Nonzero-Sum Games for IoT Systems: An Event-Triggered Reinforcement Learning Approach

This paper investigates Nash equilibrium seeking for non-zero-sum games (NZSG) in internet of things systems. A resilient event-triggered control mechanism was proposed, which reduces false triggers under Denial-of-Service attacks while simultaneously lowering the communication and computational load of the system. Moreover, reinforcement learning is employed to seek the Nash equilibrium point within the NZSG framework, where both cooperative and competitive interactions among the distributed agents coexist. To enhance critic learning process, an experience replay method is introduced to relax the restrictive persistent excitation condition. Finally, the effectiveness of the proposed scheme is demonstrated through simulation studies conducted on microgrids.