Event-triggered synchronisation of master-slave neural networks with interchange attack

This paper investigates the event-triggered synchronisation of master-slave neural networks subject to interchange attack. In the proposed strategy, the synchronisation output error is sampled by a sensor, and the data is transmitted to the controller via an event-triggered mechanism (ETM). The proposed ETM reduces the amount of transmitted data, preventing communication congestion. In the communication channel between the sensor and the controller, a novel type of cyber-attack, termed an interchange cyber-attack, is considered. This represents the first introduction of such an attack in this system configuration. The attack's incidence is modelled as a Bernoulli stochastic process, and it is shown to possess the capacity to destabilise the system and compromise overall control performance. To mitigate these cyber threats and external disturbances, an output-feedback H∞ controller is proposed. The control law is designed to preserve closed-loop stability, enhance robustness against the interchange attack, and ensure a prescribed disturbance attenuation level. By constructing a tailored Lyapunov–Krasovskii (LK) functional and applying LK stability theory, sufficient conditions in terms of linear matrix inequalities (LMIs) are derived to guarantee master-slave synchronisation of the neural networks. Finally, two numerical examples are presented to demonstrate the effectiveness of the proposed method.