Deep Reinforcement Learning for Delay and Energy-Aware Task Scheduling in Edge Clouds

Edge computing is proving to be a promising model, offering low-latency and high-bandwidth services to the end-users. However, due to the dynamic nature of the network and the heterogeneous computing resources, task scheduling in edge clouds remains a challenging problem. In order to solve this problem, we propose a novel task scheduling algorithm for edge clouds based on deep reinforcement learning, which combines a deep Q-learning network with a priority-based action selection strategy. This approach aims to optimize computing resource allocation while minimizing energy consumption in edge nodes. We evaluated the effectiveness of our algorithm using a simulated edge cloud environment and compared it with other advanced task scheduling algorithms. Experimental results indicate that our algorithm outperforms baseline algorithms in terms of delay and energy consumption. In particular, our method improves task completion time and energy efficiency compared to traditional scheduling algorithms.