Adaptive safe resilient consensus protocol of leader–follower multi-agent systems subject to hybrid false data injection cyberattacks and directed network topology

Abstract

This article introduces a safe consensus protocol of second-order unknown nonlinear multi-agent systems (MASs) subjected to hybrid false data injection cyberattacks (FDICAs), external disturbance, and directed network. All followers can be prone to FDICA and external disturbance. In an MAS, agents measure their state variables by sensors and then send them to neighbor agents through wireless systems. Therefore, both sensors and wireless systems can be prone to FDICAs. These attacks can deteriorate the closed-loop performance, reliability, and safety or even cause instability since they can easily propagate errors among the agents. So that, it is vital to design control policies enhancing resilience against FDICA. It is assumed that different bounded attacks are applied to position and speed channels. Our purpose is to plan a global safe adaptive resilient consensus protocol to guarantee the asymptotic stability. To this aim, by employing artificial neural networks (ANNs), the nonlinear dynamics of each agent is estimated. Afterward, a safe adaptive control law is planned to compensate for the effects of nonlinearity, disturbance, and FDICA to achieve the consensus. By means of the Lyapunov’s second theory, the global asymptotic stability (GAS) of the system is proved. Different numerical results confirm the theoretical investigations and illustrate the efficiency of the presented approach on controlling agents in the proposed second-order nonlinear MAS.