State Estimation for 2-D Roesser Model with Multiple Network Induced Phenomena

This paper is concerned with the state estimation problem for a class of 2-D Roesser model, where the multiply network induced phenomena including time delay, sensor saturation and missing measurement are simultaneously considered under the network environment. By introducing a 2-D stochastic variable, these network induced phenomena are successfully described by an uniform measurement model. The time delay in the measurement and the probability of the stochastic binary sequence characterizing the time delay are all assumed to be time-varying. With help of the obtained measurement, the gain-scheduling state estimator for the 2-D Roesser model is constructed. In order to enhance the adaptability to the stochastic factors in network environment, the gains are designed to be two portions of constant and variation, which thus make the state estimator can be scheduled on line. By resorting to a sequence of mathematical operations, the error dynamics related to the original 2-D system and the state estimator is established. The mean-square asymptotical stability of the estimation error system is analyzed on basis of the Lyapunov stability theorem and the stochastic analysis techniques. The acquired stability criteria in terms of time-varying form are effectively converted to a finite number of line matrix inequalities (LMIs). Feasibility of the studied state estimation technique is confirmed by its good performance on the numerical example.