Video traffic will be a major load for future integrated services multimedia networks such as ATM networks, the Internet, and high-speed wireless networks. However, the bursty nature and strict Quality of Service (QoS) requirements of video traffic have been obstacles to efficient video transmission. Particularly, two properties of multiple time-scale burstiness and Long-Range Dependence (LRD) in VBR video traffic have been recently reported and are considered to significantly affect network performance.
In this dissertation, we address these two issues from the perspective of traffic control and network design for video transmission by investigating the relationship between the time scale in the correlation of video traffic and the queue buildup. Our main contributions are the proposal of a flexible and mathematically tractable LRD video traffic model, and introduction of two key time scales, cutoff and dominant time-scale which play a key role in Call Admission Control (CAC) and Usage Parameter Control (UPC).
We begin this dissertation by investigating the statistics and origin of the multiple time scale burstiness and LRD of VBR video traffic and relating with the current research activities and the status of traffic control standards.
As the first result, we present a LRD video traffic model based on the shifting-level (SL) process with an accurate parameter matching algorithm for video traffic. The SL process has its strength that it captures all key statistics of an empirical video trace, i.e., short- and long-term correlations and rate-distribution, while still retaining mathematical tractability. We devise a queueing analysis method of SL loaded system, named quantization reduction method. This method provides queueing results over all range of queue size, not just an asymptotic solution. Especially, we found that for most available traces its ACF can be accurately modeled by a compound correlation (SLCC): an exponential function in short range and a hy...