Multi-packet Hybrid ARQ: Closing gap to the ergodic capacity

TL;DR

This paper introduces multi-packet HARQ, allowing multiple packets to share channel blocks, optimizing resource allocation, and significantly improving throughput over conventional HARQ in fading channels.

Contribution

It proposes a novel multi-packet HARQ scheme with optimized resource sharing and feedback strategies, outperforming traditional HARQ in fading environments.

Findings

Multi-packet HARQ outperforms conventional HARQ by over 5 dB at high spectral efficiencies.

Optimized resource sharing improves throughput in block-fading channels.

Rich multi-bit feedback enhances decision-making and performance.

Abstract

In this work we consider incremental redundancy (IR) hybrid automatic repeat request (HARQ), where transmission rounds are carried out over independent block-fading channels. We propose the so-called multi-packet HARQ where the transmitter allows different packets to share the same channel block. In this way the resources (block) are optimally assigned throughout the transmission rounds. This stands in contrast with the conventional HARQ, where each transmission round occupies the entire block. We analyze superposition coding and time-sharing transmission strategies and we optimize the parameters to maximize the throughput. Besides the conventional one-bit feedback (ACK/NACK) we also consider the rich, multi-bit feedback. To solve the optimization problem we formulate it as a Markov decision process (MDP) problem where the decisions are taken using accumulated mutual information (AMI) obtained from the receiver via delayed feedback. When only one-bit feedback is used to inform the transmitter about the decoding success/failure (ACK/NACK), the Partial State Information Markov Decision Process (PSI-MDP) framework is used to obtain the optimal policies. Numerical examples obtained in a Rayleigh-fading channel indicate that, the proposed multi-packet HARQ outperforms the conventional one, by more than 5 dB for high spectral efficiencies.