Research On Target Tracking Algorithm Based On WSN

Wireless sensor network(WSN), which is precise, reliable, timely and hidden in tracking, is suitable for tracking targets in complex environment. In consideration of the restrictions for WSN nodes, an study of the hotspots of WSN target tracking is conducted in this dissertation. The main contents are stated as follows:Aiming at the contradiction between tracking precision and energy consumption in the process of WSN collaborative tracking, a reinforcement learning based collaborative tracking algorithm is presented. In this algorithm, the optimal cluster head action strategy is made by Q-learning method to make the cluster head changed and cluster members selected optimally. Under the premise of satisfying the tracking precision, the number of cluster member is reduced and the sampling time interval is selected dynastically to reduce energy consumption. Simulation results show that the proposed algorithm can not only guarantee the tracking accuracy, but also can reduce the energy consumption of the network.A measure-driven probability hypothesis density filter is proposed to solve the challenge of the unknown target birth intensity in multi-target tracking. The dependence of the prior knowledge of the target birth intensity is avoided by using the measure-driven method. Using the augmented state space, the interference of clutter to the real target intensity estimation is avoided. Simulation results show that the proposed algorithm is sensitive to the change of target number and improves the accuracy of tracking obviously when the computation complexity is reduced.A quantized innovation based target tracking algorithm is proposed to solve the energy restriction in WSN. By quantizing the measure innovation, the quantized error is reduced. Meanwhile, energy consumption is decreased by setting data censoring threshold. Simulation results show that there is small change in the tracking accuracy while the energy consumption of the network is cut down.A target tracking simulation platform based on WSN is constructed, including the six-element array ultrasonic sensor network and the monitoring software platform. Asynchronous sensor data acquisition and wireless network data transmission are realized by the former part. The target tracking trajectory can be calculated with the acquisition data and the performance of the target tracking algorithm under different parameters can be studied by the latter part. Tracking of moving targets is preliminarily realized by testing on the platform.