Heterogeneous Network Traffic Modeling and Analysis in Wavelet Domain

随着多媒体流量开始主导宽带网络，视频和数据流量的准确建模对许多重要的网络应用变得至关重要。视频和数据流量最近被发现具有复杂的自相似性，使先前开发的流量模型失效。其中一个最重要的问题是如何开发计算效率高且准确的模型，以捕获网络设计/控制的视频和数据流量的复杂统计特性。我们的研究目标是开发一个统一的模型，该模型能够(1)模拟包括VBR视频和数据在内的异构多媒体流量，(2)计算效率高，(3)分析可行。为了实现我们的目标，我们建议使用小波。我们将表明，尽管异构多媒体流量具有复杂的短期和长期时间依赖性，但相应的小波系数不再是长期依赖的。因此，可以使用简单的方法在小波域对网络流量进行建模。特别是，独立的小波模型已经被证明是足够精确的，简洁的，并且具有最低的计算复杂度。我们将集中精力研究如何使用独立小波模型进行分析，包括反相关和缓冲损失率。我们还将研究如何将小波模型应用于网络控制，如准入控制。

As multi-media traffic begins to dominate broadband networks, accurate modeling of video and data traffic becomes crucial to many important network applications. Video and data traffic has recently been found to possess complicated self-similarity that invalidates previously developed traffic models. One of the most important problem is how to develop computationally efficient and yet accurate models which could capture the complicated statistical properties of video and data traffic for network design/control. Our objective of this research is to develop a unified model which is (1) able to model heterogeneous multi-media traffic including VBR video and data, (2) computationally efficient, and (3) feasible for analysis. To accomplish our goal, we propose to use wavelets. We will show that although heterogeneous multi-media traffic has the complicated short- and long-range temporal dependence, the corresponding wavelet coefficients are no-longer long-range dependent. Therefore, simple methods can be used to model network traffic in the wavelet domain. In particular, independent wavelet models have shown to be sufficiently accurate, parsimonious, and have the lowest computational complexity attainable. We will focus our efforts on investigating how to use independent wavelet models for analysis including anto-correlation, and buffer loss rate. We will also investigate how to apply wavelet models to network control such as admission control.