Minimal Overhead ARQ Sharing Strategies for URLLC in Multi-Hop Networks

TL;DR

This paper introduces a cooperative ARQ sharing strategy for multi-hop URLLC networks that improves reliability without increasing latency by allowing nodes to borrow unused ARQs from their predecessors.

Contribution

It proposes a novel cooperative ARQ sharing scheme where nodes share ARQ information, optimizing reliability in multi-hop URLLC networks with low complexity algorithms.

Findings

The cooperative strategy reduces packet-drop probability compared to non-cooperative methods.

The proposed algorithms achieve near-optimal ARQ distribution efficiently.

The scheme maintains low latency while enhancing reliability.

Abstract

The problem of achieving ultra-reliable and low-latency communication (URLLC) in multi-terminal networks has gained traction in the recent past owing to new wireless applications in vehicular networks. In the context of multi-hop networks, which is a classic example for multi-party communication, recent studies have shown that automatic-repeat-request (ARQ) based decode-and-forward (DF) strategies are suitable for URLLC since the idea of distributing a given number of ARQs across the nodes provides fine control on the features of reliability and latency. Inspired by these developments, in this work, we propose a cooperative ARQ sharing strategy for URLLC in multi-hop networks. At the heart of the proposed scheme lies the idea that every node is given the knowledge of the number of ARQs allotted to its preceding node in addition to the ARQs allotted to itself. As a result, each node only needs to count the number of unsuccessful attempts of its preceding node, and then borrow the unused ARQs, thereby improving the reliability feature with no compromise in the latency constraint. Using packet-drop-probability (PDP) as the reliability metric for the proposed cooperative strategy, we formulate an optimization problem of minimizing the PDP subject to a sum constraint on the total number of ARQs allotted across all the nodes. Supported by theoretical analysis on the behavior of PDP, we present low-complexity algorithms to compute near-optimal ARQ distributions for our strategy, and show that our strategy outperforms the existing non-cooperative strategies.