H.264/SVC parameter optimization based on quantization parameter, MGS fragmentation, and user bandwidth distribution

Xu Chen, Jihong Zhang, Wei Liu, Yongsheng Liang, Ji-qiang Feng

Abstract

This paper presents an optimization framework for H.264/SVC (Scalable Video Coding) that aligns encoding parameters with the statistical distribution of user bandwidth. By jointly optimizing the enhancement layer's quantization parameter ($QP_e$) and Medium Grain Scalability (MGS) fragmentation, the proposed approach maximizes average video quality (PSNR) across diverse network environments. Experimental results demonstrate significant quality gains and high rate-estimation accuracy, making it a viable solution for context-aware video streaming.

Introduction

In modern video streaming, a "one size fits all" approach to encoding is often inefficient due to the wide variance in user internet speeds. Scalable Video Coding (SVC) addresses this by providing a base layer for low-quality video and enhancement layers for higher detail. However, traditional methods for setting enhancement layer parameters—specifically the Quantization Parameter ($QP_e$) and MGS fragmentation—often overlook the actual transmission environment. This research proposes a method to optimize these parameters by matching them to the statistical distribution of the target audience's bandwidth, ensuring the bitstream is most efficient for the actual population of viewers.

Methods

The authors propose a three-stage optimization framework designed to find the mathematical "sweet spot" for video delivery:

  1. $QP_e$ Determination: A mapping is established between the user bandwidth range and the quantization parameter. Using Rate-Quantization (R-Q) models, the system calculates a $QP_e$ that ensures the total bit rate aligns with the upper bounds of the users' network capacities.
  2. MGS Fragmentation Optimization: Medium Grain Scalability allows enhancement layers to be split into up to 15 fragments. While more fragments offer finer granularity for bandwidth adaptation, they also introduce header overhead. The authors utilize a pre-coding process to analyze quality increments and determine the optimal number of fragments.
  3. Joint Optimization via Probability: The core of the approach is an objective function that maximizes average PSNR across all users. By modeling user bandwidth as a probability density function $f(x)$, the system iterates through configurations to find the highest expected quality.
Figure 1: Sketch map of the optimization approach, illustrating the flow from video input to encoding scheme determination.
Figure 2: The pre-coding process, showing the MGS stack structure across I, B, and P frames.

Results and Discussion

The study utilized the Joint Scalable Video Model (JSVM) to test 28 standard video sequences, including "Foreman," "Bus," "Mobile," and "Soccer." Key findings include:

  • Quality Improvements: The optimized parameter selection outperformed fixed-parameter schemes by up to 3 dB in average PSNR for specific bandwidth ranges in the "Mobile" and "Bus" sequences.
  • Estimation Accuracy: The rate estimation model proved highly reliable, showing less than a 5% deviation from actual coding results.
  • Efficiency and Trade-offs: The research identified an approximately linear relationship between coding bit rate and fragment quality in MGS. It also quantified the overhead penalty; while scalability improves adaptation, excessive fragmentation eventually degrades overall PSNR due to NAL unit header overhead.

Conclusion

This research demonstrates that the "best" encoding settings are not inherent to the video content alone but are a function of the audience's network statistics. By integrating user bandwidth distribution into the optimization loop, significant gains in delivery efficiency can be achieved. Future work will look to incorporate spatial and temporal scalability into the model and explore real-time adaptation for dynamic audience distributions.

Summary updated

Abstract

In the situation of limited bandwidth, how to improve the performance of scalable video coding plays an important role in video coding. The previously proposed scalable video coding optimization schemes concentrate on reducing coding computation or trying to achieve consistent video quality; however, the connections between coding scheme, transmission environments, and users’ accesses manner were not jointly considered. This article proposes a H.264/SVC (scalable video codec) parameter optimization scheme, which attempt to make full use of limited bandwidth, to achieve better peak signal-to-noise ratio, based on the joint measure of user bandwidth range and probability density distribution. This algorithm constructs a relationship map which consists of the bandwidth range of multiple users and the quantified quality increments measure, QP e , in order to make effective use of the video coding bit-stream. A medium grain scalability fragmentation optimization algorithm is also presented with respect to user bandwidth probability density distribution, encoding bit rate, and scalability. Experiments on a public dataset show that this method provides significant average quality improvement for streaming video applications.
BibTeX
@article{chen2013h,
  author = {Xu Chen and Jihong Zhang and Wei Liu and Yongsheng Liang and Ji-qiang Feng},
  title = {{H.264/SVC parameter optimization based on quantization parameter, MGS fragmentation, and user bandwidth distribution}},
  journal = {EURASIP J. Adv. Signal Process.},
  year = {2013},
  doi = {10.1186/1687-6180-2013-10},
}
Figure