PIRA: Pan-CDN Intra-video Resource Adaptation for Short Video Streaming

TL;DR

This work tackles the dynamic QoE-cost tradeoff in short-video delivery by introducing PIRA, a client-side Model Predictive Control framework that jointly optimizes intra-video pan-CDN choices and video-range durations. It builds a media-list level, stochastic optimization model of QoE and traffic cost, and implements two pruning strategies to enable real-time operation. Through both simulations and large-scale production deployment (450,000 users over two weeks), PIRA achieves measurable improvements: startup delay reductions of 2.1%, rebuffering duration reductions of 15.2%, and a 10% reduction in unit traffic cost, along with a 13.4% drop in rebuffering occurrences. These results demonstrate that intra-video, dynamic CDN adaptation can significantly enhance user QoE while lowering operational expenses at scale, addressing a critical need for heterogeneously sourced edge resources in modern short-video platforms.

Abstract

In large scale short video platforms, CDN resource selection plays a critical role in maintaining Quality of Experience (QoE) while controlling escalating traffic costs. To better understand this phenomenon, we conduct in the wild network measurements during video playback in a production short video system. The results reveal that CDNs delivering higher average QoE often come at greater financial cost, yet their connection quality fluctuates even within a single video underscoring a fundamental and dynamic trade off between QoE and cost. However, the problem of sustaining high QoE under cost constraints remains insufficiently investigated in the context of CDN selection for short video streaming. To address this, we propose PIRA, a dynamic resource selection algorithm that optimizes QoE and cost in real time during video playback. PIRA formally integrating QoE and cost by a mathematical model, and introduce a intra video control theoretic CDN resource selection approach which can balance QoE and cost under network dynamics. To reduce the computation overheads, PIRA employs state space pruning and adaptive parameter adjustment to efficiently solve the high dimensional optimization problem. In large scale production experiments involving 450,000 users over two weeks, PIRA outperforms the production baseline, achieving a 2.1% reduction in start up delay, 15.2% shorter rebuffering time, and 10% lower average unit traffic cost, demonstrating its effectiveness in balancing user experience and financial cost at scale.

Figures (7)

• Figure 1: Figure (a) illustrates the average throughput of pan-CDNs across all users on the video platform, while Figure (b) shows the proportion of users who experience the best performance from each pan-CDN. For instance, pan-CDN1 delivers the highest average bandwidth for $65\%$ of users.
• Figure 2: An example of network throughput traces for different pan-CDNs in a short time period.
• Figure 3: Overview of PIRA and associated client-side streaming modules.
• Figure 4: An illustrative example of the media-list download process in short-video streaming.
• Figure 5: Normalized performance and computation overheads of different planning horizon steps in PIRA. Performance $=$ QoE-$\gamma$Cost, where $\gamma = 0.3$.