Double-layer game-based optimal event-triggered consensus control for stochastic multiagent systems against deception attacks

This paper studies the optimal event-triggered control issue for stochastic multiagent systems (MASs) against deception attacks in a double-layer game framework. First, estimated outputs are obtained by the Bay- esian estimation method, thereby facilitating the output feedback consensus control for the stochastic MASs subject to sensor deception attacks. Subsequently, to obtain the local optimality of event-triggered condition (ETC) and the global optimality of the consensus control strategy, a double-layer game method is designed by defining a new cost function using adaptive dynamic programming (ADP) method. In zero-sum games, the controller and the ETC are viewed as two players with opposing interests. In graphical games, communication among all players is constrained by the topology. The proposed double-layer game scheme has demonstrated that, at the maximum triggering interval, the optimal control strategy, derived from the solution of the Hamilton equation, not only achieves Nash equilibrium but also serves as the best response to the control strategies for neighboring players in the sense of mathematical expectation. The simulation results are finally given to demonstrate the validity of provided technique.