Integral Event-Triggered Based Secure Formation Control for Multi-Quadrotors Under False Data Injection Attack

This paper addresses an integral event-triggered secure formation control approach for unmanned aerial vehicles (UAVs) subject to false data injection (FDI) attacks. The formation-learning closed loop is shown to be semiglobally practically uniformly ultimately bounded, while the critic-weight error is semiglobally practically fixed-time UUB. To handle FDI disturbance on actuators of UAVs, a zero-sum differential game framework is utilized to formulate the optimal control problem against the attack process, where secure formation controllers seek to improve the system performance of formation and the attacker’s objective is the converse. Additionally, for obtaining the optimal secure formation control policy at the Nash equilibrium, which requires solving the coupled Hamilton–Jacobi–Isaacs equation associated with the game, a critic-only reinforcement learning (RL) scheme with online weight adaptation was developed to approximate the optimal policy, using a fixed, preloaded, full-row-rank historical regression dataset. In order to conserve computational resources and reduce network transmission burdens, a conventional event-triggered control is initially proposed along with a more efficient integral form developed in this study. The critic-weight error is shown to be semiglobally practically fixed-time UUB, while the formation and critic-weight states are shown to be semiglobally practically UUB. Zeno behavior is also excluded. A simulation experiment is provided to demonstrate the efficacy of the devised scheme.