A QoS-Aware Energy Optimization Framework for Mobile Devices Using DRL and Edge Computing

Abstract

The proliferation of computationally intensive mobile applications like Augmented Reality (AR), Speech Recognition, and Mobile Gaming has led to an alarming rise in mobile energy consumption, raising the need for frequent recharging of mobile devices. This frequent recharging leads to degradation of mobile device’s battery health resulting in reduced battery capacity, shorter usage cycles, and ultimately increased costs associated with battery replacements. The current mobile energy optimization frameworks are unable to effectively handle the dynamic and realtime nature of mobile environments with unpredictable variables. This research proposes a Deep Reinforcement Learning (DRL) model to optimize energy efficiency in mobile devices using Mobile Edge Computing (MEC) to handle varying workloads and dynamic environment while maintaining Quality of Service (QoS) where computationally intensive applications are optimally offloaded to edge cloud servers. The result shows that the proposed outperforms current works by reducing energy consumption and average latency by 27.5% and 35% respectively.