R1-2601869
discussion
Discussion on aspects of uplink-based CSI acquisition
From ZTE
Summary
This ZTE contribution discusses the design of the 6G RAN (6GR) SRS, providing 4 observations and 18 proposals across 6 sections. The document advocates for a lean, unified, and flexible SRS framework covering resource configuration, coverage/capacity, frequency hopping, DL CSI acquisition, and AI/ML use cases, with a strong emphasis on enabling a single SRS resource for multiple usages and introducing successive frequency hopping.
Position
ZTE proposes a lean and streamlined 6GR SRS design where one SRS resource or transmission can be dynamically reused for multiple usages, including UL CSI acquisition, DL CSI acquisition, and beam management, to avoid the inefficiency of 5G NR's separate configurations per usage. ZTE requires explicit antenna port mapping mechanisms between SRS ports and PUSCH, PUCCH, and DL Rx ports to ensure UL Tx coherency, port selection, and accurate DL CSI acquisition. ZTE argues that the 5G NR nested tree-like frequency hopping pattern is infeasible for wide-band CSI acquisition due to phase discontinuity and channel aging, and proposes introducing a successive frequency hopping pattern as a supplement, supported by BLER and SE simulation gains. ZTE questions the necessity of SRS carrier switching in 6G to avoid redundant features and requires that SRS carrier switching mechanisms not be considered for the non-CA single CC operation of DL CBW 400MHz around 7GHz. ZTE presents simulation results showing AI/ML-based SRS with comb-12 achieves comparable performance to legacy comb-4 SRS but proposes a feasibility study focusing on labeled data construction, model scalability/generalization, and computational complexity.
Key proposals
- Proposal 1-1 (Sec 1): Proposes SRS is essential for both downlink and uplink channel measurements and needs to be a mandatory uplink reference signal adopted in 6GR.
- Proposal 2-1 (Sec 2.1): Proposes striving to enable that one SRS resource or transmission can be applied to multiple usages.
- Proposal 2-3 (Sec 2.2): Proposes supporting explicit mapping of antenna ports between SRS port and PUSCH port for UL Tx coherency, SRS port and PUCCH port for port-selection, and SRS port and DL Rx ports for antenna switching/DL CSI acquisition.
- Proposal 2-4 (Sec 2.3): Proposes supporting dynamic update of SRS bandwidth and consecutive symbols including cross-slot scheme.
- Proposal 2-5 (Sec 2.4): Proposes studying whether and how to facilitate power control related parameters configured in SRS resource specific, where more than one separate closed-loop power control adjustment states for SRS.
- Proposal 3-1 (Sec 3.1): Proposes studying the mechanism of determining group number and base sequence number for SRS sequence generation to ensure both coverage and capacity.
- Proposal 3-3 (Sec 3.2): Proposes supporting adaptive combination of comb offset and cyclic shift to each SRS port in 6GR.
- Proposal 3-4 (Sec 3.3): Proposes supporting the hopping of group/base sequence, comb offset, and cyclic shift for SRS in 6GR.
- Proposal 3-5 (Sec 3.4): Proposes supporting sub-band sounding scheme with repeated SRS transmissions in 6GR.
- Proposal 3-6 (Sec 3.5): Proposes supporting precoded SRS transmission along with UL Tx coherency in 6GR.
- Proposal 4-1 (Sec 4): Proposes introducing successive SRS frequency hopping pattern as one supplement scheme in 6GR.
- Proposal 5-2 (Sec 5.2): Proposes the necessity of SRS carrier switching should be discussed first in 6G day 1 to avoid redundantly specifying the same functionality in multiple features.
- Proposal 5-3 (Sec 5.3): Proposes striving for efficient and unified SRS sounding procedure for DL CBW 400MHz & UL CBW 200MHz around 7GHz, and studying how to mitigate adverse effects in terms of frequency hopping, antenna switching, and RF chain switching.
- Proposal 6-1 (Sec 6.2): Proposes studying the practical needs with consideration of the feasibility investigation of AI/ML based method for SRS transmission.
- Proposal 6-2 (Sec 6.2): Proposes studying evaluation aspects for AI/ML-based UL-based CSI acquisition including KPIs (computational complexity, model size) and scalability/generalization.