Intelligent Online Computation Offloading for Wireless-Powered Mobile-Edge Computing
Yanting Wang, Zhuo Qian, Lijun He, Rui Yin, Celimuge Wu
Abstract
In the Internet of Things (IoT) ecosystem, optimizing processing capabilities of devices through Wireless Powered Mobile Edge Computing (WP-MEC) is crucial. This research addresses the challenge of efficiently scheduling task offloading from devices to an edge server, which is vital for enhancing system performance. Prior studies often overlook the necessity for rapid adaptation to changing wireless conditions, resulting in suboptimal offloading strategies. Our work introduces the Intelligent Online Computation Offloading (IOCO) algorithm, leveraging Deep Neural Networks (DNNs) to make informed, real-time offloading decisions based on previous experiences. This approach not only optimizes the allocation of wireless and computing resources but also incorporates novel quantization and sampling methods to improve robustness and adaptability. Simulation results demonstrate that IOCO can achieve near-optimal efficiency swiftly and adapt effectively to significant resource changes, highlighting its practicality in dynamic WP-MEC environments.