Security Event-Triggered Filtering for Delayed Neural Networks Under Denial-of-Service Attack and Randomly Occurring Deception Attacks

This paper investigates the problem of event-triggered filtering for delay neural networks (DNNs) under hybrid cyber attacks. First, a novel hybrid cyber attack model is established, which consists of randomly occurring deception attacks and Denial-of-Service attack. Secondly, by considering DNNs, event-triggered mechanism, and hybrid cyber attacks into a unified framework, a novel filtering error model is established by introducing two sets of random variables satisfying Bernoulli distribution and defining a switched filter system. Then, thanks to Lyapunov stability theory and linear matrix inequality (LMI) technology, a sufficient condition to ensure the exponentially mean-square stability is derived for this mathematical model. Among them, a reasonable boundary technique is selected to process the delay-dependent terms in the derivative of Lyapunov–Krasovskii functional. Moreover, a new sufficient condition are derived with LMI forms to co-design both the filter and the event-triggering parameters. Finally, an example verifies the effectiveness of the proposed method.