Fair Rate Degradation in Flow-Aware Networks

Abstract

In this paper, it is shown that the admission control routine in Flow-Aware Networks (FAN) may lead to severe fair rate degradation, which negatively impacts the performance of streaming applications. In order to prevent this negative behavior, the limitation mechanism is proposed. The aim of the mechanism is to limit the maximum number of new flows that may be admitted on a link between any two consecutive network's automeasurements. The solution is efficient, viable and dramatically reduces the fair rate degradation.

Enhanced measurement-based admission control for flow-aware networks

“…However, as shown in [12], this assumption does not hold true, as when many new flows arrive at the same instant, the thresholds can be exceeded significantly.…”

“…This approach protects the MBAC from over-admitting, i.e., from allowing too many new flows to acquire access to the link. Through simulations, it was shown in [12] that reducing the inter-measurement time even 10 times did not provide better performance than introducing even a very simple limitation mechanism. This shows that increasing the frequency of measurements is a considerably worse option for combating the FR degradation problem.…”

“…It was shown in [12] that the standard admission control routine in FAN may lead to severe quality degradation. Also, a simple remedy to the problem was proposed.…”

“…It was shown in [12] that frequent degradations of the FR may occur on FAN links when there are too many flows attempting to acquire access to the link bandwidth. The root of the FR degradations in FAN lies in the very design of the admission control block.…”

“…The method presented in [12] is the simplest, and surprisingly very efficient. In this section, we propose to enhance the method by adjusting the admission limit according to current FR values.…”