Average Consensus Control of Multi‐Agent Systems With Hybrid Privacy‐Preserving Schemes

ABSTRACT In this paper, the problem of privacy‐preserving average consensus (PPAC) is investigated for discrete multi‐agent systems (MASs) with hybrid privacy‐preserving mechanisms. To this end, a hybrid privacy mechanism consisting of noise injection and quantization‐based encryption‐decryption rules (EDR) is first constructed to protect the privacy of the initial state of the agent, where the utilization of a dynamic private key enhances the private capability. Then, a PPAC scheme is introduced by resorting to the constructed hybrid mechanism. By means of matrix properties, a sufficient condition is provided to determine the controller gain and the noise parameters, ensuring the performance of average consensus. Additionally, the size of the encrypted data is analyzed in the framework of the hybrid privacy mechanisms. Furthermore, the privacy of the initial state is demonstrated with respect to both honest‐but‐curious agents and external eavesdroppers. Finally, the effectiveness of the proposed PPAC scheme is illustrated through a numerical example.