Research On Target Tracking Algorithm In Wireless Sensor Networks

Wireless Sensor network, which is highly practical, involves the wireless sensor technology, electronic information, computer network, automatic control, embedded technology and so on. And target tracking technology based on wireless sensor network is a hot research issue in recent years. Target tracking technology based on wireless sensor networks, which is not only applied to military in enemy military targets detection and tracking, but also applied to civil aspects, such as ecological environment monitoring, intelligent traffic control, and infrastructure security, has important application value. In this thesis, the target tracking algorithm in wireless sensor network(WSN) are studied, and the main contents are as follows:Aiming at the problem of particle impoverishment phenomenon caused by the particle filter resampling, in WSN the target tracking algorithm based on the improved resampling particle filter is presented. The proposed algorithm avoids resampling residual particle in the residual resampling thus reducing the calculation time; by producing new particles, the diversity of the particles is increased, and so as to improve the particle impoverishment phenomenon and the tracking accuracy.In the wireless sensor networks the conventional target tracking algorithm cannot give dual attention to the tracking accuracy and energy consumption. For this problem, two algorithms are proposed. One is the target tracking algorithm based on dynamic cluster, the other is based on face-structured. In the target tracking algorithm based on dynamic cluster, a new dynamic cluster is generated; and then the fruit fly optimization algorithm is used to optimize particle filter algorithm of the importance sampling; Finally, according to the target’s predicted location coordinates in next moment and the cluster head replaced policy, the cluster heads are determined whether to be replaced. In the target tracking algorithm based on face-structured the network could be divided faces. When a target moves into the monitoring area, sensor nodes that can sense the target estimate the initial position of the target based on weighted least square, and the expanded kalman algorithm is employed to predict the next moment. In order to compare the tracking performance of two algorithms, the simulation experiments are carried. The results show that both algorithms can maintain the tracking accuracy, and extend the life of the network.