Synchronization by modified broadcast gossip algorithm in multi-agent system

Abstract

This paper proposes a modified broadcast gossip algorithm that reaches synchronization of multi-agent systems. The broadcast gossip algorithm is used to reach a consensus on the scalar value of each agent in a static environment. This algorithm is suitable for the wireless multi agent system because it can reach a consensus by uniformly random broadcasting communication. In this paper, we modified the broadcast gossip algorithm to synchronize each system in a discrete time dynamic multi-agent system. The proposed algorithm reduces usage of network resources, and it has a robust property against topology change. In the simple case, we prove that this algorithm almost surely reaches a consensus on a dynamic system, and the expectation of consensus is the dynamics of average initial state. Simulation results show that the algorithm works in an oscillated system and that it can be applied to decentralized and topology-changed environment.