TCP congestion window analysis

[Vineet]

Hi,

I have a network with a UDP and a TCP traffic source. I am comparing the TCP friendliness of the UDP source equipped with different rate control techniques namely rate control 1 and rate control 2..The TCP source starts pumping in data at 1000ms at which point in time the UDP source triggers the rate control experiments. I plot (shown below) the congestion window variation of TCP source with time under the influence of each rate control technique. But i am unable to comprehend the graph. I will be grateful if someone could help me in analyzing which of the rate control techniques is better in terms of TCP friendliness.

Thanks in advance!

[Nat P]

Il giorno venerdì 19 giugno 2015 11:41:32 UTC+2, Vineet ha scritto:

Hi,

I have a network with a UDP and a TCP traffic source. I am comparing the TCP friendliness of the UDP source equipped with different rate control techniques namely rate control 1 and rate control 2..The TCP source starts pumping in data at 1000ms at which point in time the UDP source triggers the rate control experiments. I plot (shown below) the congestion window variation of TCP source with time under the influence of each rate control technique. But i am unable to comprehend the graph. I will be grateful if someone could help me in analyzing which of the rate control techniques is better in terms of TCP friendliness.

The graph shows the congestion window over the time. Rate control 1 seems the old good NewReno, while I don't know what is the black line.

After re-reading your email, I think the question is bad-formatted:

- What are you asking for ?
- Do you have provided all information which could be useful to comprehend your system, experiment, and analysis ?

Nat

[Vineet Gokhale]

Sorry for the incomplete details!

The major questions i have about this graph are the following:

1. The graph is the response of the TCP source to the customized rate control techniques (rate control 1 and rate control 2) implemented on top of UDP. So, as far as TCP friendliness is concerned which of these rate control techniques perform better? What is the parameter to measure it?

2. The TCP congestion control technique (ns3 default) is unchanged for both experiments (rate control 1 and rate control 2). Then, why is the congestion window behavior different in both cases? For the black graph, it always makes a very sharp increase while for the red graph it alternates between slow and sharp increase. 

Thanks in advance!

--
You received this message because you are subscribed to a topic in the Google Groups "ns-3-users" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/ns-3-users/DIV0robWYaA/unsubscribe.
To unsubscribe from this group and all its topics, send an email to ns-3-users+...@googlegroups.com.
To post to this group, send email to ns-3-...@googlegroups.com.
Visit this group at http://groups.google.com/group/ns-3-users.
For more options, visit https://groups.google.com/d/optout.

[Tommaso Pecorella]

Hi,

On Friday, June 19, 2015 at 12:55:40 PM UTC+2, Vineet wrote:

Sorry for the incomplete details!

The major questions i have about this graph are the following:

1. The graph is the response of the TCP source to the customized rate control techniques (rate control 1 and rate control 2) implemented on top of UDP. So, as far as TCP friendliness is concerned which of these rate control techniques perform better? What is the parameter to measure it?

TCP friendliness is a vague concept, meaning that is not defined formally. You have to decide what is "friendly", possibly in a mathematical way.

I could suggest to measure the TCP retransmissions. The less, the better.

 

Another parameter (formal this time) could be the TCP fairness, which is the bandwidth ratio shared by the flows, i.e., two flows should (ideally) share the available bandwidth equally.

2. The TCP congestion control technique (ns3 default) is unchanged for both experiments (rate control 1 and rate control 2). Then, why is the congestion window behavior different in both cases? For the black graph, it always makes a very sharp increase while for the red graph it alternates between slow and sharp increase. 

See above. It depends on the quantity of lost segments and by their distribution (if they are lost in batches or once in a while). Please study the particular TCP flavour you're using for further details.

However, mind that this is your own research. It's very strange to ask us to explain to you what your simulation results means.

Have fun,

T.

[Vineet]

Thanks for the insightful replies!