一种基于缓冲区阈值调整的动态自适应码率选择算法

A Dynamic Adaptive Rate Selection Algorithm based on Buffer Threshold Adjustment

关键词:  流媒体  动态自适应  缓冲区阈值  自适应算法

Key words:  streaming media  dynamic adaptive  buffer threshold  adaptive algorithm

摘要:

基于HTTP的动态自适应流媒体（Dynamic adaptive streaming over HTTP，DASH）是一种可利用Web服务器提供在线高质量视频流的自适应码率流技术，决定了视频播放性能。传统动态自适应码率选择算法存在考虑因素单一、播放缓冲区长度受限、带宽不稳定等问题，因此，改进与优化动态自适应码率选择算法十分必要。针对客户端缓冲区长度受限和带宽不稳定造成的视频播放卡顿现象，综合考虑网络带宽和客户端缓冲区因素，提出一种基于缓冲区阈值调整的动态自适应码率选择算法（Dynamic adaptive rate selection algorithm based on buffer threshold adjustment，BT-DARA）。该算法首先通过获取不同视频段大小和下载速率，计算网络吞吐量并对视频下载时间进行预测，然后结合缓冲区视频片段长度状态进行码率选择。特别地，在视频播放过程中对不同阶段采取不同的视频码率选择策略，并在客户端缓冲区达到临界值时动态调整缓冲区阈值参数，以此增加缓冲区长度，减少码率切换次数，确保视频播放的稳定性。实验结果表明，该算法能提高视频播放的码率，保证视频播放质量和稳定性，减少了视频开始播放的启动时间，为用户提供良好的视频播放体验。

Abstract:

Dynamic Adaptive Streaming over HTTP(DASH) is an adaptive rate stream technology that can provide online high-quality video streams using web servers, which determines video playback performance. There are some problems in the traditional dynamic adaptive rate selection algorithm, such as considering only a single factor, limited play buffer length and bandwidth instability. It is necessary to improve and optimize the dynamic adaptive rate selection algorithm. Aiming at the video playback carton caused by the limited buffer length and bandwidth instability of the client, and comprehensively considering the network bandwidth and client buffer factors, a dynamic adaptive rate selection algorithm based on buffer threshold adjustment (BT-DARA) is proposed. The algorithm first calculates the network throughput to predict the video download time by acquiring different video segment sizes and downloading rates, and then combines the buffer video segment length state to select the bit rate. In particular, different video bit rate selection strategies are adopted for different stages during video playback, and the buffer threshold parameter is dynamically adjusted when the client buffer reaches a critical value. Therefore, the algorithm can increase the buffer length, reduce the number of code rate switching, and ensure the stability of video playback. The experimental results show that the algorithm can improve the bit rate of video playback, ensure the quality and stability of video playback, reduce the startup time of video playback, and provide users with a good video playback experience.