Distributed Filtering for Discrete Time Varying System Using Maximum Error First Protocol

Abstract

The problem of distributed filtering is designed for discrete time varying system considering sensor network with certain topological structure. Generally, the topology of sensor network is sparse in nature and the information are quite difficult to collect. In this paper, the sensor network is considered as discrete time varying system and the distributed recursive filtering problem is addressed. Stochastic nonlinearities are introduced into the system with gaussian inputs. Communication burden can be reduced by introducing Maximum Error First protocol (MEF) and also it saves the communication resource. The optimal distributed filter is designed with minimum variance for the considered discrete time varying stochastic nonlinear system. An upper bound of the error covariance matrix is arrived in terms of solving the Riccati type difference equation. The filter gain is derived in virtue of minimizing the upper bound of filtering error covariance. To deal with the sparsity of the sensor network a new matrix simplification technique is used. Results are derived by considering some of the sample values of time varying matrixes and nonlinear functions are simulated for the proposal and outputs are plotted in graph.