Delay and latency in Lte Scenarios

[Madan Pande]

        Using a scenario with one Enb, and multiple UEs, placed randomly at different distances within one cell, I am trying to find how best to saturate the Enb (using either PfFf or PssFf Schedulers). As otherwise,  given a Propagation Loss Model and Fading model, the delay in packet in most of the flows does not change significantly.

 2.         Two approaches are obvious. One can increase the number of UEs and the number of Tcp/Udp applications...While increasing number of UEs is quite easy, impact of increasing the number of applications has an opacity in the PGW and Enb applications... [Using NetAnim,FlowMonitor and Tracemetrics tool for the analysis].

 3. IMHO,  most likely the errors of Packet loss is caused by some latency in Pgw / Enb or insufficient routing precautions...It may be incorrect to attribute packet losses coming up while running Lte-Epc scenarios, with number of UEs crossing 30, just to either pathloss or mutual interference.The opacity of Pgw/Sgw is a cause of concern here...

4.  Questions that I am seeking answers for are:

i) is the overall end-to-end LteEpc delay/latency impacted by some buffer becoming full in the Pgw/Sgw or Enb before the packet gets routed to the UE or vice-versa...as the number of flows increase? any simple way to figure this out?

ii) Is it due to the inherent latency in the Pgw and Enb, and is such a latency/delay constant or variable with loading? how to figure this out?

iii) Does one need to add more Pgw/Sgw (and even Remote Host) while increasing the number of flows?

Any suggestions are welcome...

With Regards

   


 

[Nicola Baldo]

Hi, please see my comments inline

On Wednesday, March 13, 2013 11:19:48 AM UTC+1, mador...@hotmail.com wrote:

i) is the overall end-to-end LteEpc delay/latency impacted by some buffer becoming full in the Pgw/Sgw or Enb before the packet gets routed to the UE or vice-versa...as the number of flows increase? any simple way to figure this out?

try to compare end-to-end measurements with PDCP measurements which are relative to the radio link only.
 

ii) Is it due to the inherent latency in the Pgw and Enb, and is such a latency/delay constant or variable with loading?

by default the link between the PGW/SGW node and the eNB is a point to point link with given bandwidth and delay. Hence this contribution to the latency will vary with load. I recommend also to check the network between the PGW and the remote host, depending on your setup you might have a bottleneck there.
 

how to figure this out?

see my previous comment
 

iii) Does one need to add more Pgw/Sgw (and even Remote Host) while increasing the number of flows?

multiples SGWs within the same LTE network are not supported in LENA. On the other hand, adding multiple remote hosts (e.g., with separate point to point links to the PGW) is a good idea.

[Madan Pande]

Nicola,
          
Thank You for your detailed reply...I will do as directed...

Regards,
madan

---

Date: Sun, 17 Mar 2013 13:05:18 -0700
From: nba...@cttc.es
To: ns-3-...@googlegroups.com
Subject: Re: Delay and latency in Lte Scenarios

--
You received this message because you are subscribed to the Google Groups "ns-3-users" group.
To unsubscribe from this group and stop receiving emails from it, 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?hl=en.
For more options, visit https://groups.google.com/groups/opt_out.
 
 

[Sokratis Barbounakis]

Dear Nicola, 

You state:

by default the link between the PGW/SGW node and the eNB is a point to point link with given bandwidth and delay. 

 

How can we get this bandwidth and delay? 

I assume from your statement, that we cannot change these parameters?

Thanks a lot!

Sokratis