IABR: An Intelligent Joint Adaptive Bitrate Selection and Communication Resource Allocation in F-RAN

Abstract

Benefiting from the caching and transcoding capacity, the fog radio access network (F-RAN) has exhibited the promising potential to promote the performance of the adaptive bitrate (ABR) at the fog server. Despite extensive research on ABR algorithms, the joint optimization of bandwidth resource allocation and bitrate selection is a significant and interesting topic to maximize the QoE for all users in F-RAN. However, the joint optimization problem is intractable to be solved by traditional methods according to complex network dynamics and large user preference variances. In this paper, we propose an intelligent adaptive bitrate algorithm (iABR) based on a hierarchical actor-critic (HAC) agent at the F-RAN server to maximize the overall QoE of multiple users. The proposed iABR can perceive user preference and dynamics of wireless communication environment to automatically adjust the bandwidth resource allocation and bitrate selection policy. Simulation results illustrate that the proposed iABR exhibits more satisfying performance on overall QoE for multiple users compared with state-of-art reinforcement learning based ABR algorithm when adopting average bandwidth allocation and greedy bandwidth allocation.