Consensus algorithm of multi-agent system with binary-valued communication

This paper studies the consensus of multi-agent systems with binary-valued communication. We consider a group of agents on an undirected graph with a fixed topology, but differing from most existing work, each agent cannot get the true value of its neighbors´ states. What information each agent gets from its neighbors is binary-valued measurement. A two-scale control algorithm is constructed: Each agent in the information network estimates the states of its neighbors´ based on the binary-valued measurements for a short time, during which every node´s state will keep constant; Based on the estimation of its neighbors´ states and its own state, each agent designs its own control, by which the states will be updated and finally the system will achieve consensus. The state estimation algorithm is analyzed to have convergent and asymptotically efficient properties. As a result, the system is shown to be weak consensus by using a stochastic Lyapunov analysis.