Resilient fixed-time synchronization of neural networks under DoS attacks

In this paper, the fixed-time synchronization of neural networks (NNs) with denial-of-service (DoS) attacks is investigated. A switching system with a stable subsystem and an unstable subsystem is utilized to describe the error dynamic behaviours between the master and slave NNs with DoS attacks. By virtue of the comparison principle, a novel fixed-time stability lemma for the switching system is established. The estimation of upper bound of setting time is provided. We further apply this lemma to investigate the resilient fixed-time synchronization of NNs subject to DoS attacks. By designing an appropriate controller, some sufficient criteria are proposed to guarantee the synchronization within a prearranged fixed time. The effects of DoS attacks on the settling time are discussed as well. Two numerical examples are given to show the effectiveness of main results and a potential application in the secure communication is provided.