Delay Minimization for Instantly Decodable Network Coding in Persistent Channels with Feedback Intermittence

TL;DR

This paper addresses minimizing multicast decoding delay in persistent channels with intermittent feedback by modeling the environment with Gilbert-Elliott channels, formulating the problem as a maximum weight clique, and proposing a greedy algorithm that outperforms existing methods.

Contribution

The paper introduces a generalized formulation for decoding delay minimization under persistent erasure channels and feedback intermittence, along with a novel greedy algorithm for NP-hard problem solving.

Findings

The proposed algorithm significantly reduces decoding delay in simulations.

It outperforms existing blind approaches across various channel conditions.

The formulation encompasses simpler models as special cases.

Abstract

In this paper, we consider the problem of minimizing the multicast decoding delay of generalized instantly decodable network coding (G-IDNC) over persistent forward and feedback erasure channels with feedback intermittence. In such an environment, the sender does not always receive acknowledgement from the receivers after each transmission. Moreover, both the forward and feedback channels are subject to persistent erasures, which can be modelled by a two state (good and bad states) Markov chain known as Gilbert-Elliott channel (GEC). Due to such feedback imperfections, the sender is unable to determine subsequent instantly decodable packets combination for all receivers. Given this harsh channel and feedback model, we first derive expressions for the probability distributions of decoding delay increments and then employ these expressions in formulating the minimum decoding problem in such environment as a maximum weight clique problem in the G-IDNC graph. We also show that the problem formulations in simpler channel and feedback models are special cases of our generalized formulation. Since this problem is NP-hard, we design a greedy algorithm to solve it and compare it to blind approaches proposed in literature. Through extensive simulations, our adaptive algorithm is shown to outperform the blind approaches in all situations and to achieve significant improvement in the decoding delay, especially when the channel is highly persistent