R1-2601767
LS in
Reply LS on PA models in 6GR waveform evaluations
From Huawei
Summary
This LS from RAN4 to RAN1 communicates RAN4's agreements on PA model usage and waveform evaluation assumptions for 6G low PAPR waveform studies. It includes 11 proposals and 4 observations, covering general evaluation alignment, detailed PA model calibration and usage guidance, and a request for RAN1 to consider these assumptions.
Position
RAN4 agrees to conduct individual assessments on PAPR-reduction techniques for 6G waveform evaluations, focusing on Tx power gain while taking RF requirements into consideration. RAN4 provides specific PA model calibration guidance distinguishing between models that include front-end insertion loss (IL) and those that do not, with output power defined as (power class – 1dB) or (power class – 1dB + x) respectively, both calibrated to meet ACLR of 31dB for PC2 or 30dB for PC3. RAN4 requires separate handling for normalized versus non-normalized PA coefficient models, where normalized models need input normalization by Vmax and output restoration via Vmax • G, while non-normalized models require input clipping at Vmax. RAN4 has not yet reached consensus on PA model(s) for 6G study and indicates ongoing effort on generating appropriate PA models. RAN4 requests RAN1 to incorporate these waveform evaluation assumptions and referenced PA models into its study on low PAPR waveforms.
Key proposals
- Proposal 2 (Sec 1): For waveform evaluations, RAN4 agreed to conduct individual assessments on PAPR-reduction techniques, primarily focusing on Tx power gain by taking RF requirements into consideration.
- Proposal 3 (Sec 1): These evaluations should be based on the agreed waveform evaluation assumptions.
- Proposal 4 (Sec 1): To ensure consistency, the evaluation assumptions should be aligned across relevant working groups.
- Proposal 5 (Sec 1): Table 1 lists the evaluation assumptions for UL low PAPR waveforms.
- Proposal 6 (Sec 1): RAN4 agreed that RAN1's agreements on evaluation assumptions should also be taken into account for RAN4's evaluations.
- Proposal 1 (Sec 2.1): If a model includes the front-end IL of x dB, the PA model's output power is (power class – 1dB) with ACLR meeting 31dB for PC2 or 30dB for PC3.
- Proposal 2 (Sec 2.1): If a model does not include the front-end IL of x dB, x is determined so the PA model's output power is (power class – 1dB + x) with ACLR meeting 31dB for PC2 or 30dB for PC3.
- Proposal 1 (Sec 2.2): For a PA model with normalized coefficients, the input signal should be normalized based on Vmax, and the output signal's amplitude should be amplified according to (Vmax • G) to restore the predicted output amplitude.
- Proposal 2 (Sec 2.2): For a PA model with non-normalized coefficients, the input amplitude range should be clipped within the PA model's maximum input amplitude (Vmax) to get the predicted output.
- Proposal 1 (Sec 2.3): The sampling rate of the input signal is suggested to be the same as the sampling rate of the model via upsampling before putting into the PA.
- Proposal 1 (Sec 2.4): It is suggested that the bandwidth of the input signal is equal to or less than the given bandwidth of the PA model.
- Proposal 1 (Sec 3): RAN4 respectfully requests that RAN1 consider the waveform evaluation assumptions and the PA models in the reference for the following study on low PAPR waveforms.