Variance-constrained resilient H∞ filtering for time-varying nonlinear networked systems subject to quantization effects

Abstract

This paper deals with the resilient variance-constrained H∞ finite-horizon filtering problem for a class of discrete time-varying nonlinear networked system with quantization effects. The nonlinearity enters the system in a probabilistic way that is characterized by a binary sequence with known distribution. The system parameters under investigation are all time-varying and the randomly occurring filter gain variations are modeled by utilizing a random variable obeying prespecified binary distribution which is uncorrelated with other stochastic variables. The nonlinearities and exogenous disturbances we adopt are non-zero mean, which makes the variance analysis become more difficult. Furthermore, the quantization effects are also taken into account to describe the unavoidable constraints imposed on the signal during the transmission in networked systems. Sufficient conditions are established for the finite-horizon filter guaranteeing the constraints imposed on both estimation error variance and H∞ specification. By means of the recursive linear matrix inequality approach, the algorithm for computing the desired filtering gains is provided. Finally, a numerical illustrative example is used to verify the effectiveness of the proposed design method.