Secure formation control for multi-UAV systems based on zero-sum game

To achieve secure formation control of unmanned aerial vehicles (UAVs), this paper proposes a fixed-time convergent reinforcement learning algorithm based on a zero-sum game under an event-triggered mechanism. Considering that actuators may be subjected to false data injection (FDI) attacks, which can cause UAVs to deviate from their trajectories, a zero-sum game framework is constructed. Within this framework, a security controller aims to minimize a performance function, while the attacker endeavors to maximize this function. The security controller aims to minimize a performance function, while the attacker endeavors to maximize this function, since obtaining the solution to the optimal security tracking controller is related to the Hamilton-Jacobi-Isaacs (HJI) equation associated with the zero-sum game, a reinforcement learning algorithm is proposed to approximate the numerical solution of the HJI equation. The proposed algorithm achieves fixed-time convergence, and its proof is given in this paper. Finally, the simulation results confirm the effectiveness of the presented methodology.