Research Of Wireless Sensor Network Information Fusion And Target Tracking Problem

With rapid development of wireless telecommunication, integrated circuits, sensor information technology, micro-electromechanical systems and other technologies, wireless sensor network emerges as the era requires. Wireless sensor network is composed of a large number of tiny sensor nodes, which constitute a multi-hop and self-organized network system. Its goal is to perceive, acquire and deal with the target in the monitoring area collaboratively and to send the results to the observer. Because of the advantages in cost, size and flexibility of wireless sensor network, it is widely used in national security, environmental monitoring, traffic management, medical care, anti-terrorism, anti-disaster and other fields. Therefore, wireless sensor network, recognized as one of the IT technologies which have unprecedented impact on human life in 21st century, has significant scientific values and broad prospects of applications.In this thesis, wireless sensor network is combined with information fusion algorithm to improve the accuracy and efficiency of the collected information based on the characteristics of wireless sensor network and the concepts and features of information fusion. In the meantime, this paper introduces a new information fusion mechanism in order to reduce the energy, decrease the storage and network resources. This distributed tracking mechanism is based on the strategy that dynamic clusters in wireless sensor network are organized according to the current position of the target. For linear system, three fusion algorithms are applied into clustering mechanism and compared with the algorithms without clustering mechanism respectively. According to the simulation results, the effectivity and reliability of this algorithm have been verified. Additionally, Grubbs guideline can be used to eliminate the abnormal data to get better accuracy when packet-dropping exists in wireless sensor network.For nonlinear system and energy problem, Unscented Kalman Filter based on clustering mechanism is proposed. Simulation results show that this algorithm has a good tracking performance, reduces network conflicts and congestion effectively, cuts down the consumption of wireless sensor network and prolongs the life period of wireless sensor network, compared to the traditional Extended Kalman Filter and Unscented Kalman Filter.