Study On Key Technologies For Target Tracking In Wireless Sensor Networks

Wireless Sensor Network (WSN) with large redundant nodes is suitable for target tracking because of its high self-organization. However, there still exist several challenges, such as limitation of energy, computing capability and short-distance communication. So, it is of theoretical and practical significance for studying on target tracking system of WSN.WSN topology and self-organization methods are analyzed first. Considering of the inadequacies of LEACH algorithm, a vote based cluster head election mechanism is proposed. This method insures only the nodes have more neighbors and remaining energy can be elected as cluster heads. At the same time, the cluster head number must be chosen according to network energy consumption.By comparing the energy consumption and reliability of single-hop and multi-hop, RMHR protocol is designed. RMHR selects the routing according to its PRR when packets need to be transmitted. Since the acknowledgement and the multi-path techniques are adopt in the protocol, energy consumption and time delay caused by data retransmission are reduced.RSS based OBS localization protocol is designed, and the concept of location efficacy is first proposed. To avoid error transmission in location process, nodes choose beacons with better location efficacy to localize themselves. It can be concluded from the experiments that OBS gets better location precision compared to maximum likelihood algorithm and Centroid algorithm when beacons are not enough.Equations of detection probability and detection delay are deduced by analyzing the target detection theory in WSN. It finds that node density, sensing radius and duty cycle are key factors impacting the detection performance. Based on the theoretical studies, MTDS algorithm has been developed to balance the energy consumption, which arranging nodes working time according to their residual energy.WSN target tracking model is established based on dynamic tracking region, in which untagged target position and trajectory can be calculated and forecasted. The dissertation optimizes the area using target's length, velocity and turning angle. So the area of tracking region and the number of nodes participating in the tracking task can be decreased efficiently.