Average Consensus Under Privacy Constraint and Unreliable Network Environments

ABSTRACT Average consensus algorithms play an important role in distributed optimization, decision‐making, sensor fusion and dynamic system control. This article investigates the privacy‐preserving average consensus problem under various unreliable network environments. Different from the classical average consensus algorithm, a new robust average consensus algorithm is designed. The first contribution of this algorithm is its robustness to network adversities. That is, when network topology switches in terms of communication links, the agent joins or leaves, there is no need to reinitialize the consensus process, only a small amount of data need to be adjusted. The second contribution is to protect the initial states of agents from other curious agents. We strictly prove that the accurate convergence and privacy protection can be achieved under some given assumptions. Finally, two practical application examples are given to verify the effectiveness of our new algorithm.