Using Steady- State TCP Behavior for Proactive Queue Management

This paper describes and evaluates the use of steady-state TCP behavior knowledge for proactive queue management (PQM). We show how routers can use flow parameters and the knowledge of TCP end-host behavior to intelligently mark packets to prevent queue build-up, and thus prevent congestion from ever occurring, while at the same time provide high link utilization and low packet loss. This is in contrast to traditional active queue management (AQM) schemes, where incipient congestion is detected early and then reacted to. Using a queue management algorithm that tracks TCP behavior allows queue sizes to be deterministically stabilized at very low levels, irrespective of the number of flows passing through the router. An advantage of our algorithm is that there are no parameters that need to be tuned to achieve optimal performance in a given scenario, thus avoiding repeated trials with different parameters – an approach common to other AQM schemes. Further we show that our scheme out-performs the RED and BLUE AQM schemes with respect to queueing delay and performs as well with respect to link utilization and packet loss. Keywords—Congestion control, active queue man-