The Research On Distributed Target Tracking For Wireless Sensor Networks

Target tracking is one of the most important applications in wireless sensor networks. Tracking a moving target requires a large number of nodes to work together, and also need to process and transmit large amounts of data based on resource-constrained wireless sensor network. One possible way to achieve the maximum utilization of those resources is to apply a distributed target tracking scheme, which must consider the trade-off among energy consumption, quantities of data transmission and tracking accuracy. This dissertation address on target tracking problem for wireless sensor network, in which the nodes self organization, target state estimate, target tracking strategy and target recovery issue are emphasized, the main contributions are as follows:Sensor nodes dynamic self-organization problems with the target movements are studied for wireless sensor network target tracking. Based on node energy consumption model, the relationship between the number of nodes, the communication distance and energy consumption is summed up. Based on the predicted position of the moving target, the node dynamically collaboration nearest neighbor (NNC) self-organizing strategy is proposed. The relationship between the nodes of the cluster members, the node remaining energy is incorporated into improved NNC which is to remove redundant nodes of the cluster members in order to reduce network the total energy consumption. The proposed method has good tracking precision while single hop communication distance between nodes can effectively reduce the power consumption. Improved node tissue can reduce the demand on the number of nodes while maintain the tracking accuracy.Multi-sensor distributed state nonlinear estimation problem is researched for wireless sensor network; distributed extended Kalman filtering method based on the consensus filtering is developed. It is proved that the distributed nonlinear filter system state estimate is equivalent to the centralized nonlinear estimation results obtained. Three extended dynamical consensus Kalman filtering algorithm was proposed, i.e., observation-based, innovation-based and estimate-based consensus kalman fitler, which is adapted to different applications. Simulation results show that, compared with central kalman filter algorithm, observation, innovation and estimate based consensus distributed kalman filter will get comparable accuracy while it will be more robust due to its distributed manner.In order to meet the requirements specified tracking while minimize network energy consumption and maximize network lifetime, adaptive target tracking protocols is studied. Sampling interval adaptive tracking and dual adaptive target tracking are proposed. The method can adjust tracking sampling rate according to tracking error, it is an effective method. When tracking accuracy cannot meet requirement while sample interval is minimum, double adaptive target tracking protocol will change nodes wake up the regional automatically. This method have fault tolerant can decrease target miss rate with require tracking accuracy.Reasons about loss of target in the wireless sensor network target tracking process was researched and then study lost target recovery mechanism. Consider the uncertainty of the target movement, a target based on the prediction of multi-step recovery mechanisms, focusing on the radius of the recovery area. The recovery mechanism is composed of four phases. The recovery mechanism will improve the fault tolerance of the system to ensure reliable operation of the system.