Study Of Target Tracking Algorithm Based On Sensor Networks

Wireless sensor networks(WSNs), which integrates the micro-electro-mechanism system(MEMS), sensor, wireless communication and distributed computing, is a brand- new technology for obtaining and processing information. With its advantages of broad covering, Self-Organizing, high fault-tolerance and high precision, as well as the distinctive features of low cost, flexible construction etc, the WSNs are utilized widely in military and civil fields. Target tracking is not only a basic function of WSNs, but also a typical application, it has a deep theoretical research value.This thesis first introduces the WSNs'concept, architecture, characteristic, key technologies and applications, and then describes the node localization algorithms, with the emphasis on range-based and range-free. Based on the detailed discussion of Kalman filter algorithm (KF) and improved KF, particle filter algorithm (PF) and improved PF, a mixed algorithm is presented, called unscented-mixed particle filter (UM-PF). This algorithm utilizes a mixed particle-propagation scheme, in the process of multi-sensor measurement fusion using a particle filter in WSNs, a certain number of particles in the particle filter are propagated by using a Gaussian distribution obtained from an unscented Kalman filter as the proposal distribution, while the rest of the particles are simply propagated by using the state transition prior distribution. Simulation results show that, at close range of tracking time, the tracking precision of UM-PF is obviously higher than that of PF, and with respect to acquire the similar tracking precision, UM-PF costs much less time than PF.