CONSENSUS CONTROL FOR LEADER-FOLLOWING MULTI-AGENT SYSTEMS WITH MEASUREMENT NOISES

This work is concerned with consensus control for a class of leader-following multi-agent systems （MASs）. The information that each agent received is corrupted by measurement noises. To reduce the impact of noises on consensus, time-varying consensus gains are adopted, based on which consensus protocols are designed. By using the tools of stochastic analysis and algebraic graph theory, a sufficient condition is obtained for the protocol to ensure strong mean square consensus under the fixed topologies. This condition is shown to be necessary and sufficient in the noise-free case. Furthermore, by using a common Lyapunov function, the result is extended to the switching topology case.