Privacy-Preserving Average Consensus via Virtual Agent Introducing Mechanism

In order to solve the problem of agent privacy information leakage during the realization of multi-agent consensus, a novel privacy-preserving method is proposed to avoid the leakage of the initial and real-time state values of agents. Firstly, a virtual agent is introduced to participate in the original multi-agent system, and the communication topology is reconstructed to establish a normal communication channel. Secondly, in order to satisfy that the final consensus value does not change, a qualification condition is given. Thirdly, the privacy proof is carried out for the internal honest-but-curious agent and the external eavesdropper respectively, and the relationship between the initial state value and the weight value satisfying the privacy protection condition is obtained. The simulation section shows in detail the performance of consensus and privacy.