Study On Target Tracking Algorithm In Wireless Sensor Network

Wireless sensor network (WSN) is one of the most important technologies of the 21st century, target tracking is an important application in WSN. Since target tracking based on WSN has the characteristics of no fixed network support, quick setup,high fault tolerance and strong survivability, it has been widely used in military, traffic control, environmental monitoring, health care, home intelligent and other fields. The study of target tracking based on WSN has important academic value and practical significance.The nonlinear problems widely spread in target tracking. Particle filter is widely used because of its flexibility to deal with the nonlinear non-Gaussian systems. Aiming at weaknesses of degradation of weights and loss of particles'diversity after resampling in elementary particle filtering, this paper studies the particle filter based on nonparametric belief propagation algorithm. Basing on the summary of wireless sensor networks in target tracking and the mathematical theory related with this topic, the major work and innovation of this paper are like the following:(1) Proposing an algorithm of multi-target tracking based on particle filter. First, the observed data from the sensors are fuzzy clustered by fuzzy C-means algorithm and integrated by linear fusion, then estimate the location of target by sampling importance re-sampling particle filter to achieve multi-target tracking. (2) Proposing an algorithm of distributed WSN target tracking based on nonparametric belief propagation algorithm. First, the nodes which have detected target estimated measurement using kernel density estimation (KDE). Then the measurements are transmitted to the leader, sampling from the product of message by Gibbs and running regularized particle filter to realize target tracking.(3)Based on the discussed earlier, the nonparametric belief propagation algorithm is extended to multiple target tracking. Through modeling the graphical model of multi-target tracking and non parameters -belief propagation, to achieve the correct data association and tracking of multiple target tracking.(4) The initial experiment Papers of the target tracking algorithm based on the ZigBee wireless location development system shows these algorithm is effective and feasible.