Research On Fusion Estimate Algorithm Of Wireless Sensor Networks Based On Descriptor System

Wireless sensor network(WSN) is an emerging domain of science since the 1970 s. The applications of the WSN include the medical treatment, the military affairs, the national defense and the high-technology research and so on. Comparing with the sensor networks whose nodes obey the traditional system model, the sensor networks whose nodes obey the descriptor system model are more general and the estimation and fusion problem have attracted more attention correspondingly.For the sensor networks whose nodes obey the descriptor system model, a distributed estimation and fused algorithm is presented in this paper. Firstly, the singular value decomposition is used to transform the original system into two equivalent reduced order non-singular sub-systems, which yields a set of multi-rate systems with different state update rate, measurement update rate and estimate update rate by using the multi-rate sampling scheme. Furthermore, the multi-rate systems with random packet loss is modeled by a Stochastic system model. The local filtering estimator, the one-step predicted estimator and the corresponding estimation error co-variance matrices are then obtained. The fusion Kalman filtered estimator is further proposed with a fusion rule weighted by matrices. Our algorithm is simulated by Matlab. Numerical examples illustrated the effectiveness of the fusion Kalman filtered estimator and a bit more precision compared with the local estimator with a single sensor.