- Journal
- COMPUTER COMMUNICATIONS
- Année
- 2026
- Volume
- 250
- Article
- 108453
- Mois
- MAR 15
- DOI
- 10.1016/j.comcom.2026.108453
Abstract
This paper presents a reinforcement learning (RL)-based hybrid error-correction framework to enhance wireless communication performance in interference-limited, high-density Internet of Things (IoT) networks for telemedicine applications. The proposed system integrates lightweight tabular Q-Learning (QL) with classical Automatic Repeat Request (ARQ) and Forward Error Correction (FEC) techniques, enabling dynamic selection of error-control strategies based on real-time channel conditions. By modeling the communication process as a Markov Decision Process (MDP), the framework jointly optimizes key network metrics, including Round-Trip Time (RTT), Polling Rounds (PR), Network Capacity (NC), and Message Delivery Timeliness (MDT), across varying Total Message Lengths (TML). Simulation results show that the QL-driven approach improves NC by approximately 93%-99% (with the highest gains observed for larger payloads under high Signal-to-Interference-plus-Noise Ratio conditions) and reduces median RTT by about 46% compared to static ARQ, FEC, and hybrid ARQ+FEC baselines. Extensive MATLAB simulations confirm substantial gains in capacity and latency reduction across diverse interference-limited telemedicine scenarios. These results underscore the framework's potential to support reliable, low-latency, and scalable biometric data transmission in real-time telemedicine applications, including wireless facial-scanning devices.