Packet delay increased from ns-3.24 to ns-3.25

classic Classic list List threaded Threaded
2 messages Options
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Packet delay increased from ns-3.24 to ns-3.25

Ben Newton
I've noticed a significant increase in the mean delay experienced by
packets in one of my simulations after moving to ns-3.26 from ns-3.24.  I
believe the changes to queues in ns-3.25 are the cause.  Does that sound
likely?  In some cases I'm seeing mean one way delays (reported by
flowMonitor) that are double their previous values, even when I set the
queue size to 1.

I'm wondering if this is expected, and if the new values are more accurate?


I can generate some simple examples if that would be beneficial.

Thanks,
  Ben Newton
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: Packet delay increased from ns-3.24 to ns-3.25

Stefano Avallone
Hi Ben,

you are correct, the increased delay is likely due to the current default
queue disc (PfifoFast) and its default size (1000 packets).

You might try to reduce the value of the ns3::PfifoFastQueueDisc::Limit
attribute to see if it helps.

Actually, there were some proposals to change the defaults but I recall no
agreement was met.

Bests,
Stefano


On lunedì 17 ottobre 2016 09:43:32 CEST Ben Newton wrote:

> I've noticed a significant increase in the mean delay experienced by
> packets in one of my simulations after moving to ns-3.26 from ns-3.24.  I
> believe the changes to queues in ns-3.25 are the cause.  Does that sound
> likely?  In some cases I'm seeing mean one way delays (reported by
> flowMonitor) that are double their previous values, even when I set the
> queue size to 1.
>
> I'm wondering if this is expected, and if the new values are more accurate?
>
>
> I can generate some simple examples if that would be beneficial.
>
> Thanks,
>   Ben Newton



Loading...