OAI Progress - next error


#1

Made some progress with OAI / LimeSDR - have a phone registered and some data flowing, can ping, etc. The last Major Malfunction was not understanding the security relationship between OP, Ki and OPc and the howto guides barely touch on them. I just picked this in hss.conf:

OPERATOR_key = “11111111111111111111111111111111”; # OP key matching your database

not knowing exactly what it should be, and using a non-related OPc and Ki in the user database.
After picking another key, and calculating the proper OPc using auchss.py { https://gitlab.fokus.fraunhofer.de/bri/osmo-sim-auth } the phone registers OK now.

Next error is this - mentioned in a previous thread, but the solution proffered there ( use the non 1v4.ini ) does not work for mine, actually seems to make it worse. Another discussion stated that could be a follow on error for ULSCH consecutive error count reached 20, triggering UL Failure

Anyone have experience how to get around this then?

[PHY][W][eNB 0, CC 0] frame 633, subframe 8, UE 0: ULSCH consecutive error count reached 20, triggering UL Failure
[MAC][I][UL_failure_indication] [eNB 0][UE 0/80e6] Frame 633 subframeP 8 Signaling UL Failure for UE 0 on CC_id 0 (timer 0)
[PHY][E]ERROR: Format 1A: rb_alloc (1ff) > RIV_max (144)


#2

Last obstacle overcome - here’s a video of a Galaxy S4 Mini playing a video from youtube over OAI, it’s been playing steady for the past hour.

What helps is running it with the xforms display and looking for a good trellis pattern, and I found with basic lmssdr power and no duplexer the ue needs to be very close, a couple of feet maybe, positioning is critical to get a good pattern and hence reliable communication. It’s using the band7 config which is 2.68GHz dl, which is only about 11cm wavelength so it does not take much to move a quarter wave!


#3

@cswiger - Chuck,

WHERE IS THE MILLION HEARTS BUTTON FOR THIS?!! INCREDIBLE FEAT…!! You rule…!!

73 de Marty, KN0CK


#4

Thanks Marty :blush:

Testing with AndSMB and a local file store gets about 150KB/sec (bytes) up/down speed test, which jives with an honest 1.5Mbps (bits) d/l speed test.

Here’s some better screenshots of working LTE UL SCOPE eNB displays, signals to look for.

Relatively idle:

Downloading data:

Moving on to researching an affordable hobbyist duplexer…
There’s a UPD007A about $23 but I cannot find quantities less than 100, and the 2mm sq cell phone duplexers at about $4 would be challenging to solder or make a breakout board.


#5

Last brag - testing with aperf / iperf and the phone positioned JUST RIGHT gets up to 8.1Mbps with this nice looking constellation:

[ 6] local 192.168.1.226 port 5001 connected with 172.16.0.4 port 39679
[ 6] 0.0- 1.6 sec 663 KBytes 3.29 Mbits/sec
[ 4] local 192.168.1.226 port 5001 connected with 172.16.0.4 port 47701
[ 4] 0.0-10.5 sec 7.51 MBytes 5.99 Mbits/sec
[ 6] local 192.168.1.226 port 5001 connected with 172.16.0.4 port 49518
[ 6] 0.0-10.6 sec 10.3 MBytes 8.10 Mbits/sec
[ 4] local 192.168.1.226 port 5001 connected with 172.16.0.4 port 59844
[ 4] 0.0-10.8 sec 10.4 MBytes 8.09 Mbits/sec
[ 6] local 192.168.1.226 port 5001 connected with 172.16.0.4 port 38994
[ 6] 0.0-10.7 sec 10.3 MBytes 8.09 Mbits/sec

Looks like channel training / adapting as each test progressed from 3.3, 6.0 and peaked at 8.10Mbps.


#6

Nice work, thanks for sharing!

Don’t suppose you’d be up for putting together a wiki page documenting how to recreate this?


#7

In progress https://wiki.myriadrf.org/OpenAirInterface_LTE


#8

Oh yes, a proper duplexer helps a lot - the folks at http://www.richardsonrfpd.com/Pages/Product-Details.aspx?productId=1152678 were kind enough to sell a small quantity (2) of their USD007A LTE Band7 ceramic duplexers for $38.60 each (incl 3% cc charge) + $21 FedEx ground from Reno NV. Using silver solder as recommended wired it up dead bug and gets much better signal around the room, nowhere near as picky in placement.


OAI and LimeSDR-Mini
#9

@cswiger, have try to using the mimo.ini for oai!!! how best it is compared to the classic ini file?


#10

Have not been in that space in a while - have not a clue.


#11

@cswiger I am emailing RidchardsonRFPD in order to order 2 pieces of USD007A but I have no reply from them, please how did you convince them to get your 2 pieces ??


#12

Here are my notes:

Karen Dallo-Shea
Inside Sales & Marketing Specialist
RichardsonRFPD/ Arrow RF & Power
Work: 631-851-2554
Toll Free: 800-737-6937 x3931
Email: kshea@richardsonrfpd.com
www.richardsonrfpd.com
Quotations are valid for a period of 30 days

Inputted a sales inquiry on the CTS website https://www.ctscorp.com/
got a reply from Thomas Strasser Thomas.Strasser@ctscorp.com that
Our recommended source for USD007A is Richardson RFPD.

www.richardsonrfpd.com
Currently they should have 27pcs available.
I replied that their minimum order is 100 units (even though only 27 are in stock!)

Karen chimed in:
Hello Chuck,
Karen Dallo-Shea here your account manager at Richardson RFPD.
CTS sent me your inquiry for USD007A.
We can support up to 27 pcs and 1-2 is fine.
They are $ 37.50 each, stock in Reno, NV.

Then:
Hello Chuck,
I am not sure if you would be able to purchase 2 pcs on our website @ $37.50 but you can try. If any issue or if you do not care to try I can help place the order for you.
For payment we accept CC but there is a 3% fee that needs to be applied and the way we do that is to include in the unit cost.

Cost for 2 pcs @ $38.63 each, plus shipping Fed Ex Ground $21.50 or we can use your freight account if you have one. Parts ship from Reno, NV

I attached the CC form you may complete or call me directly with the information.
If you wish to pay ACH/Wire I can send you our bank info.
Thank you
Karen

So I called her up from work and placed a Credit Card order for 2 - with shipping that’s about $100
Mounting app note - recommends silver solder: https://www.ctscorp.com/wp-content/uploads/2015/10/AN1016.pdf
from: https://www.ctscorp.com/resource-center/application-notes/

Board for mounting:


#13

Hey @cswiger, did you have warning like this on the connection of your phone to the enodeB : Failed to find ue context associated with eNB ue s1ap id: 0 without duplexer? About time, how long the phone could connect to your eNB without duplexer?