Using MPI in LTE

[Luis Diez]

Hello all,

first of all, sorry if this issue is duplicated, I don't think so.
I'm currently working with the LTE module and wonder if it is possible to use MPI for a LTE scenario. I would like to speed up the simulation time.
Following the MPI documentation, the nodes (or node containers) can be assigned to different logical processes (LP). Then a we can use a point-2-point link and connect different bunches of nodes. This is easy to use for topologies such as the third-distributed.cc,but I'm not able to see how to do it for LTE.

While we (following the examples) use a point-2-point link between the PDN gateway and the remote network, the heavy part of the simulation is all in the access network. In the access network we could assign each eNB to different LPs, but the eNBs are not independent and the UEs may change its access element.
Surely I missed something. Any idea, suggestion (lesson :) ) is more than welcome.

Best regards,
Luis

[Konstantinos]

Hi Luis,

As mentioned in the ns-3 documentation, pure distributed wireless is currently not supported. 

This means that you could have your LTE module in one LP and the remote host to another. 

However, I do not think that would increase the performance.

[Marco Miozzo]

Hi Luis,

the problem in assigning different LPs to to eNBs is that the eNBs strongly interact with each other in terms of interference and possibly handover, so I'm a bit skeptical it works.
Recently, Nicola and me were discussing on this, based on our very preliminary and informal analysis probably it should better to generate LPs per PHY link basis, where actually is concentrated most of the computational complexity (in detail in the interference evaluation and in the TBLER evaluation, but the latter can be disabled).

Best regards,

marco.

--
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.
For more options, visit https://groups.google.com/d/optout.

[Luis Diez]

Hello,

maybe I closed the thread to early, but the answer from Konstantinos was pretty clear. Thank you so much for your response, I was quite sure that it was not easy to use MPI with LTE but I had to ask. I'll try to follow your indications, but it would require diving into how the physical components get connected rather than using the scenario as in other examples, right?
Nice to hear that you are, at least, considering this option.

BR,
Luis

[Nicola Baldo]

On Tuesday, September 23, 2014 11:07:06 AM UTC+2, Luis Diez wrote:

maybe I closed the thread to early, but the answer from Konstantinos was pretty clear.

Kostantinos is right in that MPI is currently not supported with LTE
 

Thank you so much for your response, I was quite sure that it was not easy to use MPI with LTE but I had to ask. I'll try to follow your indications, but it would require diving into how the physical components get connected rather than using the scenario as in other examples, right?

Exactly, basically you would need to develop an LTE-specific SpectrumChannel that partitions transmissions/receptions on different LPs. It should work, but it's hard to say how much speedup you might get, if any.

[Ismael Al-Shiab]

Hello All,

I started working on this and would like to get the updates. did you achieve successful implementation of MPI with LTE in ns-3 ? any statistics for the simulation time enhancement?

[NS3SudoUser]

I am also currently working on the same subject. Any progress on the successful implementation of MPI with LTE in ns-3?

[Franci Gjeci]

Hello, 

I was asking the same question as all of you. Has anybody made any substantial progress?

Thanks