R1-2509387
discussion
FLS#3 on evaluation assumptions for 6GR air interface
From Huawei
Summary
This document, moderated by Huawei, presents 3GPP RAN1 discussion and decision proposals for 6G Radio (FS_6G_Radio) evaluation assumptions. It covers approximately 30 proposals and 18 observations across antenna modeling for TN and NTN, system-level simulation (SLS) parameters, new traffic models (AI/ML, immersive communications, extended FTP, IoT), and link budget updates.
Position
Huawei (as moderator) proposes specific parameter updates for BS antenna modelling at 700MHz, 2GHz, and 7GHz, and presents proposals for 15GHz BS/UE antenna models. They propose a generic AI/ML traffic model that explicitly preserves the concept of a 'Token' as the minimum application-layer data unit, defines packet size as an integer multiple of Token size, and requires a 'Token success rate' metric distinct from packet success rate to directly reflect AI/ML service quality. They propose extending the XR traffic model (TR 38.838) for immersive communications by adding values for immersive gaming data rate/frame generation rate and UL-heavy video uploading packet characteristics, and defining a haptics model with co-generated packets using specific pseudo-code. They propose extending FTP Model 3 with PDB and multiple packet sizes, preferring Option 1 where multiple independent traffic flows are modelled per UE.
Key proposals
- Proposal 2.1.2-1 (Sec 2.1): Updates BS antenna modelling for 700MHz, 2GHz, and 7GHz carrier frequencies by specifying (M,N,P,Mg,Ng; Mp,Np) tuples.
- Proposal 2.1.2-2rv3 (Sec 2.1): Proposes BS antenna modelling configurations for around 15GHz carrier frequency, including indoor/outdoor combinations and TXRU mapping notes.
- Proposal 2.2.2-rv4 (Sec 2.2): Models UE antenna for 6GR evaluation with specific combinations (1T2R up to 16T16R) using Alt 1 (UPA array parameters) or Alt 2 (device antenna model from TR38.901).
- Proposal 3.1.2-rv1 (Sec 3.1): Assumes layout for system-level simulation including single and two-layer hex grids with random micro drops for various deployment scenarios.
- Proposal 3.2.2-1rv2 (Sec 3.2): Assumes total transmit power per BS for system-level simulation across frequency ranges and deployment scenarios, with notes on maximum BS Tx power and EIRP limits.
- Proposal 3.2.2-2rv1 (Sec 3.2): Assumes UE power class for system-level simulation, including 29dBm (not for smartphone) and EIRP limitations.
- Proposal 3.3.2-rv2 (Sec 3.3): Assumes UE distribution and UE speed for system-level simulation, including indoor/outdoor ratios, UE speeds, and number of UEs per TRxP.
- Proposal 3.4.1-rv1 (Sec 3.4): Assumes other SLS parameters such as ISD, BS/UE antenna height, noise figure, scheduling, interference modelling, and channel estimation.
- Proposal 4.1.2-rv1 (Sec 4.1): Proposes a working assumption for a generic AI/ML traffic model parametrized by packet size (integer multiple of token size), packet arrival (bursts with Type-1/Type-2 importance), PDB, and Token success rate.
- Proposal 4.2.2-rv1 (Sec 4.2): Proposes two amended eXR models based on TR 38.838 for immersive communications: Model-1 (without Haptics) with updated parameters for immersive gaming and UL-heavy video uploading, and Model-2 (with Haptics) defining co-generated haptics packet generation and PDB.
- Proposal 4.3.2-1rv1 (Sec 4.3): Extends FTP Model 3 to incorporate packet delay budget (PDB), defining candidate PDB values and latency measurement methodology.
- Proposal 4.3.2-2rv1 (Sec 4.3): Proposes options for extending FTP Model 3 to incorporate multiple packet sizes, including independent traffic flows per UE (Option 1) or per cell (Option 2) with session/packet structures.
- Proposal 4.1.2 (Observation 1) (Sec 4.1): Observes that the traffic model for AI/ML services can be derived based on application layer simulation to extract Token success rate, delay budget, Token arrival rate, and Token size requirements.
- Proposal 6.1.2 (Sec 6): Proposes carrier frequency for NTN Ku-band as 14GHz for UL and 11GHz for DL.
- Proposal 6.2.2 (Sec 6): Models satellite antenna for NTN as an array using TXRU virtualization model Option-2B from TR 36.897, with Mg=Ng=1 and the Bessel function from 38.811 as FFS for simplified evaluations.