mptcp integration with ns3.23

[HA]

Hi,

I have the following questions regarding mptcp module:

1. Does the current mptcp module work and integrate perfectly with ns3.23? 

2. Regarding the performance is it close to the real implementation of mptcp in Linux Kernel? Does it have any limitations or bugs?

3. Are you willing to merge the code with ns3 later on?

Thanks a lot.

[Benevid Felix]

Hi, i am made this update (3.19->3.23) with https://github.com/mkheirkhah/mptcp and it's work fine.

But, has too the ns3-dce what use kernel modules emulation.

bfs

[Vikas Raunak]

Hi

How do acknowledgements work in ns3 ? Based on the acknowledgements, the ns3 model for MPTCP developed by Chihani et al. is totally wrong, I guess. Consider this case:

I run 2 subflows with different delays: The congestion window behavior is attached.

Why is one congestion window so suppressed ? 

I am very confused regarding this issue.

Please help. 

Thanking you

Regards

[Vikas Raunak]

I meant (if Data Acks are used to penalize subflow congestion windows then only that behavior is possible) that only subflow level acks can lead to a subflow penalization right ?

--
Vous recevez ce message, car vous êtes abonné à un sujet dans le groupe Google Groupes "mptcp".
Pour vous désabonner de ce sujet, visitez le site https://groups.google.com/d/topic/mptcp/SK9BEa3Cbys/unsubscribe.
Pour vous désabonner de ce groupe et de tous ses sujets, envoyez un e-mail à l'adresse mptcp+un...@googlegroups.com.
Pour obtenir davantage d'options, consultez la page https://groups.google.com/d/optout.

[Vikas Raunak]

As MPTCP has 2 levels of ACKS.

[Dejene Boru]

Hi Benevid Felix,

Does this https://github.com/mkheirkhah/mptcp  MPTCP code in ns-3 work correctly can you confirm if you have done some simulations or made changes? 

I found out that this implementation of MPTCP is not compatible with the standard especially when it shares link with TCP flows. It is also very much dependent on the link buffer (queue) settings.

Thanks,