[srsran-users] 5G NR "Phase compensation"

[Xavier Arteaga]

Hi Sylvain,
I am delighted to answer your question the best I can.

> 1) What is it for ? What's the purpose of that "special" up conversion
> instead of the way it's done in 4G for instance ?

The phase compensation allows the transmission and reception of the OFDM
waveform without centering the resource grid on a carrier DC. For example,
this enables the UE to detect and decode the SS/PBCH without centering the
baseband in the same gNb center frequency.

If the phase compensation was not in place, the UE would see a different
phase in each OFDM symbol. The phase shift is relative to the difference of
center frequencies between the gNb and itself and it would be difficult to
detect.

> 2) My understanding of where this comes from ( from the eq in 5.4 to
> the implementation ) is that :
> ...
> Is my understanding correct ?

Yes, it is correct.

To conclude, as TS38.211 Section 5.4 says, the phase compensation is
applied relative to the beginning of the subframe. This makes each of the
OFDM symbol phases within a subframe coherent for transmission and
reception, independently of the resource grid center frequencies.

This allows the UE to select the most suitable center frequency for the
bandwidth of interest without having knowledge of the gNb center frequency.

I hope that my message enlightens you. If you or anyone else has any
further doubts or queries, please do not hesitate to ask for clarification.

Kind regards,
Xavier



On Wed, 1 Mar 2023 at 22:23, Sylvain Munaut <246tnt at gmail.com> wrote:

> Hi,
>
> I'm trying to understand what the 5G "phase compensation" thing is and
> what it's for.
>
> I see in the implementation (
> lib/phy/lower/modulation/phase_compensation_lut.h ) the reference to
> TS38.211 Section 5.4.
> But I must admit I don't find that section to be super explicit, so I
> have a couple of questions :
>
> 1) What is it for ? What's the purpose of that "special" up conversion
> instead of the way it's done in 4G for instance ?
>
> 2) My understanding of where this comes from ( from the eq in 5.4 to
> the implementation ) is that :
>    - In the formula the phase of the exponential used to up-convert
> "resets" at every symbol
>    - But of course in the downconversion done in hardware by the
> mixer, that's not the case, phase would be continuous, so that's where
> the phase shift at every symbol comes from.
>
> Is my understanding correct ?
> And if it is, something I still don't get is that the same kind of
> discontinuity would also happen between sub-frames ( i.e going from
> symbol N_sym-1 of subframe 0 to symbol 0 of subframe 1 for instance ).
> But I don't see that being accounted for, AFAICT it just applies a
> static LUT to each symbol in a subframe.
>
> Thanks for any light anyone could shine on the matter.
>
> Cheers,
>
>      Sylvain
> _______________________________________________
> srsran-users mailing list
> srsran-users at lists.srsran.com
> https://lists.srsran.com/mailman/listinfo/srsran-users
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.srsran.com/pipermail/srsran-users/attachments/20230302/def27fd6/attachment.htm>