Issues in Instantiating POWDER-OAI

114 views
Skip to first unread message

Ibrahim Moazzam

unread,
Aug 25, 2021, 3:59:24 AM8/25/21
to Powder Users
Screen Shot 2021-08-25 at 12.48.25 PM.png
Hi! I've been wanting to try out the POWDER-OAI profile, but I've been getting this error whenever I try to instantiate it. The parameters I've chosen are:
  1. UE Shuttle Line: Red Line
  2. First eNodeB site: Friendship Manor
  3. Second eNodeB site: Friendship Manor
  4. No second UE
FM was the only resource that was available, which is why I chose the Friendship Manor. 

Any help would be appreciated. Thank you!

Kirk Webb

unread,
Aug 25, 2021, 11:46:31 AM8/25/21
to powder...@googlegroups.com
Hi Ibrahim,

Only the CBRS radios at POWDER are currently available to users for
outside over the air work at this time (TDD mode, CBRS frequencies,
and no RF front-ends). This profile is incompatible with those
resources. We will not have an OAI profile that works outside for at
least a month, at which point we hope to have band 41 TDD
functionality available with COTS UEs and OAI eNB/gNBs.

In the meantime, if you want to use OAI on POWDER, you can do so via
the PhantomNet controlled RF resources. Dustin from our team should
be able to point you at an appropriate profile.

-Kirk
> --
> You received this message because you are subscribed to the Google Groups "Powder Users" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to powder-users...@googlegroups.com.
> To view this discussion on the web visit https://groups.google.com/d/msgid/powder-users/9ea525c2-cd57-407c-8629-3c31911897c7n%40googlegroups.com.

Dustin Maas

unread,
Aug 25, 2021, 5:36:09 PM8/25/21
to powder...@googlegroups.com

Hi Ibrahim,

I’d suggest you try out the develop branch of the following profile (be sure to select the develop branch):

https://www.powderwireless.net/instantiate.php?profile=2f49dd30-aea9-11eb-b1eb-e4434b2381fc&from=manage-profile&refspec=refs/heads/develop

The default parameters for this profile/branch will deploy an OAI 5G SA experiment using one of our controlled RF test benches (NI X310s with wired RF connections and a shared OctoClock reference). Be sure to wait for all startup scripts to finish before you proceed with the included instructions, which show how to sync the OAI UE with the OAI gNB/5GCN, and pass some user-plane traffic back and forth.

-Dustin

Ibrahim Moazzam

unread,
Aug 26, 2021, 7:48:32 AM8/26/21
to Powder Users
Hi Dustin,

Thank you for the prompt response. I tried initializing this profile but I've been getting the following error. I believe I don't have access to the X310 nodes, if possible, could you please give me access to these nodes?

Best,
Ibrahim
Screen Shot 2021-08-26 at 4.42.41 PM.png

Dustin Maas

unread,
Aug 26, 2021, 12:42:15 PM8/26/21
to Powder Users

Hi Ibrahim,

You should now have access to these radios.

-Dustin

Ibrahim Moazzam

unread,
Aug 30, 2021, 4:20:43 AM8/30/21
to Powder Users
Hi Dustin,

Thank you, I have access to these radios now. Unfortunately this profile isn't really what I'm looking for, since I was trying to simulate handover using OAI (with 2 eNBs). Would you know any profiles that fit that description?

Best,
Ibrahim

Dustin Maas

unread,
Aug 30, 2021, 6:51:35 PM8/30/21
to Powder Users

Hi Ibrahim,

I’m not aware of a public profile that deploys an OAI-based handover experiment in our controlled RF environment, but we’re checking around within the group to see if somebody has one we could make available. I’ll let you know if we find something suitable. Here’s another option though..

I think you are already familiar with the srsran-handover profile, which implements S1 handover using srsRAN/Open5GS and provides the same topology and attenuation control you’d need for OAI-based S1 or X2 handover. I think it should be relatively straightforward to adapt this profile to use OAI instead of srsRAN.

To get started, you could instantiate an experiment based on the srsran-handover profile, clone/build OAI at each eNB, and follow the relevant steps in this OAI X2 handover tutorial. You could fork/adjust the srsran-handover profile to automate all of the OAI setup after you get to a working solution. I’d be happy to help you debug if you go this route and run into issues getting it to work.

