Fully-distributed Consensus Control of Multi-agent Systems Under Stochastic Hybrid Attacks on a Directed Graph

Abstract

This study presents a consensus control method for generic linear multi-agent systems (MASs) subject to stochastic deception attacks on actuators and random denial of service (DoS) attacks over communication channels. In this case, the MASs are communicating over a compromised network where the attackers are performing random DoS attacks and injecting data into the control inputs. A fully-distributed robust-adaptive consensus protocol is designed for both types of stochastic cyber-attacks by taking into account the immediate partial information per node rather than global information and by accounting for the signals added by adversaries. To the best of the authors’ knowledge, a fully-distributed consensus control approach with adaptive gains (without requiring any global knowledge for the design) under hybrid cyber-attacks of stochastic nature over a directed graph has been addressed for the first time. Finally, simulation results are presented to further illustrate the theoretical results’ effectiveness.