Zhika Liu: Multi-agent Federated Reinforcement Learning Strategy for Mobile Virtual Reality Delivery Networks


Elm Lecture Theatre, Nucleus Bldg


Thursday, May 23, 2024 - 13:00 to 14:00


Virtual reality (VR) services are gaining popularity but pose challenges for wireless communications due to their high data demands. This talk will introduce a joint caching, computing, and communication (3C) strategy tailored for dynamic VR scenarios. The strategy aims to minimize long-term discounted delay and energy consumption for VR projection, adhering to constraints on latency, power, caching, and computing. This system features a three-layer communication setup that includes a cloud server, UAV (Unmanned Aerial Vehicle) base stations equipped with massive Multiple-Input Multiple-Output (mMIMO) technology serving as edge servers, and mobile user devices. Eight service routes for 3C decisions have been designed to cater to diverse user needs. A federated multi-agent deep reinforcement learning (RL) algorithm is proposed using actor and critic networks. This algorithm optimizes service routes based on users' locations, orientations, and content preferences. Here, edge servers function as learning agents, facilitating the adaptation of service delivery to user-specific conditions.


Zhikai Liu received a B.S. degree in communications engineering from Hohai University in 2014 and an M.S. degree in Electromagnetic Field and Microwave Technology from South China Normal University in 2021. He is currently pursuing a Ph.D. degree at The University of Edinburgh. His research interests include edge caching, distributed learning, signal processing, and wireless communications.

Zhika Liu

Event Contact Name: 

Tao Xu

Event Contact Email: