Multi-Phase Recoding for Batched Network Coding

Batched network coding is a practical realization of random linear network coding that resolves the issue of high computational and storage costs at the intermediate nodes by restricting recoding to the packets belonging to the same batch. Although there are various recoding approaches in the literature, they consider a one-shot approach which decides the number of recoded packets to be generated and transmits them without considering the reception status of these packets at the next node. In this paper, we divide the recoding process into multiple phases. In each phase, recoding depends on the amount of innovative information left at the current node after the transmission of the previous phases, where this knowledge can be obtained via feedback mechanism. This way, we can reduce the uncertainty during recoding and utilize the network more efficiently.We formulate the optimization problems for multi-phase adaptive recoding and show by simulation that two-phase recoding can already enhance the throughput significantly