PROJECT TITLE :

When Heavy-Tailed and Light-Tailed Flows Compete: The Response Time Tail Under Generalized Max-Weight Scheduling

ABSTRACT:

This paper focuses on the planning and analysis of scheduling policies for multi-category queues, like those found in wireless networks and high-speed switches. During this context, we have a tendency to study the response-time tail underneath generalized max-weight policies in settings where the traffic flows are highly uneven. Specifically, we have a tendency to contemplate a setting where a bursty flow, modeled using serious-tailed statistics, competes with a a lot of benign, lightweight-tailed flow. During this setting, we tend to prove that classical max-weight scheduling, which is known to be throughput optimal, ends up in the sunshine-tailed flow having significant-tailed response times. However, we show that via a careful design of inter-queue scheduling policy (from the category of generalized max-weight policies) and intra-queue scheduling policies, it is potential to keep up throughput optimality, and guarantee light-weight-tailed delays for the light-tailed flow, while not affecting the response-time tail for the serious-tailed flow.