A Framework of an RL-Based Task Offloading Mechanism for Multi-users in Edge Computing

Abstract

AbstractIn recent years, the Internet of Things (IoT) has been growing rapidly, and new applications using the IoT includes autonomous driving systems, mobile health, smart homes, VR/AR technologies, and many more. While IoT applications plays important roles in enriching our lives, the size of each generated task is increasing compared to tasks generated by traditional mobile devices, which can cause high latency when deployed in traditional cloud computing. Edge computing has emerged as a method to mitigate this problem. Edge computing can provide services with lower latency than conventional cloud computing methods because the tasks are processed in the vicinity of the user’s device. However, when the tasks are concentrated in one of the distributed servers, the low processing power of these servers becomes a bottleneck and high latency may occur.In this paper, we proposed a deep reinforcement learning based offloading mechanism in an edge computing environment which can dynamically offload a task based on the performance and availability of nearby edge servers. Preliminary experiment results are promising and offered insights on related issues.KeywordsEdge computingDeep reinforcement learningTask offloading