Project

General

Profile

Bug #1855

provide actual BER or C/I values from osmo-bts-trx into the PCU

Added by laforge over 2 years ago. Updated 4 days ago.

Status:
New
Priority:
High
Assignee:
Category:
osmo-bts-trx
Target version:
Start date:
11/18/2016
Due date:
% Done:

0%

Spec Reference:

Description

see fo rexample 4b76a323b3bb71f8d3f4dc7439ecd9bad0f13bcf which introduces various FIXMEs explaining what's missing.


Related issues

Related to OsmoPCU - Bug #3395: Uplink CS/MCS control is broken osmo-pcu is used with osmo-bts-trx/osmo-trxStalled2018-07-14

Related to OsmoBTS - Bug #1616: osmo-bts-trx / osmo-bts-octphy doesn't provide C/I information to PCU Stalled2016-02-23

Blocked by Cellular Network Infrastructure - Feature #4006: TRX protocol: wind of changeNew2019-05-17

History

#1 Updated by laforge 10 months ago

  • Related to Bug #3395: Uplink CS/MCS control is broken osmo-pcu is used with osmo-bts-trx/osmo-trx added

#2 Updated by laforge 9 months ago

  • Related to Bug #1616: osmo-bts-trx / osmo-bts-octphy doesn't provide C/I information to PCU added

#3 Updated by laforge 9 months ago

#4 Updated by laforge 8 months ago

  • Priority changed from Normal to High

#5 Updated by laforge 7 months ago

  • Assignee set to tnt

#6 Updated by laforge 25 days ago

tnt, any news here?

I just re-investigated this topic with Lynxis. The C/I value can be computed from the training sequence of each burst, where we can compare the "ideal" training sequence with the actual training sequence and then express that in dB.

As osmo-trx is already doing the correlation againt the training sequence, it may make sense to do it there. In fact, detectBurst() appears to already return the complex normalized amplitude of the correlation of the training sequence, which is probably almost exactly the value we're interested in here? If this is true, we should probably extend the trx protocol to provide the data to osmo-bts-trx, rather than having osmo-bts-trx do another correlation against the training sequence...

#7 Updated by tnt 6 days ago

I got the math worked out that computes a value that might be the C/I ... (at least it seems to have the right properties and it's also in the right range of value).

http://git.osmocom.org/osmo-trx/commit/?h=tnt/ci&id=d40f11962a4d0e2c0ea9e3bde4c568c6c4b62a12

How/Where to actually pass it all the way to the PCU is TBD.

#8 Updated by ipse 6 days ago

tnt wrote:

I got the math worked out that computes a value that might be the C/I ... (at least it seems to have the right properties and it's also in the right range of value).

http://git.osmocom.org/osmo-trx/commit/?h=tnt/ci&id=d40f11962a4d0e2c0ea9e3bde4c568c6c4b62a12

How/Where to actually pass it all the way to the PCU is TBD.

Great to finally see some progress here :)

By "right properties" - have you tested increasing interference to check that it's correctly accounted?

Do you have test vectors which could be used as a test case when/if someone decides to write an optimized version of the code?

#9 Updated by tnt 6 days ago

I checked that :
- It's independent of scaling (so if you multiply the input vector by any value, it doesn't change the result, except for 'float' precision of course).
- Adding noise tends to decrease the value. I say "tends" because it's all based on only 16 samples ... which doesn't provide much averaging at all, so I ran it through thousands of runs, adding noise to some off the air captures and just looked at the shape of the graph.

#10 Updated by tnt 4 days ago

I think it was pespin and fixeria were discussing how to extend the TRX protocol ?
Just ping me when you came up with that extended protocol and where to plug that value ...

#11 Updated by fixeria 4 days ago

Also available in: Atom PDF

Add picture from clipboard (Maximum size: 48.8 MB)