Asynchronous dissipative filtering for nonhomogeneous Markov switching neural networks with variable packet dropouts.

Abstract

This work focuses on the problem of asynchronous filtering for nonhomogeneous Markov switching neural networks with variable packet dropouts (VPDs). The discrete-time nonhomogeneous Markov process is adopted to depict the modes switching of target plant, where time-varying transition probabilities are revealed by utilizing a polytope technology. By means of the Bernoulli distributed sequence, the randomly occurring packet dropouts are presented, where VPD rates are mode-dependent and remain variable. Unlike the existing results, the hidden Markov model scheme is formulated to describe the asynchronization between nonhomogeneous neural networks and filter, and resilient filters are presented, which makes the designed filters more general. Eventually, a simulation example is established to verify the effectiveness of the developed filter scheme.