Model-based resilient control for a multi-agent system against network intrusion
Abstract
Cyber physical systems (CPS) integrate information technology and physical entities. The computerization of critical infrastructure such as control systems means that several real life physical systems interact with information technology (IT) to an extent which makes them susceptible to malicious attacks. These attacks cannot be addressed like conventional cyber-attacks on IT systems. The impact of cyber-attacks on the physical infrastructure of cyber-physical systems is not well studied and as such opens up a whole new field of studies that researchers are only now looking at. While IT security places emphasis on data accuracy for example, control systems emphasize timely and accurate transmission of control signals with no delays or retransmissions as these can have detrimental effects on the system. This motivates the need for resilient control algorithms to protect such critical infrastructure from malicious attacks. In this research, a cyber-physical system is implemented as a team of networked autonomous agents with formation control by cyclic pursuit as their team objective. In the cyclic pursuit formation control strategy a distributed control algorithm is employed in which each agent depends only on information received from its next neighbor to cooperatively carry out the mission. A model-based resilient control algorithm is developed that enables the team of autonomous agents to accomplish their task even in the presence of a malicious attacker disrupting inter-agent communication. An experimental validation process for the model-based resilient control strategy is also implemented. Using a mathematical model of the dynamic system representing a formation task, the algorithm predicts dynamical states at any given time, detects anomalies in states due to corrupted data, and corrects the behavior of agents.
Subject Area
Computer Engineering|Information Technology|Robotics
Recommended Citation
Esther Marion Amullen,
"Model-based resilient control for a multi-agent system against network intrusion"
(2015).
ETD Collection for Tennessee State University.
Paper AAI10003137.
https://digitalscholarship.tnstate.edu/dissertations/AAI10003137