Event-Triggered Predefined-Time Consensus Control of High-Order Multi-agent Systems Subject to Deception Attacks

Deceptive attacks compromise the availability of accurate system states, posing challenges for the design of adaptive controllers and nonlinear multi-agent systems. These attacks introduce significant control errors, making it difficult to achieve consensus among multiple agents while effectively addressing uncertainties. To address the challenge of the true system state being unavailable for controller design due to deceptive attacks, a predefined-time consensus control is proposed against deceptive attacks. A predefined-time state observer is developed to compensate for the effects of these attacks. The controller of each individual subsystem utilizes estimates from its own and neighboring estimators. Additionally, a corresponding reasonable adaptive law is designed post-attack. This research proposes a novel predefined-time method incorporating an acceleration term. By avoiding complex calculations such as exponential functions or trigonometric functions, the method simplifies computations and reduces computational resource requirements. Compared to commonly used functions, the inclusion of an acceleration term results in a theoretical upper bound of settling time that is less conservative. This approach utilizes a flexible formation task strategy based solely on neighborhood information, thereby improving tracking and coordination performance. It imposes fewer restrictions while reducing the need for information processing and exchange. The designed control protocol and event-triggered mechanism are fully distributed. Simulation results validate the compensation mechanism and predefined-time consensus control method can reduce control time costs and achieve a 25% reduction of estimation error compared to prescribed-time ones, and a 20% reduction of estimation error under deception attacks compared to traditional predefined-time ones.