[Iccrg] Use of CUBIC in Rate-based Congestion Control

[Chamil Kulatunga]

Dear Rhee,

In TFRC throughput equation, RTT stays in the denominator (lower part). 
When the bottleneck queue fills, RTT increases. Then the sender reduces 
its rate according to the developing queue when moves from one 
loss-event to the next. Therefore queuing delay is considered.

But in CUBIC, RTT stays in the numerator (upper part). According to RTT 
increase due to queuing delay, the sender increases the rate after a 
loss-event until it meet the next loss.

This may give some contradictory representation if used in any 
equation-based congestion control. Will this restrict using CUBIC with 
equation-based approaches?

Chamil Kulatunga
Communications and Imaging Research Group
Department of Engineering
University of Aberdeen
Aberdeen, AB24 3UE
United Kingdom

Tel: + 44 1224 272813
Fax: + 44 1224 272497