Video-aware measurement-based admission control

TL;DR

This paper proposes using the average aggregate arrival rate instead of the instantaneous rate for video admission control, improving decision stability and efficiency especially with fewer flows.

Contribution

It introduces a mathematical model comparing average and instantaneous rates and demonstrates the benefits of using the average rate for admission decisions in video traffic.

Findings

Average rate improves decision stability for few flows.

No significant advantage for moderate/large flows.

Smoothing burstiness enhances admission control reliability.

Abstract

Using instantaneous aggregate arrival rate as an admission control parameter will contribute to either bandwidth under-utilization or over-utilization. Being bursty in nature and variable in rate, video flows might encode any rate between a range of minimum and maximum values. At the time the decision is made, if the measured rate is at the minimum value, the bandwidth might be over-utilized due to accepting more sessions than the link can accommodate. In contrast, it might be under-utilized if the measured rate is at the maximum value due to rejecting more sessions than the link can accommodate. The burstiness can be taken into account by considering the past history of the traffic. This paper investigates the suitability of the average aggregate arrival rate instead of the instantaneous aggregate arrival rate for video admission decisions. It establishes a mathematical model to predict the relationship between the two rates. Simulation results confirm that the average aggregate arrival rate is a more efficient decision factor for a small number of flows. Although it has no additional advantage for moderate and large number of flows, it still can stabilize the admission decision by smoothing the burstiness of a set of the instantaneous rates (within the measurement period) over a period of time.