Adaptive Cooperative Transmission Design for Ultra-Reliable Low-Latency Communications via Deep Reinforcement Learning

TL;DR

This work tackles URLLC in a two-hop relaying scenario by framing per-hop transmission parameter selection as a dual-agent Markov decision process. A novel DRL-CoLA framework enables source and relay to learn latency-aware policies for numerology, mini-slot size, and MCS using only local CSI and ARQ feedback, avoiding global CSI overhead. The approach employs decentralized deep Q-learning with delay-outage-rate-based rewards and demonstrates near-optimal reliability under strict latency constraints, validated by simulations against a global-CSI one-shot baseline. The findings highlight the practicality of distributed learning for reliable, low-latency communications in relay-assisted 5G NR and beyond.

Abstract

Next-generation wireless communication systems must support ultra-reliable low-latency communication (URLLC) service for mission-critical applications. Meeting stringent URLLC requirements is challenging, especially for two-hop cooperative communication. In this paper, we develop an adaptive transmission design for a two-hop relaying communication system. Each hop transmission adaptively configures its transmission parameters separately, including numerology, mini-slot size, and modulation and coding scheme, for reliable packet transmission within a strict latency constraint. We formulate the hop-specific transceiver configuration as a Markov decision process (MDP) and propose a dual-agent reinforcement learning-based cooperative latency-aware transmission (DRL-CoLA) algorithm to learn latency-aware transmission policies in a distributed manner. Simulation results verify that the proposed algorithm achieves the near-optimal reliability while satisfying strict latency requirements.

Figures (3)

• Figure 1: System model for two-hop relaying with ARQ protocols under the latency constraint $T_{\rm th}$.
• Figure 2: Comparison of end-to-end reliability between the proposed DRL-CoLa scheme and one-shot transmission.
• Figure 3: Comparison of accumulated rewards achieved by DRL-CoLA when implemented with different reinforcement learning algorithms.