Deep Reinforcement Learning Based Delay-Sensitive Task Scheduling and Resource Management Algorithm for Multi-User Mobile-Edge Computing Systems

With the arrival of the 5G era, a new service paradigm known as mobile-edge computing (MEC) has been introduced for providing high quality mobile services by offloading the delay-sensitive and computation-intensive tasks from mobile devices to nearby MEC servers. In this paper, we investigate the problem of delay-sensitive task scheduling and resource (e.g. CPU, memory) management on the server side in multi-user MEC scenario, and propose a new online algorithm based on deep reinforcement learning (DRL) method to reduce average slowdown and average timeout period of tasks in the queue. We also design a new reward function to guide the algorithm to learn directly from experience to scheduling tasks and managing resources. Simulation result shows that our algorithm outperforms multiple traditional algorithms and have a big advantage of intelligence and good understanding towards workload and environment.