Fusion Estimation For Sensor Network Systems With Correlated Measurements And Oosms

With the development of the concepts"Internet of Things"and"Experience China", the broad prospects of the sensor nets is emphasized again in many domains, such as military, industry, agriculture and so on. The research of the sensor nets has been widely concerned both in home and abroad. One of the most important research fields is on the information fusion technology, the research of which is proposed important theoretic significance and extensive application.Considering the physical condition of sensor nets, the fusion systems based on sensor networks will face several problems, when it is collecting and processing the information from local sensors. Such as the correlation of local sensor information, the sampling asynchronously, and the random delay disordered arrival of information transmission, and so on. In fact, how to design fusion algorithms under the constrained conditions have become the key problems in network information fusion field. Therefore, what is mainly proposed in this paper is the research of fusion estimation methods with restrictions of correlation noises and delay measurements in the sensor net systems.(1) An optimal recursive fusion estimation algorithm is proposed in this paper in the sense of Linear Minimum Mean Square Error (LMMSE). This method can solve the filtering problem for the synchronous system with the measurement noises correlated with the process noise one time step apart and the measurement noises auto-related optimally.(2) An optimal recursive fusion method is proposed in the central frame for the multi-sensor asynchronous sampling system with correlated noises. This proposed asynchronous fusion method transforms the asynchronous fusion problem into a pseudo-synchronous fusion problem, firstly. Then, deal with the summation correlation of the system noises with a technology named decorrelation. Finally, the optimal estimation in the sense of LMMSE can be worked out by using the thought of recursive fusion.(3) A multi-sensor system with multiple one-step-delay OOSMs in the sensor network is considered in this paper. And a novel OOSM directly sequentially update algorithm is proposed. The novel method extends the optimal A1 algorithm which is one of the important filtering methods proposed by Y. Bar-Shalom to more general situation with multiple OOSMs in a fusion period. And the algorithm is optimal in Linear Minimum Mean Square Error (LMMSE).(4) The variables in Kalman filter in the linear time invariant (LTI) systems can be computed off line. Take advantage of this feature of the LTI system, a novel representation of Kalman filter is proposed in this paper, which is the sum of the measurements and the initial state. A hybrid filter is proposed in this paper for the LTI system with multiple OOSMs. The filtering algorithm considered the essential characters of the LTI system and LMMSE, perfectly.