一种嵌入视频流的丢包自测量方法  

Embedding Method for Packet Loss Self-Measurement of Video Streaming

作　　者：胡治国[1,2] 张大陆[1] 谷丽丽[1] 张起强[1] 陈志伟[1] 周华磊[1] 曹孝晶[1] 

机构地区：[1]同济大学计算机科学与技术系 [2]中国人民解放军65583部队

出　　处：《软件学报》2013年第9期2182-2195,共14页Journal of Software

基　　金：国家自然科学基金(61073154)

摘　　要：现有的丢包主动测量方法是通过探测流的丢包信息去推测网络的丢包特性,进而推测特定应用流的丢包,测量结果不能准确获知某一给定应用流的丢包情况.由于丢包通常属于短时间、小概率事件,要更加准确地测量丢包就意味着需延长测量时间,或者提高探测流的发送速率以及时发现丢包,这将不可避免地增加网络的额外负载.分析了不同类型帧损伤的影响,并以MPEG-4,H264视频为研究对象,通过对其码流结构特点及RTP封装策略的分析,提出一种将测量信息嵌入到视频用户数据域(User_Data)的丢包测量方法 PLBU(packet loss measurement based on User_Data).该方法利用视频码流信息完成对丢包的探测,不影响视频的正常播放,不需要注入新的探测流,极大地降低了因测量而引入的额外负载.NIST Net及Planetlab等实验结果表明,该算法不仅丢包测量准确性高,且可提供丢包所属视频帧类型等信息,如视频中I,P,B帧的数据包丢失的情况.借助该测量方法,服务提供商可评测网络视频流丢包,并分析视频体验质量(QoE)变化情况,且不受视频流在网络传输中的优先级影响.The accurate packet loss measurement of application traffic has been challenging, since the frequently-used active monitoring approaches that use probe packets cannot capture the packet loss experienced by the traffic of individual user applications. Moreover, measurement and estimation of packet loss characteristics are challenging due to the relatively rare occurrence and typically short duration of packet loss episodes. Because it is often impossible to measure the entire data passing across a network, the most widely used method of measurement works by injecting probe packets. In order to overcome these shortcomings, this paper based on the analysis of MPEG4 video structural features and RTP encapsulation strategy, a novel packet loss probing technique, packet loss measurement based on User Data （PLBU）, is proposed. The PLBU complete packets loss measurement by using User Data field of MPEG-4 video. This does not affect the normal video playback and dose not need to inject new probing stream, which makes it, less intrusive. NIST Net andPlanetlab experiments verify the effectiveness of this method, and show that PLBU estimates packet loss more accurately, and with less measurement overhead than other existing technique such as BADABING.

关 键 词：主动测量 视频 用户数据域 丢包 视频帧 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]