-Dustin

Ibrahim Moazzam

unread,
Aug 31, 2021, 3:29:07 AM8/31/21
to Powder Users
Thanks for the suggestion, Dustin! Unfortunately nucs 1, 2, and 4 are unavailable until the 31st of October according to the Available Resources page. Could we use any other nucs instead for the srsran-handover profile?

Dustin Maas

unread,
Aug 31, 2021, 12:13:27 PM8/31/21
to Powder Users

Those nodes were in a funny state for some reason. They should be available again now.

In general, there are two overlapping sets of NUCs that have the necessary connections in the matrix for this profile: {nuc1,nuc2,nuc4} and {nuc2,nuc3,nuc4}.

-Dustin

Ibrahim Moazzam

unread,
Sep 3, 2021, 7:30:33 AM9/3/21
to Powder Users
Hi Dustin,

I hope you're doing well. Since I'm relatively new to networks, I've been having some trouble figuring out how to clone/build OAI at each eNB. I tried following a tutorial online but it leaves my eNBs unresponsive. If possible, could you please send me instructions on building OAI at eNBs? 

Thank you!
Ibrahim

Dustin Maas

unread,
Sep 3, 2021, 12:02:41 PM9/3/21
to Powder Users

Hi Ibrahim,

Can you say what you mean by unresponsive? Sometimes the OAI gitlab server can be a little slow, so it might take a while to clone the repo.

Here is how to build the OAI eNB, assuming you are starting with the srsran-handover profile:

git clone https://gitlab.eurecom.fr/oai/openairinterface5g
cd openairinterface5g
source oaienv
cd cmake_targets
./build_oai -I
./build_oai -w USRP —eNB

Best,
Dustin

Dustin Maas

unread,
Sep 3, 2021, 4:31:31 PM9/3/21
to Powder Users

Hi Ibrahim,

This paper and the accompanying profile might be useful to you. The profile implements (among other things within the broader context of the paper) inter-frequency handover using OAI-based eNBs and a COTS UE in a controlled RF environment.

-Dustin

Ibrahim Moazzam

unread,
Sep 6, 2021, 2:12:16 PM9/6/21
to Powder Users
Hi Dustin,

Thank you for finding the paper and accompanying profile. However, I was trying out the code you'd sent for building the OAI eNB, and I keep getting the following error when I type in "source oaienv". Would you happen to know why I've been getting the error?
Screen Shot 2021-09-06 at 11.11.50 PM.png
Best,
Ibrahim

Kirk Webb

unread,
Sep 7, 2021, 11:09:15 AM9/7/21
to powder...@googlegroups.com
The default shell on nodes is tcsh. Run "/bin/bash" first before sourcing that file.


Ibrahim Moazzam

unread,
Sep 23, 2021, 4:37:04 PM9/23/21
to Powder Users
Hey Dustin,

I was trying to configure the BoTM profile that you'd recommended, however I've been getting the following error whenever I log into enb1 and run the following command: sudo -E ~/openairinterface5g/targets/bin/lte-softmodem.Rel14 -O ~/openairinterface5g/targets/PROJECTS/GENERIC-LTE-EPC/CONF/enb.band7.tm1.50PRB.usrpb210.conf
All of the dependancies were installed smoothly too. Please let me know if you have any fixes for the error. Thank you!

Screen Shot 2021-09-24 at 1.24.17 AM.png

Best,
Ibrahim

Dustin Maas

unread,
Sep 28, 2021, 6:12:23 PM9/28/21
to Powder Users
Hi Ibrahim,

It looks like the FlexRAN agent is unable to connect to the RAN or MANO controller for some reason. I'd double-check that the rest of the processes have been started and are still running before starting the eNB. I'm not very familiar with FlexRAN or this particular profile, but @Aashish Gottipati (paper author and profile creator) might be able to help troubleshoot if you continue to see this issue.

Best,
Dustin

Ibrahim Moazzam

unread,
Sep 30, 2021, 8:19:10 AM9/30/21
to Powder Users
Hi Dustin,

Yeah, I made sure that the rest of the processes were up and running before I started the eNB. I'll try running the experiment again, and will contact Aashish if the issue persists. Could you please let me know how I can contact Aashish in that case?

Best,
Ibrahim

Reply all
Reply to author
Forward
0 new messages