Fault-Tolerant Consensus Control With Privacy-Preserving Virtual Layer for Heterogeneous Multi-Agent System

The paper investigates the privacy-preserving cooperative tracking problem under actuator faults, system uncertainties and unmatched disturbances for the heterogeneous multi-agent system (MAS). The virtual system is constructed to preserve the initial states and transient processes with the dynamic perturbation method against internal and external eavesdroppers. The real system achieves tracking to the virtual system and heterogeneous consensus by the adaptive fault-tolerant (FT) controller. It is demonstrated that for groups of agents with different initial states, the designed privacy-preserving algorithm can achieve consistent output to the eavesdropper while ensuring ultimate consensus. Compared with the existing results in the privacy-preserving problem, the requirements for an accurate fault-free model are avoided by the hierarchical design approach. Simulation examples validate the effectiveness of the suggested control scheme.
Note to Practitioners
—The study is motivated by the privacy-preserving problem of communication between MASs, and it applies to groups like UAVs, vehicles and robots that need to exchange information to achieve common goals. There are hostile parties who eavesdrop on messages sent between agents via network interception in order to get sensitive private information. To prevent eavesdroppers from obtaining privacy of the sent information and achieve the agent group’s common objective, communication information with perturbation and corresponding control mechanisms are established. The information protected by the paper is the agent’s initial states and transient process, and the genuine states will be wrapped with dynamic perturbation signals. However, due to the interference of the real measuring mechanism and the actuator faults, precisely measuring and controlling the genuine state is challenging. As a result, this research proposes a virtual system with privacy-preserving performance. The disguised signals of virtual system are communicated between agents, while the real system is designed to track the virtual system while suppressing disturbances and faults. Future studies will concentrate on privacy preservation with security constraints.