AODT 1.5 Material Calibration question

Accelerated Computing Aerial Forum
1

Hi Nvdia,

We are evaluating the official AODT 1.5 calibration workflow.

For calibration, we followed the NVIDIA documentation:

Call “client.run_calibration()”:

* 100 position-aligned GPX and RSRP samples
* Full ray-path export
* Reflection, transmission, diffuse scattering, and diffraction enabled
* Only cal.targets.Materials: true
* RU power, position, orientation, and UE settings fixed

Result:

GT average RSRP:             -96.17 dBm
Original AODT average CFR:   -89.21 dBm
Calibrated AODT average CFR: -89.06 dBm

The calibrated result follows nearly the same curve as the original result, and the approximately 7 dB absolute offset remains. Throughput accuracy also did not improve.

  1. Is the centered calibration loss expected to correct only the route-dependent signal shape, not the absolute RSRP level?
  2. Is there an official method for calibrating the global power offset without modifying RU transmit power?
  3. What is the recommended way to compare smartphone-reported NR RSRP with AODT CFR or post-calibration predicted RSRP?(and throughput)

*Below plot show Calibration NOT good

image
image1367×618 121 KB

Thanks
BR
Bryan Chen

2

Hi @bryan_chen I was looking at your data, and looks like your field measurements are noisy. We would need additional information on your data collection methodology. I will send you a DM.

Thanks

Kalpendu