Distributed Set-Membership Filtering for Nonlinear Time-Varying Systems With Dynamic Coding-Decoding Communication Protocol

This article studies the distributed set-membership filter design problem for a class of general nonlinear discrete time-varying systems whose measurements are collected by a multitude of distributed sensors. A scheme is exploited to estimate the state by using the measurements from not only the local sensors but also the neighboring ones, whose information is propagated by codewords through the communication network on basis of the fixed topology. A dynamic coding-decoding technique is applied during the data transmission process among sensing nodes. The purpose of the investigation is to limit the estimation error into a predetermined allowed region characterized by an ellipsoid, in spite of the existence of the so-called unknown-but-bounded (UBB) disturbances. Sufficient conditions are established for the existence of required filter, and the filtering gains can be determined by virtue of solving certain set of matrix inequalities. A suboptimal problem is discussed to guarantee the locally optimal filtering performance. Finally, the provided theoretical framework is demonstrated by an illustrative numerical example.

Automated summary

This article investigates the design of a distributed set-membership filter for a class of general nonlinear discrete time-varying systems with measurements collected by distributed sensors. A dynamic coding-decoding technique is applied during data transmission among sensing nodes to limit estimation error. Conditions for the existence of the required filter and determination of filtering gains are established through solving matrix inequalities, and a suboptimal problem is discussed to guarantee locally optimal filtering performance. The provided theoretical framework is demonstrated through a numerical example.