Video Streaming with Kairos: An MPC-Based ABR with Streaming-Aware Throughput Prediction

TL;DR

This work tackles the ill-posed throughput prediction problem in MPC-based ABR by introducing Kairos, a streaming-aware predictor that explicitly accounts for irregular sampling, uncertainty, and throughput smoothness. It combines a multi-time attention network (mTAN) to regularize irregular throughput histories with a quantile-based predictor and a buffer-aware uncertainty module, coupled with a smoothness regularizer. These components feed into an uncertainty-aware MPC that optimizes QoE over a horizon while respecting buffer dynamics and rebuffer penalties. Across trace-driven and real-world experiments, Kairos consistently outperforms state-of-the-art ABR schemes, delivering robust QoE gains and smoother bitrate transitions in diverse network environments, highlighting the practical impact of incorporating streaming-aware throughput predictions into MPC-based ABR.

Abstract

In this paper, we present Kairos, a model predictive control (MPC)-based adaptive bitrate (ABR) scheme that integrates streaming-aware throughput predictions to enhance video streaming quality. Kairos features an attention-based throughput predictor with buffer-aware uncertainty control, improving prediction accuracy and adaptability to network conditions. Specifically, we introduce a multi-time attention network to handle the irregularly sampled sequences in streaming data, creating uniformly spaced latent representations. Additionally, we design a separate prediction network that estimates future throughput at multiple percentiles and incorporates a buffer-aware uncertainty adjustment module. This module dynamically selects the appropriate throughput percentile based on the buffer size, enhancing robustness to varying network conditions. Lastly, to mitigate QoE smoothness penalties caused by predictors focused solely on accuracy, we introduce a smoothness regularizer. By embedding streaming-aware characteristics, such as sampling irregularity, buffer occupancy, and smoothness, into the throughput predictor design, Kairos significantly improves bitrate decision-making within the MPC framework. Extensive trace-driven and real-world experiments demonstrate that Kairos outperforms state-of-the-art ABR schemes, achieving an average QoE improvement of 1.52% to 7.28% under various network conditions.

Figures (10)

• Figure 1: Performance gap between MPC-based ABRs.
• Figure 2: The main bottleneck limiting MPC-based ABR performance is the performance of throughput predictor.
• Figure 3: Overview of Kairos augmented video streaming system.
• Figure 4: Architecture of the multi-time attention based throughput predictor.
• Figure 5: Comparing QoE performance of Kairos with existing ABR algorithms on all considered network traces.