Research Of Multi-target Tracking Based On Particle Filter Algorithm In Wireless Sensor Network

Target tracking is one of the hotest issues in modern science and technology.The application scope of target tracking involves military,industrial,commercial,medical and other fields.Target tracking has been developing for more than fifty years which led to the improvement of related theories and technologies.With the rapid development of sensor technologies in recent years,Wireless Sensor Network(WSN)has attracted the attention of different parties.Since WSN has many features like wide coverage,low cost and diversity of detection methods,WSN is considered to be one of the most important technologies in the 21 st century.The application of WSN in target tracking has become the frontier research direction.This paper first introduce the architecture of WSN,the technology of node locating and target tracking,then summarize the factors that need to be considered in WSN locating and tracking,including the modeling of target motion state and some common motion models and the problems in tracking the WSN include the difficulty to determine the initial state of the target and the filter,the loss of intermittent targets,and the poor performance of the filter in the case of clutter and error measurements.Aiming at these shortcomings,this paper proposes a particle PHD filter for multi-target tracking in WSN based on the Random Finite Set theory(RFS)and the Finite Set Statistical method(FISST).A distributed particle PHD filter is given to reduce the computational complexity of the traditional PHD filter and improve its real time tracking performance.In this paper,a 360 ° ultrasonic sensor hardware node based on HC-SR04 array is designed for WSN.Regarding to the synchronic requirement of distributed particle filter,an AODV routing protocol based time synchronization algorithm TS-AODV is proposed and based on which a distributed particle PHD is researched and realized.The experimental results show that the particle PHD filter algorithm can effectively estimate the target state and target number of multi-target motion in sensor network target tracking applications,whose performance is stable under multi-clutter environment.Compared with the centralized particle PHD filter algorithm,the singlestep calculation time of the distributed particle PHD filtering algorithm is shorter,which improves the real-time performance of the algorithm.