Research Of Distributed Estimation Algorithm Based-on Consensus

Wireless sensor networks (WSN) are the international research hotspots, which are interdisciplinary and have very broad application prospects. In the near future, WSN will effect on all aspects of society, life and production. In recent years, many researchers have concerned about the application of WSN to monitor a particular area or to track a target. As the sensor nodes are limited by their own energy, storage and processing capacity, many traditional estimation algorithms, such as central and distributed, can not be applied to complex environment of WSN. In this paper, the background is the application of WSN, and we focus on the research of consensus based distributed estimation algorithm.First of all, we study on the basic principles of consensus based distributed estimation algorithm, and then focus on a kind of classical Kalman filter algorithm. In the practical application, since WSN has the problem of packet loss, and to minimize the steady-state mean square estimation error, we propose a optimal consensus filter algorithm with the packet loss, and use the matrix theory to analysis the convergence. We use numerical simulation to verify the validation of the algorithm in several aspects:(1) the stability of the algorithm, and the new algorithm is compared with the classic filtering algorithm; (2) the estimation performance of the algorithm when the network topology is fixed and the system is stable or unstable; (3) the effect of network topology on the algorithm, and the accuracy of estimation is better as the communication radius increases. Finally, since the characteristics of time-varying of network topology when WSN is used in some conditions, the optimal consensus filter algorithm is applied to the dynamic network topology, and through numerical simulation, the results demonstrate that the new algorithm in this paper has better performance of estimation and robustness compared with the dynamic consensus filter algorithm.