Cellular Network Infrastructure » Radio Access Network » OsmoBSC

Added http://git.osmocom.org/osmo-bsc/commit/?h=keith/uplink_power_loop2

A few bits after reading your email from the ML:

• It makes sense to implement the MS Power Control Loop in the BSC too, for the case you describe: BTS not supporting (or doing a bad work with) the MS Power Control Loop. For that, you need to indeed update configuration to add a "dyn-bsc" option, and avoid sending the IE "RSL_IE_MS_POWER_PARAM" to the BTS in that case. In that mode, BSC runs the MS Power Control Loop algo based on Measurement Reports received from BTS and sends an MS POWER CONTROL message (see rsl_chan_ms_power_ctrl()) to the BTS each time it wants it to change the requested MS Power level for the MS.

• It makes no sense to share code between BSC and BTS regarding the MS Power Control Loop. The loop algorithm code itself is quite small actually. Furthermore, the BTS may have more/different information available than the BSC: for instance, the BTS can use C/I values measured internally, but those are never sent to the BSC, only RxQual, which is similar ut has less granularity.

• Make sure you base your BSC patches on current osmo-bts master, I have been adding fixes/improvements in related code over last weeks/days.

pespin wrote:

A few bits after reading your email from the ML:

Thanks pespin
As can happen, slightly conflicting advice here and a reply on the ML from laforge

For that, you need to indeed update configuration to add a "dyn-bsc" option,

Yes, the other nice option is making it transparent, really the only model I know of for now is the rbs2000, so I could simply make the BSC power loop run for that model and not bother with dyn-bsc. That is to say, one would use dyn-bts, thereby leaving the power control parameters available in the vty. Or I could implement dyn-bsc.

and avoid sending the IE "RSL_IE_MS_POWER_PARAM" to the BTS in that case.

of course.

In that mode, BSC runs the MS Power Control Loop algo based on Measurement Reports received from BTS and sends an MS POWER >CONTROL message (see rsl_chan_ms_power_ctrl()) to the BTS each time it wants it to change the requested MS Power level for >the MS.

Yes, that's what the above linked patch does.

• It makes no sense to share code between BSC and BTS regarding the MS Power Control Loop.

OK. I'll take that as definitive, as probably the amount of work to make the code compatible is more than to keep 2 copies updated with any work on the loop.

The loop algorithm code itself is quite small actually. Furthermore, the BTS may have more/different information available than the BSC: for instance, the BTS can use C/I values measured internally, but those are never sent to the BSC, only RxQual, which is similar ut has less granularity.

Yes. In the patch above, I removed references to the C/I. I considered making use of RxQual, but I think it's OK just to use level for now.

• Make sure you base your BSC patches on current osmo-bts master, I have been adding fixes/improvements in related code over last weeks/days.

Of course, as always.. I saw your patches.

https://gerrit.osmocom.org/c/osmo-bsc/+/25654 was merged.

TODO: The power loop implemented in the merged patch could probably also make use of RxQual.