Q2oE: Balancing QoE Fairness and Preference for Video Streaming from Cooperative Edge Servers

Quality of Experience (QoE) is a pivotal factor that defines the overall quality within the realm of video streaming services. To improve users' QoE indicators, existing methods have proposed many QoE models and Adaptive Bitrate (ABR) algorithms. However, these advanced methods do not account for the tradeoff between users' personalized preferences and fairness, leading to resource wastage on the edge server and an inefficient bitrate allocation. Given resource constraints, non-cooperative edge servers frequently request the cloud to enhance user QoE, consequently resulting in high-latency transmissions. In this paper, we address this issue by introducing collaborative edge servers, which alleviate the burden of video chunk transcoding and caching on individual servers. Specifically, we design a novel QoE model named Q²0E that incorporates user-level QoE preference and QoE fairness in this scenario. Furthermore, we propose Q2D3, a Q²oE-aware reinforcement learning algorithm based on a video chunk transcoding strategy for mobility-aware vehicular networks, to improve Q²oE model performance.