Impact of Packet Loss Rate Estimation on Blockwise Adaptive Recoding for Batched Network Coding

Abstract

Batched network coding is a solution to reliable communications in multi-hop networks with packet loss. Adaptive recoding is a technique to enhance the throughput of batched network coding. Two pieces of information is needed to apply adaptive recoding: the distribution of the information remained in the received batches and the channel condition of the outgoing link. A simple way to obtain the former information is to make a short observation by grouping a few batches into a block. This way to apply adaptive recoding is called blockwise adaptive recoding (BAR), which can achieve a nice throughput even when the block size is small. Previous literature assumes that the latter information is known in advance and would not be changed over time. However in practice, these assumptions do not hold in general. In this paper, we investigate the impact of inaccurate channel condition on BAR and show by numerical evaluations that the throughput is very close to the one with accurate channel condition. To adapt the varying channel condition, we also propose a feedback scheme for BAR and a method to reduce the computational time for BAR.