Node Localization And Target Tracking In Wireless Sensor Network

Wireless sensor network (WSN) is a novel technology for acquiring and processing information, and it can be widely used in environmental detection, traffic control, target tracking, and smart home, etc. Localization technology which can be divided into node localization and target tracking is a hot research topic in WSN. Node localization is the precondition of target tracking. Sensor node is subject to many restrictions such as computing power, communication ability, energy consumption, which bring a serious challenge to the localization technology. Based on analysis and summary of existing research results, the contributions of this article can be stated as follows:1. In order to solve the problem of location positioning errors in WSN, an effective semi-definite relaxation programming method is proposed. Based on the distance measurement model of the time difference of arrival, the optimization function is established, and the node location is determined by the semi-definite relaxation programming method based on convex optimization theory.2. Aimed at the problem of node localization in WSN, a novel localization approach is proposed based on twin support vector regression. The position information and hop counts among anchor nodes are taken as training samples. Combined with the Lagrange method and KKT (Karush-Kuhn-Tuchker) conditions, twin support vector regression is adopted to attain mapping model between hop count and distance by converting the optimization of the original problem into dual form. The least square method is used to obtain estimation position of unknown node. The simulation results show that the location accuracy is improved.3. Aiming at the quality of coverage problem in WSN, to improve the performance of network monitoring, deployment location strategy is proposed based on the connectivity and the evaluation function of Voronoi partition. The experimental results show that the algorithm ensures the network higher coverage and nodes move distance less.4. In order to improve the prediction accuracy for the target, target position prediction method is proposed based on Gauss mixture model. The motion trajectory of the target is considered as the Gauss mixture model. The previous position data are taken as the training set to determine the regression function parameters, and then predict the position of the target trajectory.5. In order to balance the energy consumption and target tracking accuracy, a sensor scheduling algorithm based on probabilistic coverage is proposed. The calculation method of noise statistical parameter is given. Based on maximum likelihood estimation, efficient tracking sensor scheduling algorithm is designed. The simulation shows that the proposed algorithm has small energy consumption and high tracking precision.