Privacy-Preserving Average Consensus in Finite Time

Abstract

Average consensus has wide applications in distributed networks ranging from computing and control, where each node continually receives information from its neighbors and updates its state to reach the average. The existing average consensus algorithm may result in the disclosure of private information about the nodes as each node sends explicit state values to its neighbors. In this paper, we propose a state decomposition based privacy-preserving average consensus algorithm that allows a strongly connected directed network system to compute the accurate average value in a finite number of time steps while each node can avoid its initial state value being disclosed. More specifically, we improve on the classical ratio consensus algorithm, in which the average value can be computed in a finite number of time steps by the final value theorem. Numerical examples verify the effectiveness of our algorithm.