Filtering, Smoothing, and Control in Discrete-Time Stochastic Distributed-Sensor Networks
In this chapter, the authors concentrate on decentralized estimation and control problems for linear discrete-time stochastic systems in a distributed-sensor network. They discusses an application of decentralized filtering, decentralized smoothing problems based on the two-filter form and a backward-pass fixed-interval smoother. The authors devote to the decentralized linear-quadratic-Gaussian control problem with a classic information pattern. Although two approaches based on local filtering and local smoothing are considered for solving the problems of decentralized smoothing, the authors are concerned only with the approach based on local smoothing. The authors assume that the estimation structure consists of a central processor (or station) and two local processors. An extension of the results to the case of m local stations is straightforward. Henceforth a different type of decentralized estimation and control must be examined further for the situation where the local measurement data are obtained sequentially by different sensors.