| Wireless sensor network (WSN), one of the21st century’s most influential technologies, which is used in a wide range of fields, such as:road traffic monitoring, airports and large area monitoring, smart home, military monitoring and tracking etc. Target tracking is an especially important application of the wireless sensor network. Therefore, we mainly research the mobile target tracking based on wireless sensor network (WSN). In this article, the research is carried out in terms of Wireless sensor network-based moving target tracking and then introduced a particle filter algorithm.The algorithm can effectively solve the target tracking problems which cannot be accurately described by Non-linear, non-Gaussian environment and target motion state.First,considering that the nodes energy is limited, this paper constructed a dynamic collaborated tree network model in order to resolve the mutual collaborating issues between the sensor nodes in wireless sensor network target tracking, in which it will produce cluster head nodes transmitted data by other perceived nodes during target moving Furthermore, it will also produce new cluster head nodes according to cluster head nodes producing strategy as target moving when the previous cluster head nodes switch into perceived nodes to track targets.。 Then, this paper also introduced and analyzed the existing target tracking algorithm. According to wireless sensor network-based target tracking problems, this paper chose particle filter algorithm due to its unique advantage in dealing with state and parameter estimation problems of nonlinear, non-gaussian systems. Finally, this paper systematically analyzed the existing wireless sensor network-based particle filter algorithm and deeply discussed the existing problems in the particle filter algorithm with presenting the solutions; meanwhile, we did a large amount of detailed research on particle degradation problems. The traditional particle filter algorithm pay majority of attention to small particles to update weights. However, particle dry phenomenons lead to the decrease of residual particle number. As a result, it is difficult to represent the state of the estimated characteristics. Based on the above focus in this paper to improve the particle filter, it introduced the price reference function when resampling and calculated the value of particles to the observed value, and then assigned different weights to particles depending on the value of particles. This method adjusts the distribution of the weight by observation value; increased the particles diversity; and weaken the degree of particle degradation.Wireless sensor network has huge application potential and it has great practical application demand in the field of target tracking. Therefore, we can clearly conclude that there is still huge developing space in this field. Looking forward to our further exploration and research. |
