HMM-based dissipative filtering for Markov jump neural networks under event-triggered scheme and stochastic cyberattacks

This paper investigates the design of an asynchronous dissipative filter for a class of discrete-time Markov jump neural networks (MJNNs) under event-triggered schemes (ETS) and stochastic cyberattacks. Since the mode information of the system mode is not easily acquired by the filter, the hidden Markov model (HMM) is employed to depict such kinds of asynchronous characteristics. The transmitted data meets specific event-triggering conditions, which can alleviate the communication burden. Owing to network vulnerabilities, two kinds of cyberattacks, deception attacks (DAs) and denial-of-service (DoS) attacks, are considered in the transmission channel. By exploring the ETS method and the stochastic cyberattacks property, a hidden networked MJNNs model with network-induced delay and hybrid cyberattacks is proposed for the first time. Sufficient conditions are derived to ensure that the resulted hidden filtering error system with hybrid cyberattacks is finite-time bounded (FTB). Based on this, a criterion for finite-time exponential dissipativity (FTED) is established and an event-triggered and asynchronous secure filter is designed. Finally, two numerical examples are presented to verify the validity of the proposed filter design scheme.