Delay Performance of Network-Coding-Based Epidemic Routing

Abstract

In this paper, we first challenge the accuracy of the so-called innovative assumption that is widely adopted in delay performance analysis of network-coding-based epidemic routing (NCER) in delay-tolerant networks (DTNs). We demonstrate that this optimistic assumption severely underestimates data delivery delay and solves this problem successfully by introducing an extra encounter factor $\delta$ . Based on this, we propose a coloring process (CP)-based analytical model to evaluate the delay performance of NCER. Numerical results show that our CP-based method outperforms traditional ordinary differential equation (ODE)-based methods on estimating the delay performance under different network sizes. Furthermore, to mitigate potential competition among multiple data streams from different nodes, we propose a feedback-based recovery protocol (FRP) that takes advantage of finite buffer space and significantly reduces the number of ineffective transmissions between nodes. As a result, FRP can achieve much better delay performance compared with existing protocols in prior work.