Multimedia traffic queueing analysis in high-speed networks: A frequency domain approach

Multimedia traffic in high speed networks possesses correlation and burstiness properties. Classical queueing theory has generally ignored such properties by making renewal assumption on message interarrival time. Developing a new traffic theory for integration of multimedia services on high speed networks becomes absolutely essential. In this dissertation, we explore a new concept of spectral characterization of wide-band input process in high speed networks. It provides us a much richer and heterogeneous input environment, while keeping the complexity of queueing analysis tractable. The correlation nature of multimedia traffic is well captured by the input spectral functions. The queue response to the power spectrum and higher order statistics has been studied. Our study shows that the input power spectrum is the most essential statistic for queueing analysis. Further, the input power in low-frequency band has dominant impact on queueing performance, whereas the high-frequency power to a large extent can be neglected. Understanding of queue response to input spectrum provides us a great deal of knowledge to develop advanced network traffic measure theory, and to introduce effective network resource allocation policies. The interrelationship between the traffic spectral statistics and link capacity allocation in high speed network has been investigated. Finally we developed a sophisticated computational tool, called SMAQ, that can integrate traffic measurement and queueing analysis for stochastic modeling. The SMAQ tool takes a unique frequency-domain approach to combine the techniques of signal processing and performance analysis. Essentially, signal processing techniques are used to obtain the steady-state and second-order statistics of a traffic stream. The focus here is on the construction of a special class of Markov chains that can statistically match with each given traffic stream (or superposition of different traffic streams). The analytical queueing solutions can therefore be obtained by the Folding-algorithm based on the Markov chain input modeling. Comprehensive numerical examples show the great potential of the SMAQ tool to solve measurement-based traffic management issues.