Differentially Private Average Consensus of Multi-Agent Systems via Privacy-Based Dynamic Self-Triggered Mechanism

A differentially private average consensus (DPAC) problem for multi-agent systems (MASs) with a dynamic self-triggered mechanism is reported in this paper. With the privacy-based dynamic self-triggered mechanism (PDSTM), a DPAC control algorithm is proposed. An adaptive coupling gain related to privacy measurement errors among agents is introduced in the PDSTM, and the triggered threshold evolves adaptively. The proposed DPAC algorithm enables agents to preserve privacy on initial states of agents from information leakage while enhancing overall network's execution efficiency. The analysis of convergence and accuracy of the algorithm are provided. Moreover, a differential privacy analysis is presented, and the privacy analysis allows agents to flexibly select their own privacy levels. Finally, simulations are conducted to verify the effectiveness of the proposed algorithm.