Convolutional Network Coding based on Systematic RS Code for Low-latency Guarantee

Abstract

This paper studies the network coding problem of end-to-end packet loss networks under the requirements of low latency and high reliability. The optimization relationship for reliable data transmission using network coding is established in terms of transmission rate, constraint delay and success rate. The analysis results show that the reliable transmission rate asymptotically approaches the capacity of end-to-end packet loss networks with the increase of the limited delay. However, when the limited delay is small, only short codes can be used although its rate still has a big gap from the capacity 1 - ∊, where ∊ is the packet loss probability of networks. MDS code can perform well and nearly achieves the upper bound for packet loss networks. Furthermore, this paper presents a systematic RS convolutional network coding scheme to achieve long code performance with small decoding delay. The simulation results show that under the same time delay limitation and reliability guarantee, the convolutional network coding can improve the transmission rate meanwhile with low maximum latency compared to block codes.