Networks

Positioning plays a critical role in many applications. It is one of the basic functions of wireless sensor networks(WSNs) to determine the position of sensor nodes and the occurred event. And the former is the basis of the latter. After the located node receives the information from many reference nodes, some better reference nodes need to be selected to estimate the position of the located node to improve the localization accuracy. It is the main research content of the choose mechanism of reference nodes for localization to how to select the best suitable reference nodes to locate the unknown nodes and achieve the best localization results.Based on the requirement of the research tasks, the thesis mainly discusses the above three problems. As a whole, the main works and creations are as the following:1. The representative self-localization algorithms for wireless sensor networks are classified and analyzed, their main merits and problems are pointed out.2. Aiming at the self-localization problem,1) an improved weighted centroid localization method is proposed. Weights are determined based on the different force of reference nodes to the located node and the localization accuracy is improved.2) a distributed localization method based on the multi-hop connectivity information between the located node and reference nodes is proposed. It utilizes the multi-hop reference nodes to locate the unknown node. After the positions of all the unknown nodes are estimated, the estimated positions are refined based on the connectivity information between the located node and the neighbor nodes to reduce the localization error further.3) a cooperative self-localization method based on probability for wireless sensor networks is proposed. The method firstly estimates the initial position of the located node based on the joint probability density function of the distance between the located node and the connected reference nodes. Further, based on the refinement principle, a cooperative localization method is studied by making the best of the neighbor nodes and giving the neighbor nodes some confidence. The method improves the estimation accuracy as well as makes more unknown nodes to be located.4) an assisting localization method that utilizes the known outside reference signal source to locate the unknown nodes in sensor networks, and the localization model based on angle is established. Simulation results show that the method has some practicability.3. Aiming at the target localization problem,1) an improved NLS localization method is proposed to locate the target based on the received signal strength indicator. And the localization performance based on two different weights is analyzed.2) a Bayesian filter based localization method using the detection model of the sensors is proposed. The method firstly establishes the probability grid based on the discrete Bayesian filter, then gives each grid point a weight based on the detection model, finally estimates the target position based on two search methods.3) a cooperative perpendicular bisector localization method based on the received signal energy is proposed. The method takes advantage of the magnitude of the sensor networks, analyzes the difference between the RSSI of every two nodes and estimates the target position based on the geometry relation that the target locates at the perpendicular bisector of the two points.4) a localization system for coal mine is studied based on the WSN localization technology.4. A new parameter called DC(Degree of Collinearity) is proposed to analyze the localization performance with respect to the different placement of the terns of the reference nodes. Two theorems also are proposed and proved to analyze the impact degree of the parameter on the localization. The selection and placement mechanisms of the reference nodes are proposed based on the two theorems. Based on the above analysis, a weighted distributed localization method is proposed based on the degree of collinearity. The method selects the reference nodes that participate in locating by considering the topology of the reference nodes and the position relation between the located nodes and the reference nodes, to avoid utilizing the bad terns of the reference nodes based the DC information and improve the localization accuracy.