R1-2600194 discussion

Discussion on downlink based CSI acquisition

From OPPO
Status: not treated
WI: FS_6G_Radio
Agenda: 10.5.3.1
Release: Rel-20
Source: 3gpp.org ↗

Summary

OPPO's Tdoc R1-2600194 provides 17 observations and 37 proposals on the 6G CSI framework, recommending the NR CSI framework as the baseline with simplifications. Key proposals include supporting up to 256 CSI-RS ports, studying AI/ML-based CSI-RS overhead reduction, adopting a unified fixed codebook with low/high-resolution types, studying JSCC/JSCM for CSI feedback, and addressing specific scenarios like mTRP and NES.

Position

OPPO proposes using the NR CSI framework as a starting point for 6G, requiring simplification of CSI-ReportConfig and CPU/timeline calculations to avoid over-design. They propose supporting up to 256 CSI-RS ports, asserting that 512 ports provide no additional performance gain, and recommend large CSI-RS periodicity (e.g., 20ms) for realistic evaluation. For codebook design, they propose a unified fixed codebook baseline with low-resolution (Type I scheme A as starting point) and high-resolution (eType II with layer-common or layer-specific SD basis) types, while also studying downloadable codebooks as a complement. They present technical evaluations showing JSCC and JSCM with two-side models alleviate cliff effects in CSI feedback, and they propose studying fusion of downlink CSI feedback with SRS measurements for TDD systems.

Key proposals

Your notes

Private to your account