Fast Average-Consensus on Networks Using Heterogeneous Diffusion

In this brief, a heterogeneous diffusion-based communication protocol is proposed to achieve a significant improvement in the rate of convergence of the average consensus among the multi-agents. We prove that using the proposed communication protocol, star network, complete network, complete bipartite, $n-$ cube hypergraph, $d-$ regular bipartite graph, and $d-$ regular Ramanujan graph has a faster consensus as compared to the considered state of the art consensus protocol on these networks. Numerical simulations also show the validity of the results on real networks and the model networks.