[srsran-users] Channel state information in TDD

[Mehdi]

Dear Matthias,

Thanks for your kind reply!

Then, the question is whether the OAI/SRS developers considered any
calibration algorithm/code in their open sources? This problem of channel
mismatch can cause further problems in the multi-antenna case, where the
transmitter needs to rely on the UL channel estimations to deploy DL
beamforming.

So I am wondering whether the OAI/SRS developers have considered any
automatic calibration, etc. to be able to obtain similar wireless channel
values??

Best, -Mehdi

On Mon, Dec 6, 2021 at 1:44 PM Mehlhose, Matthias <
matthias.mehlhose at hhi.fraunhofer.de> wrote:

> Hi Mehdi,
>
>
> nice question.
>
>
> Please take into account, that your channel is not only the wireless one
> from one antenna to another.
>
>
> It's also all RF components, like cable, plugs, filters, splitters,
> switches, circulators, and amplifiers or attenuators.
>
> -> All the things between your antennas and the RF Transceiver chip
> ADC/DAC.
>
>
> In most cases, it will not be the same path! And therefore, you will not
> see the same channel, because each RF component will add an individual
> response.
>
>
> I hope you find a way to calibrate and measure the hopefully static
> differences on TX and RX sides, for your project.
>
>
> BR Matthias
> ------------------------------
> *Von:* srsran-users <srsran-users-bounces at lists.srsran.com> im Auftrag
> von Mehdi <studentatsharif at gmail.com>
> *Gesendet:* Sonntag, 5. Dezember 2021 16:55:53
> *An:* srsran-users at lists.srsran.com
> *Betreff:* [srsran-users] Channel state information in TDD
>
>
> Dear all,
>
> I am working on TDD-LTE, using SRS-UE and OAI-eNB. For my research
> project, I require the channel state information values (CSIs) which are
> estimated at eNB and UE. But I am observing an abnormal behavior that the
> channel estimations obtained at the two sides are very different! (I am
> working on TDD config. 1, while I select the estimated values with the same
> frame number and adjacent subframes 8 and 9 for UL/DL)  Specifically, the
> channel estimations at eNB are very different from the values at UE. That
> is, the eNB shows much bigger absolute values than that of the UE. This
> seems to be in contradiction with the main principles of UL-DL reciprocity
> in a static environment. The values which are reported from OAI-eNB do not
> even follow the same trend with the channel estimation values reported by
> SRS-UE!!!
>
> FYI, I am examining ul_ch_estimates as the channel estimations at eNB
> <https://gitlab.eurecom.fr/oai/openairinterface5g/-/blob/v1.0.3/openair1/PHY/LTE_ESTIMATION/lte_ul_channel_estimation.c> side
> and ue_dl.chest_res.ce as the channel estimations at UE
> <https://github.com/srsran/srsRAN/blob/master/lib/src/phy/ch_estimation/chest_dl.c>.
> The corresponding data I get, together with some initial plots of the
> values are attached to this email.
>
>
> This is very critical for my project to find possible reasons for such
> surprising behavior. So, I appreciate the community letting me know if
> anyone has had any similar experience on inspecting the channel estimations.
>
> Thanks,
>
> Mehdi
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.srsran.com/pipermail/srsran-users/attachments/20211207/3baaee82/attachment-0001.htm>