libosmocore[master]: gsm0503: migrate transcoding routines from OsmoBTS

This is merely a historical archive of years 2008-2021, before the migration to mailman3.

A maintained and still updated list archive can be found at https://lists.osmocom.org/hyperkitty/list/gerrit-log@lists.osmocom.org/.

Vadim Yanitskiy gerrit-no-reply at lists.osmocom.org
Wed Sep 7 13:07:32 UTC 2016


Patch Set 4:

First of all, thank you for detailed review!

 > > In OsmoBTS there are already generated convolutional codes.
 > > So, I spent some time finding corresponding polynomial sets for
 > > some data types, which wasn't defined in the conv_gen.py
 > generator.
 > 
 > Sounds like excellent work! Did you find them mathematically, or in
 > specs out there? :)

This might actually sound strange, but I am still have not clear
understanding, how does the polynomial generator work in details.
Some polynomial sets I have been able to find exactly in the GSM
05.03 specification, some others (mostly TCH/AHS*) aren't covered
there. After several unsuccessful attempts to find something useful
in Google, I found, that some TCH/AHS codes match with the TCH/AFS
codes with different length and punctures.

 > > Despite the EDGE's MCS levels use the same generator polynomial
 > > with different lengths, I wasn't able to find any details about
 > > EDGE convolutional transcoding, so I merely copied those tables
 > > from OsmoBTS into a separate file.
 > 
 > Ah, so there's two parts here. The commit log sounded at first like
 > there was one set of codes. Sounds like it makes sense to split
 > in two commits to make things perfectly clear.

Yeah, I really need to separate this commit... Maybe this way?

 - utils/conv_gen.py: code style changes (line width, tabs, etc.)
 - utils/conv_gen.py: improve output formatting (4 blocks / 12 numbers per line)
 - utils/conv_gen.py: generate a single file

 - gsm/gsm0503: migrate transcoding routines from OsmoBTS (without EDGE for now)
 - ??? EDGE specification ???: add EDGE/MCS transcoding routines
   As Sylvain said, the MCS convolutional codes may be generated using
   the conv_gen.py. So, as soon as I find corresponding polynomials, I'll do it.

 > > I still have some doubts, related to MCS codes...
 > > They aren't covered by GSM 05.03, so should I migrate them
 > > in separate commit? Should I use a separate file? Which name to
 > > use for this file?
 > 
 > Yes and yes -- feel free to invent a new file name. A GSM spec
 > number would be best to match the other file names -- but you say
 > the EDGE codes are not covered by a specific GSM spec? As a
 > fallback I'd suggest something like 'gsm_edge_coding.c' or
 > 'gsm_mcs_coding.c'? I assume you have a better understanding of
 > what name would be proper.

They aren't covered by GSM 05.03 :(
Tom Tsou probably knows the answer, but he is a bit busy right now.

-- 
To view, visit https://gerrit.osmocom.org/816
To unsubscribe, visit https://gerrit.osmocom.org/settings

Gerrit-MessageType: comment
Gerrit-Change-Id: I257a5d015798ee9e690fd035ca97fd971cf9f60a
Gerrit-PatchSet: 4
Gerrit-Project: libosmocore
Gerrit-Branch: master
Gerrit-Owner: Vadim Yanitskiy <axilirator at gmail.com>
Gerrit-Reviewer: Jenkins Builder
Gerrit-Reviewer: Neels Hofmeyr <nhofmeyr at sysmocom.de>
Gerrit-Reviewer: Vadim Yanitskiy <axilirator at gmail.com>
Gerrit-HasComments: No



More information about the gerrit-log mailing list