R1-2508641
discussion
Views on Modulation for 6GR
From AT&T
Summary
This AT&T contribution presents 6 proposals (Proposals 1-6) and 5 observations (Observations 1-5) on modulation design for 6G Radio (6GR). The document argues for focused study of high-order non-uniform constellations (geometric/probabilistic shaping), prioritization of designs leveraging existing NR modulation architecture, evaluation under SU-MIMO, emphasis on constellation scalability and MCS selection procedures, and deprioritizing AI/ML-based joint source/channel coding/modulation LCM procedures.
Position
AT&T proposes focusing the 6GR non-uniform constellation study on high-order modulation values (up to 4096-QAM for PDSCH and 1024-QAM for PUSCH), where geometric shaping and probabilistic shaping gains are most realized against classical uniform constellations. They require prioritization of proposals that leverage existing NR modulation architecture at both transmit and receive chains, citing block diagram approaches where non-uniform constellations are derived from a selection of uniform constellation points of a larger uniform QAM. They propose that evaluation must include SU-MIMO scenarios with PAPR assessment and optional MU-MIMO under SLS evaluation. They argue that non-uniform constellation designs must demonstrate scalability across different numbers of constellation points and clarify the UE-based MCS selection procedure (CQI reporting) currently made trivial by NR's scalable uniform 4^M QAM. They deprioritize AI/ML-based LCM procedures for joint source coding, channel coding, and modulation, leaving for further study whether such joint coding/modulation should be signaled explicitly without underlying LCM.
Key proposals
- Proposal 1 (Sec 2, Observation 1): The 6GR study on performance/complexity tradeoff of non-uniform constellations based on geometric shaping and/or probabilistic shaping is focused on high-order modulation values.
- Proposal 2 (Sec 2, Observation 2): For 6GR proposals on non-uniform constellation and probabilistic shaping, prioritize designs that leverage the existing NR modulation architecture at both the transmit and receive chains.
- Proposal 3 (Sec 2, Observation 3): For 6GR proposals on non-uniform constellation and probabilistic shaping, evaluation is needed for scenarios including SU-MIMO.
- Proposal 4 (Sec 2, Observation 4): For 6GR proposals on non-uniform constellation, further emphasis is needed on the scalability of the constellation to include different values of constellation points, in addition to the underlying MCS selection procedure needed at the receiver side.
- Proposal 5 (Sec 2, Observation 5): LCM procedures for AI/ML-based joint source coding, channel coding and modulation is deprioritized in the 6GR study.
- Proposal 6 (Sec 2, Observation 5): FFS: whether the joint source coding, channel coding and/or modulation is signaled/specified explicitly without an underlying LCM procedure.
- Proposal 1 (Sec 3): The 6GR study on performance/complexity tradeoff of non-uniform constellations based on geometric shaping and/or probabilistic shaping is focused on high-order modulation values.
- Proposal 2 (Sec 3): For 6GR proposals on non-uniform constellation and probabilistic shaping, prioritize designs that leverage the existing NR modulation architecture at both the transmit and receive chains.
- Proposal 3 (Sec 3): For 6GR proposals on non-uniform constellation and probabilistic shaping, evaluation is needed for scenarios including SU-MIMO.
- Proposal 4 (Sec 3): For 6GR proposals on non-uniform constellation, further emphasis is needed on the scalability of the constellation to include different values of constellation points, in addition to the underlying MCS selection procedure needed at the receiver side.
- Proposal 5 (Sec 3): LCM procedures for AI/ML-based joint source coding, channel coding and modulation is deprioritized in the 6GR study.
- Proposal 6 (Sec 3): FFS: whether the joint source coding, channel coding and/or modulation is signaled/specified explicitly without an underlying LCM procedure.