Distributed Task Scheduling for Wireless Powered Mobile Edge Computing: A Federated-Learning-Enabled Framework

Abstract

Sixth generation (6G) wireless communication networks are expected to provide ultra-low latency and energy-efficient services for global users. The paradigm of wireless powered mobile edge computing (WP-MEC) provides a feasible framework to real-ize the above purposes by leveraging the advantage of both wireless power transfer and MEC. Howev-er, massive network traffic and control information overburden the centralized server. In this article, we integrate federated learning and deep reinforcement learning with WP-MEC to jointly optimize computing and communication resources. Specif-ically, we design an online learning framework to schedule computation tasks of mobile clients, where the learning model can be distributed over edge servers, and parameter synchronization happens on the cloud. The superiority of the designed frame-work is evaluated and compared with benchmarks in terms of average task completion delay and task completion ratio. Finally, we discuss some research challenges and opportunities for future WP-MEC.