🤖 AI Summary
This paper addresses QoE optimization for real-time video streaming over multi-operator networks. Methodologically, it jointly models client-side resource allocation and cross-operator bandwidth sharing, formulating the problem as a Markov decision process that captures the nonlinear impact of resource allocation on perceived video quality. A theoretically grounded online bandwidth sharing strategy is proposed—first to provide rigorous optimality guarantees under dynamic network conditions. Analytical results reveal that arrival-rate imbalance and channel heterogeneity are decisive factors governing performance gains. Extensive large-scale simulations demonstrate up to 90% QoE improvement over state-of-the-art baselines, with practical parameter tuning guidelines provided for deployment. The core contributions are: (i) the first cross-operator online bandwidth sharing framework with provable theoretical optimality guarantees, and (ii) a quantitative characterization of how key system parameters—particularly arrival-rate skewness and channel diversity—directly influence achievable QoE gains.
📝 Abstract
We consider the problem of optimizing Quality of Experience (QoE) of clients streaming real-time video, served by networks managed by different operators that can share bandwidth with each other. The abundance of real-time video traffic is evident in the popularity of applications like video conferencing and video streaming of live events, which have increased significantly since the recent pandemic. We model the problem as a joint optimization of resource allocation for the clients and bandwidth sharing across the operators, with special attention to how the resource allocation impacts clients' perceived video quality. We propose an online policy as a solution, which involves dynamically sharing a portion of one operator's bandwidth with another operator. We provide strong theoretical optimality guarantees for the policy. We also use extensive simulations to demonstrate the policy's substantial performance improvements (of up to ninety percent), and identify insights into key system parameters (e.g., imbalance in arrival rates or channel conditions of the operators) that dictate the improvements.