Bug #1616
closedosmo-bts-trx / osmo-bts-octphy doesn't provide C/I information to PCU
100%
Description
osmo-bts-trx doesn't pass singal quality information to the PCU socket, meaning that link/rate adaption will of course not work as it i supposed to.
Related issues
Updated by msuraev almost 8 years ago
How exactly quality information is passed from DSP to L1 in case of osmo-bts-sysmo? I have trouble locating that place in the code - there doesn't seem to be particular function dedicated to that.
Updated by msuraev almost 8 years ago
According to void GprsMs::update_cs_ul(const pcu_l1_meas *meas) in src/gprs_ms.cpp the switch to higher/lower CS depends only on link_qual value. Not sure if some spec covers which measurement is supposed to be used for link rate adaptation. Also, it's unclear yet how to convert the values available in particular BTS to be similar to what is reported by sysmobts.
Updated by msuraev almost 8 years ago
- Status changed from New to In Progress
Note: the DL CS is selected separately in update_error_rate()
Updated by laforge almost 8 years ago
- Related to Feature #1543: link/rate adaption as per spec added
Updated by laforge almost 8 years ago
- Related to Feature #1536: Implement adaptive CS selection added
Updated by laforge almost 8 years ago
TS 45.008 Annex D contains an "Example Selection of Modulation and Coding Schemes" for 8PSK EGPRS (MCS5-MCS9) based on CV_BEP and MEAN_BEP. Those two values should be reported by a BTS model to the PCU.
For classic GPRS or EGPRS with GMSK, I'm still looking to find similar reference values.
Chapter 8.2 of TS 45.008 contains information on how RxQual and CV_BEP / MEAN_BEP values are computed and encoded.
Updated by laforge almost 8 years ago
Some papers:
- https://www.netlab.tkk.fi/opetus/s38310/01-02/li_280502.pdf
- http://eeweb.poly.edu/dgoodman/icc04gprs.pdf
- http://queseth.se/olav/pubs/AlgorithmsForLinkAdaptationInGPRS.pdf
- http://www.cin.ufpe.br/~nomadic/vtc2000/A18522.PDF
- http://investigacion.ac.upc.edu/conferencies/EW2004/papers/156.pdf
In genreal, it seems there are BLER and CIR (C/I) based schemes, and that C/I is generally perceived to be superior.
I think the parameters RxLev/Rssi, BER, RxQual (averaged BER) should be provided by ever PHY/L1 and thus be a good candidate for reporting measurements to the PCU for GPRS/GMSK. Plus CV_BEP and CV_MEAN for EGPRS/8PSK
Updated by msuraev almost 8 years ago
Version for octphy submitted for review in gerrit # 622 - 624.
Updated by msuraev almost 8 years ago
- % Done changed from 0 to 50
Changes for l1sap and bts-pcu protocols have been merged to master. Octphy implemntation have been merged, osmo-trx is partially implemented (C/I estimation is missing, extensive testing required).
Updated by laforge about 7 years ago
- Subject changed from osmo-bts-trx / osmo-bts-octphy doesn't provide signal quality information to PCU to osmo-bts-trx / osmo-bts-octphy doesn't provide C/I information to PCU
- Priority changed from Normal to Low
Updated by laforge almost 6 years ago
- Related to Bug #3395: Uplink CS/MCS control is broken osmo-pcu is used with osmo-bts-trx/osmo-trx added
Updated by laforge almost 6 years ago
- Related to Feature #1855: provide actual BER or C/I values from osmo-bts-trx into the PCU added
Updated by fixeria almost 5 years ago
- Category changed from osmo-bts-octphy to osmo-bts-trx
- Status changed from Stalled to Resolved
- Assignee set to fixeria
- % Done changed from 50 to 100
Please see https://gerrit.osmocom.org/r/Ia58043bd2381a4d34d604522e02899ae64ee0d26, which is now merged.