Robust \(\mathcal{H}_\infty\) filter design with pole constraints for discrete-time systems (Q5952978)

The robust \(H_{\infty}\) filtering problem consists in the design of a stable linear filter for an uncertain system where the noise sources are assumed to be arbitrary signals with bounded energy. The problem of robust filtering is described mainly by Riccati equations parametrized by positive scales with norm-bounded uncertainty [see, for example, \textit{L. Xie, C. E. de Souza} and \textit{Y. Wang}, Int. J. Robust Nonlinear Control 6, No. 4, 297-312 (1996; Zbl 0851.93030)]. \textit{H. Lee} and \textit{M. Fu} [IEEE Trans. Signal Process. 45, 2338-2350 (1997)] proposed equivalent linear matrix inequality (LMI) conditions for both discrete- and continuous-time systems. The problem of robust filtering for uncertain systems in convex bounded domains has been investigated recently via LMI methods. Strictly proper filters were proposed by \textit{J. C. Geromel} and \textit{M. C. de Oliveira} [Proc. 37th IEEE Conf. on Decision and Control, Vol. 1, Tampa, FL, 146-151 (1998)] (continuous-time systems), \textit{J. C. Geromel, J. Bernussou, G. Garcia} and \textit{M. C. de Oliveira} [Proc. 37th IEEE Conf. on Decision and Control, Vol. 1, Tampa, FL, 632-637 (1998)] (discrete-time systems). This paper presents sufficient LMI conditions for the robust \(H_{\infty}\) filtering design for discrete-time systems with polytope type uncertainty. The main purpose is to obtain a stable and proper linear filter such that the filtering error system remains robustly stable within a prescribed \(H_{\infty}\) level. Sufficient conditions for the existence of such a robust filter are proposed in terms of linear matrix inequalities, which can be efficiently solved by means of high performance convex optimization procedures. Furthermore, additional pole constraints can be imposed to the filter dynamics.