Event-Triggered Privacy-Preserving Dynamic Average Consensus via State Decomposition

We study the problem of dynamic average consensus for a network of agents under the dual constraints of limited communication and privacy preservation. To address this problem, we propose a distributed event-triggered privacy-preserving algorithm based on state decomposition. In the proposed algorithm, each agent communicates with its neighbors only when a local event-triggering condition involving a dynamic trigger variable is satisfied, thereby reducing the communication overhead. At the same time, privacy is preserved by decomposing the state of each agent into visible and hidden substates, enabling collaborative computation without disclosing sensitive information about the reference signals. Rigorous theoretical analysis shows that the proposed algorithm ensures convergence to an adjustable neighborhood of the average of the reference signals and provides privacy preservation against any external eavesdropper. It is also shown that the Zeno behavior is excluded by guaranteeing the existence of a positive minimum inter-event time, which is locally tunable through design parameters. The effectiveness and applicability of the proposed algorithm are demonstrated through simulation studies on a networked battery energy storage system.