Observer-Aided Adaptive Consensus Control of Cyber-Physical Multi-Agent Systems With Hybrid Attacked Topology

This paper maidenly formulates a distributed secure consensus tracking control framework for a leader-following cyber-physical multi-agent system (CP-MAS) subject to malicious cyber-intrusions. The proposed consensus tracking control framework synergistically combines model reference adaptive control (MRAC) and super-twisting terminal sliding mode control (ST-TSMC), applied to a partially unknown CP-MAS, to enable cyber-resiliency with minimal tracking error and improved chattering performance. Due to stealthy deception attacks on sensor measurements, the states of follower agents are compromised. Thus, to minimize the conservativeness in realizing the proposed cyber-resilient control framework, a radial basis function neural network (RBF-NN)-assisted state observer has been formulated to estimate the uncompromised states of CP-MAS. Using graph theory and the Lyapunov argument, online adaptation laws have been derived that guarantee the boundedness of all signals and finite-time global asymptotic stability of consensus errors. Numerical simulations have been conducted on a group of four single-link robot manipulator agents subjected to hybrid cyberattacks. The presented qualitative and quantitative assessment demonstrates the proficiency and cyber-resiliency of the proposed hybrid control framework over MRAC, ST-SMC, and other reported methodologies in the face of unknown malicious attacks.