A Unified Adaptive Recoding Framework for Batched Network Coding

Batched network coding is a variation of random linear which has low computational and storage costs. In order to adapt fluctuations in the number erasures individual batches, it not optimal recode transmit same packets for all batches. Different distributed optimization models, are called adaptive recoding schemes, were formulated this purpose. The key component these problems expected value rank distribution batch at next node, also known as rank. paper, we put forth unified framework with an arbitrary field size. We show that functions concave when packet loss pattern stationary stochastic process, covers but limited independent Gilbert-Elliott model. Under concavity assumption, there always exists solution only can minimize randomness on recoded tolerate errors due inaccurate measurements or precision machine. provide algorithm obtain such solution, propose tuning schemes turn any feasible into desired solution.