Observer-based adaptive neural asynchronous H ∞ Control for fuzzy Markov jump systems under FDI attacks

The issue of observer-based adaptive neural asynchronous H infinity control for discrete-time networked fuzzy Markov jump systems (FMJSs) under the influence of fake data injection (FDI) attack and random packet loss is investigated. Given modeling accurate systems in cyber- physical systems (CPSs), two independent Markov chains are merged into a single joint Markov chain using FMJSs and mapping techniques. Owing to the fragility of the network layer, this paper investigates the problem of FDI attacks by external attackers and the problem of packet loss caused by internal channel redundancy and network congestion. In this paper, neural networks (NNs) techniques are used to approximate FDI attacks information and model random packet loss in the feedback channel using the Bernoulli distribution. An output feedback asynchronous control is presented to mitigate the influence of attacks and data dropouts on the system. The H infinity performance of the closed-loop system is ensured by solving matrix inequalities to obtain the output feedback controller and observer gains. In addition, an algorithm to determine the optimal parameters is proposed. Lastly, the validity of the presented control methods is illustrated through a simulation case.