Memory-based dynamic event-triggered filtering for fuzzy Markov jump systems under quantisation and deception attacks

This paper is concentrated on the issue of neural-networks-based filtering for fuzzy Markov jump systems with memory-based dynamic event-triggered protocol under dynamic quantisation and stochastic deception attacks. In the perspective of decreasing the consumption of communication resources and obtaining better performance, a novel memory-based dynamic event-triggered strategy is proposed, which the event-triggered condition is related to a number of the previous triggering signals and the quantisation variable. Besides, the random deception attacks are characterised by another Markov chain. For observing the mode of plant and stochastic deception attacks, the dual hidden Markov models are applied. By leveraging neural-network-based techniques, the false signals injected by attacks are estimated, and their adverse effects are effectively mitigated through the designed filter, resulting in more accurate state estimation. The mean-square exponentially ultimately bounded condition for filtering error has been adequately discussed. At the end, the effectiveness of the proposed filtering strategy is verified through a numerical example.