[Paper Review] Anticipatory Buffer Control and Quality Selection for Wireless Video Streaming

Martin Dräxler, Johannes Blobel+3 more|arXiv (Cornell University)|Sep 21, 2013

Advanced Wireless Network Optimization35 references12 citations

TL;DR

This paper proposes an anticipatory buffer control and quality selection scheme for wireless video streaming that uses predicted future data rates to optimize segment downloading and video quality. By proactively adjusting buffer levels and quality based on forecasts, the approach eliminates playback interruptions while minimizing over-buffering, achieving near-zero stalling and high video quality in simulations and testbeds.

ABSTRACT

Video streaming is in high demand by mobile users, as recent studies indicate. In cellular networks, however, the unreliable wireless channel leads to two major problems. Poor channel states degrade video quality and interrupt the playback when a user cannot sufficiently fill its local playout buffer: buffer underruns occur. In contrast to that, good channel conditions cause common greedy buffering schemes to pile up very long buffers. Such over-buffering wastes expensive wireless channel capacity. To keep buffering in balance, we employ a novel approach. Assuming that we can predict data rates, we plan the quality and download time of the video segments ahead. This anticipatory scheduling avoids buffer underruns by downloading a large number of segments before a channel outage occurs, without wasting wireless capacity by excessive buffering. We formalize this approach as an optimization problem and derive practical heuristics for segmented video streaming protocols (e.g., HLS or MPEG DASH). Simulation results and testbed measurements show that our solution essentially eliminates playback interruptions without significantly decreasing video quality.

Motivation & Objective

• To address playback interruptions (buffer underruns) and inefficient resource use (over-buffering) in wireless video streaming due to fluctuating channel conditions.
• To improve Quality of Experience (QoE) by intelligently scheduling video segment downloads and quality levels based on anticipated future data rates.
• To design a scheduling framework that operates independently of physical-layer resource allocation, focusing on application-layer buffer and quality control.
• To validate the approach through both simulation and real-world testbed measurements under realistic wireless conditions.
• To demonstrate that anticipatory scheduling can maintain high video quality while eliminating playback stalls without excessive buffering.

Proposed method

• Formalizes the problem as an optimization task to determine the optimal download time and quality level for each video segment based on predicted data rates.
• Introduces a novel anticipatory scheduling mechanism that plans ahead using predicted data rate estimates over time intervals matching video segment durations.
• Employs heuristics such as the MIQCP and \proc Fill scheduler to solve the scheduling problem efficiently in practice.
• Uses a discrete time model with time slots to simulate segment download schedules, converting real-time testbed data into comparable discrete intervals.
• Relies on channel anticipation—estimates of future data rates over tens of seconds—rather than real-time channel measurements.
• Designs the system to be independent of physical-layer scheduling, assuming fair radio resource allocation is already in place.

Experimental results

Research questions

• RQ1 How can video streaming systems proactively manage buffer levels and video quality to prevent playback interruptions in wireless networks?
• RQ2 What is the impact of over-buffering on wireless resource utilization, and how can it be minimized without increasing stalling risk?
• RQ3 To what extent can predicted future data rates improve scheduling decisions in HLS and MPEG-DASH video streaming?
• RQ4 How do timing side effects in real testbeds (e.g., TCP, CPU contention) affect the performance of anticipatory scheduling compared to simulation?
• RQ5 Can anticipatory scheduling maintain high video quality while achieving near-zero playback interruptions in practical deployments?

Key findings

• The proposed anticipatory scheduling scheme eliminates playback interruptions in both simulation and testbed environments, achieving near-zero stalling.
• The MIQCP and \proc Fill schedulers significantly reduce buffer underruns compared to greedy approaches, even under variable channel conditions.
• Testbed measurements show only minor deviations from simulation results in average video quality and buffer levels, validating the model's accuracy.
• The higher lateness observed in testbeds for MIQCP and \proc Fill schedulers is attributed to buffer minimization making them more sensitive to real-world timing effects like TCP congestion and CPU contention.
• The agreement between simulation and testbed results confirms the feasibility and robustness of the anticipatory scheduling approach in real-world wireless conditions.
• The system maintains high video quality while minimizing over-buffering, demonstrating efficient use of wireless capacity and improved QoE.