R1-2601836
discussion
On downlink transmission schemes for downlink shared channels in 6GR
From Nokia
Summary
This Nokia contribution to 3GPP RAN1 #124-bis (FS_6G_Radio, Rel-20) provides a comprehensive technical stance on 6G downlink shared channels, covering PDSCH transmission schemes, resource allocation, DM-RS design, and AI/ML use cases. The document contains 21 Observations and 39 Proposals.
Position
Nokia proposes maintaining 6GR PDSCH resource allocation compatibility with 5G NR to facilitate MRSS, supporting slot-contained PDSCH TDRA with flexible S and L, and deferring DM-RS mapping type decisions pending 6G DM-RS design clarity. Nokia proposes raising the maximum number of orthogonal DMRS ports to 48, selecting a single DM-RS type with a scalable comb pattern, using Rel-18 e-type 1 DMRS (UL) as baseline, and specifically supporting Comb-2 DMRS RE pattern based on presented LLS results showing its robustness in frequency-selective channels. Nokia requires revisiting codeword-to-layer mapping to enable a second codeword from rank 2 and requires revisiting MCS table design to decouple higher modulation orders from higher code rates as layer imbalance mitigation. For AI/ML, Nokia proposes studying sparse DMRS and DMRS-free transmissions within a unified 6G DMRS framework, proposes using end-to-end training with known transmitted sequences for all AI receiver options, and proposes specific modular splits (non-AI channel estimation/equalization with AI-based demodulation) for receiver complexity reduction.
Key proposals
- Proposal 1 (Sec 2.1): Study the performance benefits achievable by increasing the maximum number of simultaneous transmission layers beyond what can be practically supported in 5G NR. If such performance benefits are justified, then study how best to increase the maximum number of simultaneous transmission layers (e.g, by increasing the maximum number of orthogonal DMRS ports to 48).
- Proposal 4 (Sec 2.2.1): 6GR to support PDSCH time-domain resource allocations within a slot (i.e. S+L≤14) having a flexible starting symbol S and an allocation length L. Valid values of S and L to be further discussed. Defer the discussions on the "PDSCH mapping type(s)" related to DM-RS placement within the PDSCH allocation after having more clarity on the 6G PDSCH DM-RS design.
- Proposal 7 (Sec 2.2.2): 6GR to support the following PDSCH TDRA operation from NR: PDSCH transmission (occasions) to be contained within a DL slot (FFS: PDSCH transmission crossing slot boundary), PDSCH repetition operation consisting of multiple PDSCH TX occasions in consecutive DL slots with the same TDRA / SLIV within each slot, "Multi-PDSCH" resource allocation of independent PDSCHs with independent TDRAs / SLIVs (each contained within a DL slot) of different HARQ processes, The PDSCH TDRA within a slot is indicated through a field (such as TDRA index) in the scheduling DCI.
- Proposal 9 (Sec 2.2.3): 6GR to support RBG-based ("Type 0") and RIV-based ("Type 1") PDSCH frequency domain resource allocation applying the related NR principles. The PDSCH scheduling using the baseline/fallback DCI format could be limited to RIV-based ("Type 1") PDSCH FDRA.
- Proposal 14 (Sec 2.3.1): Support up to 48 orthogonal DMRS antenna ports.
- Proposal 18 (Sec 2.3.2): Study multi-slot based PDSCH DM-RS design in 6GR.
- Proposal 20 (Sec 2.3.3): Reuse pseudo-random sequences in 6GR DMRS design.
- Proposal 25 (Sec 2.3.4): For 6GR, study how to simplify the PT-RS configuration options of NR targeting to the practical scenarios including FR2 and applicability to the other spectrum bands around 7GHz and 15GHz.
- Proposal 27 (Sec 2.4.2): In 6G, codeword-to-layer mapping schemes enabling a second codeword from rank 2 (for up to 4 codewords) is an attractive solution and should be further studied for downlink considering rank-adaptive operation and different receiver capabilities (e.g., MMSE-IRC and SIC), with the goal of mitigating layer imbalance and improving MIMO performance.
- Proposal 29 (Sec 2.4.2): Revisit MCS table design in 6G to consider layer imbalance in downlink.
- Proposal 31 (Sec 3.2): To ensure fairness in the evaluation/comparison between AI receivers and legacy receivers, the study on DMRS overhead reduction in RAN1 must consider realistic assumptions in both. For instance, both receivers should have access to the same information (e.g. DMRS resources and noise variance).
- Proposal 34 (Sec 3.3): For AI receivers, RAN1 to study sparse DMRS configurations for 6G that reduce DMRS overhead while maintaining compatibility with the unified 6G DMRS framework (single DMRS type, scalable RE mapping, and support for large port counts) and preserving flexibility to address the specific needs and constraints of channel estimation using sparse DMRS.
- Proposal 37 (Sec 3.4): RAN1 to further study DMRS-free transmissions at least for the case of multi-slot PDSCH transmission, where at least some slots can be DMRS free.
- Proposal 38 (Sec 3.5): RAN1 to consider all AI receiver options for further studies on sparse DMRS based PDSCH transmission considering in all options an end-to-end training using known transmitted sequences with the following updates: Updated Option 4: non-AI based {Channel estimation + equalization} + AI-based {demodulation/LLR calculation}; Updated Option 5: AI-based {channel estimation} + non-AI based equalization + AI-based {demodulation/LLR calculation}.