Decoding Delay Reduction in Broadcast Erasure Channels with Memory for Network Coding

This paper studies feedback based instantly decodable network coding with the aim of minimizing decoding delay per transmission over wireless broadcast erasure channels with memory. We model such channels with a Gilbert-Elliott two-state Markov model with good and bad states. We first present a weighted sum generalized instantly decodable network coding (G-IDNC) scheme, where the aim is to service a subset of receivers with expected good channel state. We then propose an improved variation of the weighted sum G-IDNC that appropriately targets a broader set of receivers (while giving initial priority to receivers with expected good channel state) to reduce decoding delay over a wider range of erasure channels with memory. Simulation results show that our proposed improved weighted sum G-IDNC algorithm always considerably outperforms an earlier approach in the literature for erasure channels with memory, namely the weighted sum strict instantly decodable network coding (S-IDNC)