Calculating Propagation delay on NS-3

739 views
Skip to first unread message

Mallesh Yadhav

unread,
Aug 28, 2015, 5:53:21 AM8/28/15
to ns-3-users
Hi,

        I am able calculate the transmission delay but not propagation delay on ns3. Is there any way to calculate the same at the MAC layer.

Thanks & regards,
   Mallesh

Konstantinos

unread,
Aug 28, 2015, 6:02:33 AM8/28/15
to ns-3-users
Hi,

Signals propagate with the speed of light, so the propagation delay for WiFi is in the order of 1-2us (common values used in the the literature).
Note also that if you use FlowMonitor (that you have as tag in your post), it accounts for both propagation and transmission delay as it is End-to-End at IP layer.

Regards,
K

Mallesh Yadhav

unread,
Aug 28, 2015, 6:10:10 AM8/28/15
to ns-3-users
Thank you very much for the quick response.

      In ns3, there are two propagation models: constant and random propagation models. I want to know the propagation delay of packets under mobile scenarios with random propagation model.

      Is it possible? Or as you said, is not worth calculate the propagation delay as the order is of single digit micro seconds.

Konstantinos

unread,
Aug 28, 2015, 6:21:30 AM8/28/15
to ns-3-users
It will depend on the random variable you use. 
If you check the source code for RandomPropagationDelay

76 Time
78 {
79  return Seconds (m_variable->GetValue ());
80 }

The delay does not relate to the position of the nodes, but it comes from the random variable you use; by default Uniform from (0,1)seconds. 
So, if you keep the default configuration, you should NOT discard it, but it is not realistic for simple WiFi at those ranges.

But if you want my advice, I would use the ConstantSpeedDelay that takes into account the distance of the nodes.

Tommaso Pecorella

unread,
Aug 28, 2015, 6:25:22 AM8/28/15
to ns-3-users
Hi,

the RandomPropagationDelayModel (please use the full name to avoid confusion) is mainly used in tests and it doesn't represent anything realistic. I mean... under this model the propagation delay between every node pair is totally random - by default between 0 and 1 second. But wait, there's more. It's not only random between each pair of nodes, it's different for each packet sent !
Basically... useful only to stress test stuff.

Have fun,

T.
Reply all
Reply to author
Forward
0 new messages