Research On Target Tracking Strategy Based On Dynamic Clusters In Wireless Sensor Networks

Wireless Sensor Networks (WSNs) are distributed, self-organizing, covert and cooperative, so it is advisable to apply them to locate and track targets. However, the tracking strategy used in conditional tracking system is not suitable for WSNs because of the resource-limited sensor nodes, fragile wireless links, and dynamic topology. It is a challenge for WSNs to fulfill tasks of target tracking with high real-time demands. Therefore, coordinated target tracking is not only one of the most important applications of WSNs, but also a difficult and key issue for research.On the basis of sufficient analysis on characters of WSNs and demands of target tracking application, a tracking strategy based on the dynamic clusters is proposed with the thought of distributed and cooperative tracking. The strategy consists of two parts: single-target and multi-target tracking. In single-target tracking, when the target appears in the surveillance area, the tracking system organizes a cluster of sensor nodes to track it. Analytic hierarchy process (AHP) is used in cluster head selection based on node management mechanism. During the target tracking process, the cluster activates suitable sensors to participate in the tracking task, and the range of the activation area adaptively changes with real-time tracking quality. Besides, an adaptive renew mechanism is proposed to avoid the target missing. After the target leaves, the cluster is deconstructed to release sensor resource. In multi-target tracking, a multi-task allocation algorithm is designed based on flexible neural network algorithm to resolve the problem of resource competition, which equalizes tracking tasks among WSNs nodes with minimizing average and total cost as a general principle.The dynamic cluster strategy is simulated in the VC++6.0environment. The result indicates that the dynamic cluster strategy performs well. In single-target tracking, it can get high tracking rate at low cost of communication overhead. In multi-target tracking, the cluster constructed by the multi-task allocation algorithm can prolong valid tracking time, and avoid overhead caused by frequent maintenance of clusters in the execution of target tracking. Therefore, the strategy is proved to be effective and credible.