Algorithm-Level Confidentiality for Average Consensus on Time-Varying Directed Graphs

Average consensus plays a key role in distributed networks, with applications ranging from time synchronization, information fusion, load balancing, to decentralized control. Existing average algorithms require individual agents exchange explicit state values their neighbors, which leads the undesirable disclosure of sensitive state. In this paper, we propose novel algorithm for time-varying directed graphs that can protect confidentiality participating agent against other agents. The injects randomness interaction obfuscate on algorithm-level and ensure information-theoretic privacy without assistance any trusted third party or data aggregator. By leveraging inherent robustness dynamics random variations interaction, our proposed also guarantee accuracy consensus. is distinctly different differential-privacy based approaches enable through compromising obtained value. Numerical simulations confirm effectiveness efficiency approach.