A Study Of Target Localization And Tracking Algorithms In Wireless Sensor Network

The wireless sensor network (WSN, Wireless Sensor Network) is just an interdisciplinary subject involving the realms of sensor, communication technology, microelectronics, distributed information processing technology, etc. because of its good flexibility, low cost, high location accuracy, WSN has been widely used in many fields. As a supporting technology of WSN, target locating and tracking technology appears particularly important and has been studied at home and abroad.The target locating and target tracking technology is just obtaining the target location and state with WSN and then predicting and estimating its next location and state. This paper improved the weighted centroid localization algorithm. Firstly, a hybrid algorithm combined with weighted centroid location algorithm and differential hybrid location algorithm is used to calculate the target location. And then the Unscented Kalman Filter algorithm is used to track the target.The traditional centroid algorithm is based on RSSI, it calculate the weighted number of the distance between the anchor node and the target node. Since giving different weighted distance between the nodes, the localization accuracy, to some extent, is improved. But indirect error is caused. Our improved algorithm do not give different weighted distance between the nodes, but weighted calculate the target node by directly using the respective RSSI values for the weights. The centroid locating algorithm is simple and its accuracy is low, so we call it coarse locating. Then, the exact coordinates of the target node can be obtained by using the differential evolution thought. The coordinate calculated by coarse locating is chosen as the initial value.The target tracking technology is mostly achieved by using Calman filter. Calman filter only work in calculating the linear equation effectively. If encountering non-linear equations, we have to change them into linear equations firstly, which undoubtedly increases the accumulation of errors. Thus, UKF algorithm, which uses UT transform and sampling function, is proposed to achieve the mapping from nonlinear to linear and improve the accuracy.Through simulation, it is found that precision is greatly raised by2%-5%, though computation is a little increased. And more effective tracking can be achieved by using UKF than extended Calman filter.