Distributed Consensus Through Network Support

Distributed consensus systems (DCSs) are increasingly important due to the often distributed realization of computational problems like voting systems and cryptocurrencies. The distributed nature of a DCS impacts the latency required to achieve consensus. In this paper, we develop an analytical model with empirically-based parameterization that provides an upper bound for that latency, thus enabling DCS operators to set expectations for its performance. We also propose a departure from the usual peer-to-peer-based DCS realization through an application-layer-based multicast approach. Our comparative analysis shows that our solution can improve convergence times by a factor of four while also ensuring targeted boundaries through the finality of the consensus convergence.