Secured formation control for multi-agent systems under DoS attacks

摘要

This paper proposes a secured formation control strategy that ensures acceptable operation of a multi-agent system in the presence of Denial of Service (DOS) attacks. We consider a team of autonomous agents whose mission is to cooperatively achieve and maintain formation by cyclic pursuit where each agent depends only on information from a single neighbor for mission completion. Agents communicate through a wireless interface that is susceptible to adversarial attacks. The attacker can target any number of communication links with the aim of corrupting information transmitted between agents disrupting their collective mission. The proposed secured formation control strategy is based on continuous on-line system identification. An Identified model of the multi-agent system is locally implemented alongside the actual system for each agent enabling agents retrieve their desired dynamic behavior in the presence of a DOS attack. The proposed technique is demonstrated through a laboratory experiment with a team of 5 pioneer 3DX robots.