Q-Learning and ML-based Energy-Efficient and Delay-Aware Routing for Multimedia Wireless Sensor Networks

Abstract

Multimedia wireless sensor networks (MWSNs) require routing strategies that simultaneously minimize energy consumption, reduce end‑to‑end delay, and preserve quality of service (QoS) for real‑time video and audio transmission. Conventional protocols such as shortest path routing and LEACH provide partial solutions: shortest path reduces distance but neglects QoS, while LEACH conserves energy through clustering but introduces delay variability. In this study, an adaptive machine learning–based routing framework using Q‑learning is developed to dynamically select optimal paths under varying network conditions. The reward function integrates energy expenditure, latency, and multimedia QoS metrics, enabling the system to learn routes that balance efficiency and reliability. MATLAB simulations compare the proposed ML routing with shortest path and LEACH protocols across multiple episodes. Results show that ML routing achieves lower energy consumption per transmission and reduced delay compared to shortest path, while maintaining competitive PSNR values. Although LEACH demonstrates superior energy conservation and packet delivery ratio, ML routing provides a balanced solution by explicitly incorporating QoS into routing decisions. This comparative analysis highlights the potential of machine learning to enhance multimedia WSN performance and motivates hybrid ML‑LEACH designs that combine energy efficiency with adaptive QoS optimization.