中文摘要：

通过对VBR视频流量进行工程建模,结合IPTV用户点播时间间隔分析聚合VBR视频流量特性,研究表明在兼容现有终端和平台系统的情况下,即使不采用任何视频平滑处理,在IPTV点播业务中通过应用大规模VBR视频流的统计随机特性进行自我平均（self-average）也可获得30%～50%的统计增益,并通过大规模现场试验验证了VBR视频自平均特性在IPTV点播业务中的可行性和有效性。

英文摘要：

Based on the engineering model for VBR video traffic, the flow characteristics of VBR video based on user requested arriving time interval was analyzed. Studies show that in compatible with the existing IPTV terminal and platform system, even without the use of any video smoothing technology, self-average can be obtained 30% 50% statistical gain through using statistical random characteristics of large scale VBR video traffic in IPTV on demand business. The feasibihty and effectiveness of VBR video ＂self-average＂ characteristics in the application of IPTV on demand services were verified.