R1-2509146
discussion
Modulation for 6GR interface
From MediaTek
Summary
This document presents 6 Observations and 2 Proposals on modulation enhancements for 6G. It covers low-PAPR modulation (O-QPSK for DFT-s-OFDM), the infeasibility of 4K-QAM DL and 1024-QAM UL, and high-dimensional coded modulation with a focus on Enumerative Sphere Shaping (ESS) for Probabilistic Constellation Shaping (PAS).
Position
MediaTek proposes supporting O-QPSK for DFT-s-OFDM to enhance coverage by reducing PAPR by approximately 4dB compared to π/2-BPSK. They present a technical case, based on field SNR logs and link-level simulations with practical EVM values from TS 38.104, that 4K-QAM is not feasible for DL and 1024-QAM is not feasible for UL. They propose RAN1 support probabilistic constellation shaping (PCS) for higher-order modulation, showing 1–2 dB T-put gain with dynamic rank and MCS adaptation in a MIMO fading channel. They argue that Enumerative Sphere Shaping (ESS) is superior to CCDM as a distribution matcher for PAS due to its smaller rate loss at small-to-medium block lengths and lower memory requirements when using bounded-precision implementation.
Key proposals
- Proposal 1 (Sec Low PAPR modulation design for DFT-s-OFDM): For DFT-s-OFDM, support O-QPSK for coverage enhancement.
- Proposal 2 (Sec T-put performance evaluation for PCS in MIMO fading channel): RAN1 to support probabilistic constellation shaping for higher-order modulation.
- Observation 1 (Sec 4096QAM for DL and 1024-QAM for UL): Based on the possible SNR in the field and the LLS evaluations for the current modulation schemes, 4K-QAM is not feasible for DL.
- Observation 2 (Sec 4096QAM for DL and 1024-QAM for UL): Due to lower SNR (compared to DL SNR) in UL and the limitation at the UE to achieve low EVM, it is expected that 1024QAM is not feasible in UL as well.
- Observation 3 (Sec Rate loss evaluation): ESS has smaller rate loss than CCDM which make it more suitable for small to medium block lengths.
- Observation 4 (Sec Performance Evaluation): ESS is a good candidate for PAS due to its good performance and small rate loss for any shaper length.
- Observation 5 (Sec Performance Evaluation): ESS storage can be further reduced, subject to acceptable performance loss.
- Observation 6 (Sec T-put performance evaluation for PCS in MIMO fading channel): 1–2 dB gain can be observed in T-put performance with dynamic rank and MCS adaptation.
- Observation 4 (Sec Conclusion): ESS is a good candidate for PAS due to its good performance and small rate loss for any shaper length.
- Observation 5 (Sec Conclusion): ESS storage can be further reduced, subject to acceptable performance loss.