Consensus control for nonlinear multi-agent systems subject to deception attacks

This paper deals with the consensus control problem for a class of discrete time-varying stochastic multi-agent systems with stochastic nonlinearities and deception attacks. A novel definition of quasi-consensus is proposed to reflect the consensus process where, at each time instant, all the agents are constrained to stay within a certain ellipsoidal region. The measurement output available for the controller is not only from the individual agent itself but also from its neighboring ones according to the given topology. A new deception attack model is proposed, where the attack signals are injected by the adversary to the measurement data during the process of information transmission via the communication network. By using a set of recursive matrix inequalities, sufficient conditions are derived for the existence of the desired control scheme for achieving the quasi-consensus. Subsequently, an optimization problem is formulated to seek the locally best consensus performance.