SCTP do not use the whole link capacity

Note: I already post this question on the ns-2 users mailing list, the Admin of this forum told me to try to ask my question in here.

The original question:

Hello,

I want to run simulation using SCTP. I’m using the ns-2.33. When I ran
successfully several simulation and analyzed the output I realized that
SCTP is not able to use the whole link capacity. I compared the link
occupation with a TCP-SACK agent which better use the link.

I’m using the default parameters on a link with those characteristics:
bw: 5Mb
delay: 15ms
queue limit: 50
Taildrop

I check at the cwnd and it do not go over 66060 bytes, it really stick
to this value no fluctuation is observed at all. I check the debug file,
this value of cwnd trigger the “congestion avoidance” mode of SCTP.

Maybe some of you already solve this issue ? Any clue is welcome 🙂

Thank you,

Now, after a lot of deep search in the NS code, I realized that the option :

set initialRwnd_

was not taken into account and always remains at the default value (65536), so block the sender to send more data.

I didn’t find the error in the code, I actually have already lost lot of time finding the problem. I have hardcoded a really huge value in the Reset() function of SCTP.cc file to remove the problem.

if you have any other solution, do not hesitate to inform me 🙄