A Distributed Consensus Protocol Based on Neighbor Selection Strategies for Multi-Agent Systems Convergence

Multi-agent systems (MASs), which consist of numerous mobile agents, is a promising research area in artificial intelligence that has been profusely applied to Engineering. A consensus problem of discrete time MASs with switching topology is investigated in this paper. First, a new distributed consensus protocol based on different neighbor selection strategies is proposed. In order to reach system consensus, the protocol requires each agent to intelligently refer to two neighbors for calculating and updating state. Compared with traditional protocol, the new protocol with different strategies can significantly reduce the cost of comparison, data storage and computation during MASs evolution. Next, three concrete neighbor selection strategies and an optimized strategy are designed. Then, we analyze and prove the stability of the protocol by Lyapunov theorem and Gerschgorin Theorem. The range of parameters, that affect reaching a consensus and equilibrium state in each strategy, is given in the proof. Finally, the experimental results demonstrate the effectiveness of the new protocol and the convergence performance of MASs under different neighbor selection strategies and parameter settings.