Concentrated differentially private average consensus algorithm for a discrete-time network with heterogeneous dynamics

This paper presents a privacy-preserving average consensus algorithm for a discrete-time network with heterogeneous dynamic nodes in the presence of Gaussian privacy noises. Rényi divergence is used to measure the privacy, and a distributed algorithm is proposed for each node in the network to protect the initial output state and ensure consensus almost surely. The convergence rate of the proposed algorithm relates to the communication topology, dynamics of systems, and decaying rates of privacy noises. Moreover, by increasing neighbors of nodes in the network, the proposed algorithm can strengthen preservation. To demonstrate the theoretical results, a numerical example is carried out on a network of one hundred nodes.