Privacy-preserving resilient bipartite consensus of multi-agent systems: A differential privacy scheme

Abstract

This paper addresses the issue of differential privacy-preserving in multi-agent systems (MASs) with the existence of misbehaving agents and antagonistic interactions over a signed digraph. Even with the existence of a maximum of $f$ faulty agents within the network, non-faulty agents pursue resilient bipartite consensus, with the requirement that their initial conditions can fulfill differential privacy. To this end, we propose the differentially private absolute weighted mean subsequence reduced (DP-AW-MSR) algorithm. Under the structurally balanced signed digraph with sufficient connectivity in terms of robustness, three essential properties of this algorithm are characterized: resilient bipartite consensus, accuracy and differential privacy. Numerical simulation is given to illustrate the effectiveness of our findings.