In large-scale short-video platforms, CDN resource selection plays a critical role in maintaining users' 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.

Mobile video streaming occupies three-quarters of today's cellular network traffic. The quality of mobile videos becomes increasingly important for video providers to attract more users. For example, they invest in network bandwidth resources and conduct adaptive bitrate techniques to improve video quality. Prior adaptive bitrate (ABR) algorithms perform well under given throughput traces on broadband and WiFi networks. They may perform poorly for mobile video streaming due to the high network dynamics of cellular networks. To study the properties of throughput traces under cellular networks, we collect 4G network throughput traces for over four months in two large cities, Beijing and Suzhou in China. We derive the environment-specific Markov property of throughputs in the dataset. Accordingly, we propose NEIVA, an environment identification based technique to adaptively predict future throughput for different types of environments. We also implement NEIVA and integrate it with the state-of-the-art ABR algorithm, model predictive control (MPC) approach in our testbed for experiments. By emulating mobile video streaming under throughput traces in our dataset, NEIVA achieves 20 - 25 percent improvement on throughput prediction accuracy comparing to baseline predictors. Meanwhile, NEIVA achieves 11 - 20 percent user QoE improvement over MPC with baseline predictors.

Today’s video service providers all desire to deeply understand the ever changing factors on user QoE to attract more users. In this paper, we study the user engagement with respect to video quality metrics and improve user engagement in adaptive video streaming systems. We conduct a comprehensive study of the real data from iQIYI, covering 700K users and 150K videos. We find bitrate switch becomes the new dominant factor on user engagement instead of rebuffering events. We also observe the impact of rate of rebuffering is more dominant than rebuffering time. We examine novel interdependencies between quality metrics in the system, e.g., the positive correlation between bitrate switch and average bitrate, which is due to the context system strategy, i.e., conservative bitrate enhancing strategy adopted by iQIYI. To improve user engagement, we propose a new engagement centric QoE function based on real data and design server side ABR algorithm which leverages our new QoE function. We evaluate our method for online test in iQIYI, with 490K real users viewing 666K streams. The results show our approach outperforms existing approaches by significantly improving the viewing time, i.e., 2.8 minutes longer viewing time per user.

Adaptive bitrate (ABR) algorithms are critical techniques for high quality-of-experience (QoE) Internet video delivery. Early ABR algorithms conducting the overall QoE function of fixed parameters are limited by the fact that the QoE of end-users are diverse such that the video bitrate is often chosen in a misleading way. State-of-the-art ABR algorithms like MPC and Pensieve utilize offline modeling techniques and result in performance degradation for online QoE diversity adaptation. To address this issue, we propose Elephanta, an online ABR algorithm for edge users, which incorporates user QoE perception interface and adaptation algorithm with flexible parameters. In order to avoid overhead from updating parameters online, we model video streaming as a renewal system and formulate the specific QoE function into flexible formats by setting constraints on corresponding QoE metrics. To validate parameter settings, we emulate Elephanta under 1500 throughput traces, including FCC broadband, $3G$ HSDPA data set from the Internet, as well as the $4G$ /LTE data set we collect. Evaluation results show that Elephanta achieves QoE improvement of 7% over MPC and 3% over Pensieve under QoE diversity in part because of its superior adaptability to QoE diversity. We implemented Elephanta in dash.js at the client side for subjective experiments. We observed the diverse QoE preferences across users and 19/21 users (strongly) agree that Elephanta is responsive to parameter changes while watching videos.

Adaptive bitrate (ABR) algorithms have been critical techniques for high quality-of-experience (QoE) Internet video delivery. Prior work designs ABR algorithms by conducting the overall QoE function of fixed parameters. However, the QoE of end users are diverse and video bitrate may be chosen in a misleading way when leaving out the diversity. State-of-the-art ABR algorithms like MPC, Pensieve utilize off-line modeling techniques and result in performance degradation for online QoE diversity adaption. To address this issue, we propose Elephanta, an online flexible ABR algorithm for edge users which incorporates (1) user QoE perception interface and (2) adaption algorithm with flexible parameters. To avoid overheads for updating parameters online, we model video streaming as a renewal system and formulate specific QoE function into flexible formats by setting constraints on corresponding QoE metrics. To validate parameter setting, we emulate Elephanta under 5 thousand throughput traces including FCC broadband, 3G HSDPA data set from the Internet and 4G/LTE data set collected by ourselves. Accordingly, we implement Elephanta in dash.js at client side for user test. Evaluation results show that Elephanta achieves QoE improvement by 21.1% over MPC, in part for its superior adaptability to QoE diversity.

With the popularization of advanced cellular networks, mobile video occupies nearly three quarters of cellular network traffic. While previous adaptive bitrate (ABR) algorithms perform well under broadband network, their performance degrades in cellular networks due to throughput fluctuation. Through real world 4G/LTE network measurement, we find that throughput in cellular networks exhibits high fluctuation. It follows Markov behaviors with different states and different transition probability among states. We further find that the transition probability is stable along time but varies significantly under different environments. This inspires us to design ABR algorithms by improving throughput prediction in cellular networks. We propose NEIVA, a network environment identification based video bitrate adaption method in cellular networks. NEIVA trains a network environment identifier based on throughput data and trains a hidden Markov model (HMM) based throughput predictor for different environments. In online video bitrate selection, NEIVA utilizes the environment identifier to select the model for corresponding environment. Then NEIVA predicts future network performance by combining offline model and online throughput data. We implement NEIVA with MPC and evaluate it in real environment. The evaluation results show that with manually identifying environment, NEIVA improves 20% -- 25% bandwidth prediction accuracy and 11% -- 20% QoE improvement over the baseline predictors. With online environment identification, online NEIVA achieves 3.8% and 11.1% average QoE improvement over MPC and HMM, respectively.