R1-2508734
discussion
Evaluation assumptions for 6GR air interface
From Huawei
Summary
This document from Huawei and HiSilicon for the 3GPP RAN1#123 meeting on 6G Radio evaluation assumptions contains 25 proposals and 1 observation covering scenario definitions, BS/UE antenna modelling, traffic models (FTP extensions, AI Token communication, immersive XR), link budget methodology, NTN deployment evaluation, and ISAC performance metrics.
Position
Huawei proposes including the urban grid scenario from TR38.901 for automotive sensing evaluation in 6GR. For traffic modeling, they require extending FTP Model 3 (not Model 1) to incorporate Packet Delay Budget, with packets exceeding PDB being dropped, and propose a new Token-based traffic model (Option-1a) parameterized by Token arrival rate, Token size, Token success rate, and Token delay budget for AI/ML services, arguing Token success rate directly reflects AI/ML service quality and cannot be replaced by packet success rate. For link budget, they require using MPL (Maximum Path Loss) as the basic performance metric and insist CDL channel models with actual number of TxRUs be used in LLS for massive MIMO scenarios. For NTN, they propose a 3D hexagonal tessellation methodology rather than UV-plane projection for evaluating beam interference in multi-satellite dense (V)LEO constellations. For ISAC, they define reconstruction accuracy as RMSE between ground truth and estimated 3D point sets, and propose adding sensing target type rows to evaluation scenario templates.
Key proposals
- Proposal 1 (Sec 2.1): The urban grid with evaluation parameters for automotive sensing as defined in TR38.901 should be included as one of the scenarios for 6GR study.
- Proposal 3 (Sec 2.2): Proposes specific TXRU mapping parameters for around 7GHz carrier frequency: Combination 1: (24,16,2,1,1; 4,16) for 768 antenna elements and 128 TXRUs; Combination 3: (48,16,2,1,1; 8,16) for 1536 antenna elements and 256 TXRUs.
- Proposal 4 (Sec 2.3): For 6GR evaluation, the maximum number of UE antenna elements should at least consider 16.
- Proposal 6 (Sec 2.4): Multiple user densities per TRxP shall be considered for 6GR evaluation, e.g., chosen from [10, 20, 30, 50] for each of the deployment scenarios.
- Proposal 7 (Sec 2.5): Multiple options for BS Tx Power assumptions can be considered for 6GR evaluations allowing for different evaluation purposes.
- Proposal 9 (Sec 3.1): Extending the FTP model to incorporate Packet delay budget (PDB) and one/multiple packet sizes should be based on FTP Model 3 instead of FTP Model 1.
- Proposal 10 (Sec 3.1.1): As an extension to FTP Model 3, FTP Model 3a models one packet size, one mean inter-arrival time value and one packet delay budget (PDB).
- Observation 1 (Sec 3.2.3): It is Token success rate directly reflecting the AI/ML service quality and cannot be replaced by the packet success rate used by the XR traffic model.
- Proposal 14 (Sec 3.2.3): Model Option-1a (parameterized by Token arrival rate, Token size, Token success rate, and Token delay budget) should be used as the traffic model representing the AI/ML services.
- Proposal 16 (Sec 3.3.1): Towards modelling the advanced XR (i.e., immersive gaming), extend the existing XR traffic model with modifications on data rate (100/300/500 Mbps), frame generation rate (90/120 fps), and packet size distribution.
- Proposal 18 (Sec 3.4): Modelling the impact of bidirectional traffic flows caused by the TCP ACK needs to be well justified first.
- Proposal 19 (Sec 4): Table 12 is used for 6G Link budget template, where the MPL should be used as the basic performance metric for coverage analysis, towards studying the coverage of around 7GHz co-site deployed with 5G mid-band.
- Proposal 20 (Sec 4): Interference density/margin should be obtained by SLS based on different scenarios. CDL channel model with actual number of TxRUs should be used in LLS for link budget analysis.
- Proposal 22 (Sec 5): The system level simulations for 6G-based NTN should consider dense/sparse (V)LEO constellation with massive beam footprints, uneven UE distribution, and a modelling methodology based on 3D hexagonal tessellation of Earth.
- Proposal 24 (Sec 6.2): Defines reconstruction accuracy for evaluating environment object reconstruction as RMSE between ground truth 3D points and estimated points of one environment object.