Model-based resilient control for a multi-agent system against Denial of Service attacks

Abstract

The computerization of critical infrastructure such as control systems means that these systems interact with information technology (IT) to an extent that makes them susceptible to malicious attacks. While IT security places emphasis on data accuracy easily attainable through simple error correction schemes such as packet re-transmission, control systems emphasize timely and accurate transmission of control signals in which delays or re-transmissions can have detrimental effects on the system. This motivates the need for resilient control algorithms that guarantee normal operation of critical infrastructure subject to malicious attacks and disturbances both at the physical layer and communication layer. In this paper, a team of networked autonomous agents whose collective objective is formation control is used to represent a cyber-physical system. A distributed formation control algorithm in which each agent depends only on its local information and that received from one neighbor to cooperatively carry out the group mission is employed. We develop a model-based resilient control algorithm that enables the team of autonomous agents accomplish their formation task even in the presence of a malicious denial of service (DOS) attack disrupting inter-agent communication. The technique is demonstrated through a laboratory experiment with 6 pioneer 3DX robots.