Target Tracking Algorithm For Binary Wireless Sensor Networks

Presently, target tracking based on Wireless Sensor Networks (WSN) has aroused wide concern in most areas, owing to its wide application prospects. Target tracking system based on WSN with Practical, efficient and superior performance has become research hotspots in the fields of science and technology.First, this paper introduces the principle and content of target tracking, studies the existing target trajectory model and establishs the state space model of target tracking system. Then, the paper studies the Kalman Filter algorithm, Extended Kalman Filter algorithm, Unscented Kalman Filter and the Particle Filter which is broadly used to deal with the non-linear, non-Gauss problem. Through the simulation, the deficiencies of the extended Kalman filter algorithm and unscented Kalman filter algorithm in nonlinear problems are analyzed, and the advantage of particle filter is verified.Secondly, the deficiencies of particle filter algorithm is analyzed:The particle degeneracy may cause the cumulative effect of particle filter prediction error. The cumulative effect of particle filter prediction error can not be completely eliminated by adopting resampling, and the resampling may cause sample impoverishment. Therefore, the variance adaptive technology is introduced in particle filter in this paper. Though modifying the variance of the process model noise of system, the performance of importance density function is improved. The formula of adaptive variance is proposed. The simulation results show that the particle filter algorithm with adaptive variance is superior to the standard particle filter in estimation precision and processing speed. The validity of the formulas proposed in this paper is verified.Finally, the target tracking based on Binary Wireless Sensor Network is studied. Target tracking system based on WSN need to consider resources of WSN, cost of system construction and precision, stability, real-time of target tracking system. The energy of network can be reduced to minimum in BSWN. Because target tracking is a typical no-linear, non-Gauss problem, it can be solved by adopting particle filter. The simulation results of non-maneuvering target and maneuvering target show the advantages of proposed particle filter in estimation precision and processing speed in target tracking based on BWSN, and the Validity of proposed formulas is verified again.