Secure Consensus of Multiagent Systems Against False Data Injection Attacks: A Distributed Model Predictive Control Approach

This conference article gives a secure distributed predictive control solution to resolve the consensus problem in false data injection (FDI) attacked multiagent systems. Cyber attacks corrupt the system outputs, which mislead the controller to receive incorrect states. Control signals based on misinformation amplify the adverse impacts on the system dynamics. To defend against unknown cyber attacks, a multivariate observer is first proposed to separate and estimate uncompromised states and FDI attacks to attenuate the effect of compromised outputs. Based on the estimated state, a predictive controller with a new cost function and a new terminal constraint is designed. Finally, a simulation of the formation cart verifies the results of this article.