Research Of Target Tracking Technology Based On Wireless Sensor Networks

With the development of wireless communication, embedded system, sensor technology and micro-electro-mechanical, with perception, computing power and wireless communications capabilities consisting of micro-sensors and wireless sensor networks have aroused great concern. The essential of target tracking is through the distribution of nodes in the wireless sensor networks of mutual cooperation to track the target, which can get independent of accurate tracking results more than a single node. Now, based on wireless sensor networks target tracking has become a research hot Point.For the problem of target tracking on wireless sensor networks, this thesis described particle filter algorithm in detail, discussion its basic principles and core technologies and discuss the problems of particle filters and solutions. Particle filtering method is applicable to any state space nonlinear system, which have the advantage of easy to implement and very flexible benefits, but it still have the problem of lack of plaque particles, filtering performance is not high, and poor real-time, it will cause the phenomenon of particle degradation. For the existing particle filter algorithm this thesis systematically analysis and proposed an new algorithm based on SIR, which is presented based on secondary model, the combination of SIR and the UPF algorithm for hybrid particle filter,The core idea of the algorithm is to generate a new importance function to solve the problem of degeneracy phenomenon to particle filter.In addition, this thesis analyzes the advantage of the clustering networks structure, based on the existing clustering sensor networks structure proposed a new cluster-tree hierarchical networks model. According to the working characteristics of sensor nodes in this model and applied the hybrid particle filter to achieve target tracking. During in many performance criteria of Target tracking in wireless sensor networks, it focuses on the tracking accuracy. Compared with the mixed particle filter, PF, and UPF algorithm, simulation results showed that the mixed particle filter has higher tracking accuracy during the process of target tracking.