R1-2601803
discussion
Discussion on channel coding for 6GR
From Spreadtrum
Spreadtrum's prior position on
10.3.1
at
RAN1#124
· AI-synthesized, paraphrased
verify sources →
Proposes that LDPC extension for 6G should maintain high consistency with existing NR LDPC, with only minor software or hardware modifications, and argues implementation-based solutions should receive priority consideration if they can meet data rate requirements. Specifically proposes prioritizing reduction of the maximum number of iterations without significantly affecting performance for data channel coding. For control channel coding, proposes directly reusing NR Polar code design for control information within NR range, and if DCI payload size must exceed the 164-bit limit, proposes removing the D-CRC interleaver and reusing the existing 1024 sequence for UCI to extend the DCI payload upper bound.
Summary
Spreadtrum/UNISOC presents 3 proposals and 2 observations on 6G channel coding, covering data channel LDPC extension and control channel Polar code design for payloads beyond NR range. The document argues for preserving NR LDPC parameters (maximum code block size of 8448 and 22 information columns) and prioritizing code block segmentation plus interleaver removal for larger DCI payloads.
Position
Spreadtrum/UNISOC proposes that the LDPC extension beyond NR range retain the maximum code block size of 8448 and 22 information columns in the base graph, arguing that the 22×384 configuration fully preserves the NR LDPC structure and enables maximum reuse of existing implementations without altering base graph design, decoding schedule, or memory access patterns. They present a technical case against increasing the number of information columns to 33 or 44, citing increased scheduling complexity, more fragmented memory access, and reduced parallel processing efficiency, especially when combined with a reduced lifting size such as 192 or 256. For Polar code design for DCI with payload size larger than NR range, they propose prioritizing code block segmentation (Option 1) and interleaver removal (Option 2) as the options offering the highest degree of compatibility with the existing design architecture, while seven other candidate options including Mid-Block Termination-Assisted Polar Codes, zero-embedding, PAC codes with zero-padding, and new data integrity check mechanisms are acknowledged but not preferred.
Key proposals
- Proposal 1 (Sec 2, 4): For the study of LDPC extension beyond NR range, the maximum code block size should remain at 8448.
- Proposal 2 (Sec 2, 4): For the study of LDPC extension beyond NR range, the number of information columns in BG should remain at 22.
- Proposal 3 (Sec 3, 4): For the study of Polar code design for DCI with payload size larger than NR range, code block segmentation and interleaver removal should be the priority options.
- Observation 1 (Sec 2): Since the maximum lifting size cannot exceed 384, and there exists a mathematical relationship among the maximum code block size, the number of information columns in BG, and the maximum lifting size, three cases need to be considered: 8448 = 22 × 384, 8448 = 33 × 256, 8448 = 44 × 192, and 16896 = 44 × 384.
- Observation 2 (Sec 3): The largest supported DCI payload size is 164 bits for NR Polar code, corresponding to 140 (DCI payload) + 24 (CRC), and there is currently great controversy over whether 6G maximum DCI payload size will exceed 140 bits.