Research On Key Technologys Of Traget Tracking On Asynchronization Condition In Wireless Sensor Networks

The technology of target tracking which based on wireless sensor network can be applied in many areas widely. Time synchronization is separated from target tracking in existing target tracking methods. Running time synchronization algorithm independly, results in increas of energy consumption in network inevitably and accuracy decline of target tracking as well. If time synchrozation is inosculated in the process of target tracking on time asynchronization condition, which will be propitious to advance the practicality evolution of target tracking in wiressless sensor network.To implement the target tracking, the issues are studied in the thesis include: (1) Correcting distance measurement errors in self-localization of nodes before tracking; (2) During the target tracking, based on time synchronization and dynamic clustering, implementing the distributed particle filter algorithm, dealing with problem of asynchronous measurement. (3) Validation of the target tracking algorithm. Mainly works as follows:In order to improve the self-localization accuracy of nodes, an error correct method in TDOA distance measurement based on TinyOS is put forward. In this method, the temperature compensation of average path is adopted to reduce affection of ultrasonic velocity caused by environmental temperature. The words asynchronizaton and time compersation are used to repair system spending such as task release, command calls and interrupt handling in operating system. Least-square principle is used in linear fitting of the measure results. Implemented in TinyOS, the method is practical, simple and effective. In the effective measure distance, experiment results show the revised errors can be less than 1%.According to demand of time registration during target tracking, an embedded on-demand time synchronization algorithm is put forward. The algorithm is based on "post-facto". The synchronization operation is drived by target tracking process, which can reduce the energy consumption in network due to periodic time and the synchronization error over time. By using the two-way message exchange as the core mechanism of synchronization and embedding synchronization packets in target tracking messages, the messages exchanged among nodes and the data traffic in network are decreased. Comparing with TPSN algorithm, results show that the algorithm has good performance in precision, and can effectively reduce the energy consumption of communication during target tracking in wireless sensor network.Aiming at the requirements for node collaboration and management during target tracking in wireless sensor networks, a dynamic clustering algorithm is proposed. Based on greedy algorithm, the cluster head is selected using competition mechanism and RSS estimation. And the covering area of cluster head is divided functionally according to the communication distances of target and cluster head. The recruitment and state adjustement for member nodes are completed as well. Simulation results show that this method can organize nodes to track target efficiently, balance the load of network nodes, and has good robustness.Particle filter takes a lot of computation, which leads to the filter unable to carry in common sensor nodes. A parallel particle filter algorithm is put forward for handling this promblem. Particle set is divided into several subsets according to the number of nodes which are distributed to each node in cluster. The subsets are sampled, weighted, resampled and aggregated parallelly on nodes. Local state is estimated by cluster head after the aggregation parameters were uploaded. Comparing with the centralized particle filter, the algorithm has good performance in tracking precision andless requirement in computing capacity. It can also balance computational load among nodes effectively.In order to solve the asynchronous measurement problem of multi-sensor in target tracking, a trust-judging based density assisted particle filter (TJ-DAPF) algorithm is presented. First, through introducing time asynchronous parameters to modify the measurement equation, asynchronously measurement model is established in tracking cluster. Then, the density-assisted particle filter is used to estimate the target states and parameters. In the filtering process, the operation of parameters estimation is controlled dynamicly by judging the credibility of parameter estimation results. Compard with DAPF, TJ-DAPF has similar tracking accuracy and performance, but lower calculation complexity and higher calculation efficiency.According to the demand of researching on indoor target tracking of wireless sensor networks, a tracking experimental system is designed. First, using the experimental method, the indoor wireless signal propagation characteristics are analyzed. Then, the structure of tracking experimental system is given, expounding the design details of nesC programming and software of monitoring and management platform. Finally, the performance of system is analyzed by test results. This system is universal and expansile, which can lay the foundation for algorithm verification and applications research of WSN.