Research On Tracking Algorithms Based On Particle Filtering In Wireless Sensor Networks

Recent decades, with the development and innovation of sensors, computer, information processing, wireless communication and other related technologies, wireless sensor networks are widely applied in various areas. While in many applications of wireless sensor networks,such as troop’s mobilizing on the battlefield and traffic control, it needs to obtain the real-time information about the target. Thus, it has great practical significance to conduct the research on the target tracking algorithm in wireless sensor networks.This paper discusses the localizing method and filtering algorithm in target tracking of wireless sensor networks, and the main contents of this paper are as follows:(1) In terms of positioning, this paper briefly compares and analyzes several traditional positioning methods, then it proposes an improved step by step location algorithm that uses the same method to select three coordinate for estimation. Simulating improved algorithm and comparing with trilateration and general step by step method, the results show that the overall performance of the improved algorithm is the best among the three.(2) In terms of filtering, simulation results shows that with regard to the nonlinear and non-Gaussian filtering in target tracking, the density estimation based particle filter outperforms the extended Kalman filter and unscented Kalman filter algorithm, so this paper studies the particle filter algorithm. With regard to the particle degradation problem in particle filter, this paper proposes a hybrid proposal distribution that combines prior distribution and the importance density function that obtained by UKF method to slow down the degradation of the particles. Besides, with regard to the sample dilution problem, the paper puts forward a improved classification resampling method that makes the particles of bigger weight having a random disturbance that control by the noise variance and the particles of smaller weight having a higher probability, which can increase the diversity of the particles. The results show that the improved particle filter is more accurate and it takes less than half time than UPF.(3) This paper also uses the improved algorithm in the target tracking in wireless sensor networks, and the simulation results verify its effectiveness under the different noise variance and target operating conditions.