Network modelling and measurement: Whence, Where and Whither

  • by

Darryl Veitch, University of Technology Sydney

There is a long established history of the interplay between telecommunication networks and queueing theory. The rise of the Internet brought with it the availability of large data sets (with surprising features), and highlighted new problems of practical importance for network operators and end applications. This led to a burst of activity in applied queueing, and interest in new kinds of problems, including those with closer links to inference. The discipline of network measurement meanwhile explored many network inference problems, including those directly related to queueing systems, as well as others inspired by, or at least informed by, queueing considerations and insights.

Twenty years on, networks have changed, and the literature has matured. We are now entering into a new era where old assumptions fail, the questions are different and more complex, but where many more opportunities are opening up. In this talk I will describe this changing landscape, beginning with an overview of the old approaches, and their limitations. I will then try to map out some of the new directions of interest in the space, which should contain rich opportunities for expansion of the queueing frontier.