ON A DETECTABILITY CONCEPT OF DISCRETE-TIME INFINITE MARKOV JUMP LINEAR SYSTEMS Eduardo F. Costa Jo˜ ao B. R. do Val ,1 Marcelo D. Fragoso ∗∗ UNICAMP - FEEC, Depto. de Telem´ atica, C.P. 6101, 13081-970, Campinas, SP, Brazil ∗∗ LNCC/CNPq, Av. Getulio Vargas 333, Quitandinha, 25651-070, Petr´ opolis, RJ, Brazil Abstract: This paper introduces a concept of detectability for discrete-time infinite Markov jump linear systems that relates the stochastic convergence of the output with the stochastic convergence of the state. It is shown that the new concept generalizes a known stochastic detectability concept and, in the finite dimension scenario, it is reduced to the weak detectability concept. It is also shown that the detectability concept proposed here retrieves the well known property of linear deterministic systems that observability is stricter than detectability. Keywords: Stochastic jump processes, Markov parameters, Markov models, systems concepts, observability. 1. INTRODUCTION This paper is concerned with the discrete-time Infinite Markov jump linear system (MJLS) defined in a fixed stochastic basis (Ω, F , (F k ), P ) by Ψ : x(k + 1)= A θ(k) x(k), k 0, y(k)= C θ(k) x(k), x(0)= x 0 , θ(0)= θ 0 (1) where x and y are the state and the output variables, respectively. The mode θ is the state of an underlying discrete-time Markov chain Θ = {θ(k); k 0} taking values in S = {1, 2,... } and having a stationary tran- sition probability matrix P =[ p ij ], i, j Z. θ 0 S is a random variable for which μ i = P(θ 0 = i), i S , and x 0 is a second order random variable. It is assumed that matrices A i and C i , i S , belong respectively Research supported in part by FAPESP, Grant 98/13095-8, by CNPq, Grant 300721/86-2(RN) and by the PRONEX Grant 015/98 ’Control of Dynamical Systems’ 1 Corresponding author. Email: jbosco@dt.fee.unicamp.br Fax: 55-19-3289 1395 to the collections of real matrices A =(A 1 , A 2 ,... ), dim(A i )= n × n, and C =( C 1 , C 2 ,... ), dim( C i )= q × n, for which sup iS A i < and sup iS C i < . We also assume that x(k) and θ(k) are observed at each time instant k. When one deals with system Ψ, the usual detectability concept is the stochastic detectability (S-detectability), which is a dual concept of stochastic stabilizability; see (Costa and Fragoso, 1995) in the same setting of this paper, or (Fragoso and Baczynski, 2001) in the continuous time case, or (Costa, 1995) and (Morozan, 1995) in the finite dimension case. However, the S- detectability concept presents the drawbacks pointed out in the sequel. Consider the weak observability (W-observability) concept that follows from the extension of the finite state space case, see Section 4. It appears in (Costa and do Val, n.d.b), (Costa and do Val, 2001) and (Morozan, 1995), and it is more general than other observability concepts for MJLS, like the ones appear- Copyright © 2002 IFAC 15th Triennial World Congress, Barcelona, Spain