An Improved Prescribed-Time Privacy-Preserving Average Consensus Algorithm for First-Order Multi-Agent Systems

Privacy issues, communication constraints and convergence performance are key factors for expanding applications of multi-agent systems, but considering the above three factors simultaneously has not been well addressed. In order to extend the practical application scenarios and fill such gaps, the paper proposes an improved prescribed-time privacy-preserving consensus method based on event-triggered mechanism. Unlike the traditional privacy protection scheme, proposed method adopts a decentralized (executed separately by every agent), time-varying transformation to encrypt the information. Agents encode their true states in virtue of independent key before transmitting them, which makes the agents' internal states indecipherable from others. On this basis, the prescribed-time theory is introduced to achieve convergence at any specified time within the range allowed by the actuator performance. In addition, event-triggered mechanism can effectively relieve the actuator pressure caused by frequent updates and reduce computational resource consumption. Finally, the validity of theoretical results is verified by numerical simulation experiments.