R1-2600137
discussion
Evaluation assumptions for 6GR air interface
From Huawei
Summary
This Huawei/HiSilicon contribution addresses remaining 6G radio evaluation assumptions, covering UE antenna modelling and various traffic models. It presents four formal proposals regarding UE antenna configurations, FTP-3 extension for mixed packet sizes, bidirectional traffic modelling, and IoT traffic model reuse.
Position
Huawei proposes specific UE antenna modelling with combinations ranging from 1T1R up to 8T16R (16R for FWA only) across carrier frequencies from 700MHz to 15GHz, with antenna element locations for transmission being a subset of reception locations. They require FTP-3 extension to use Alt2 with X=2 mixed packet sizes per UE and independent Poisson arrival rates, rejecting Alt1 due to unbalanced traffic load severely impacting UPT performance. They question the necessity of bidirectional traffic modelling with TCP ACK impact, citing that 5G evaluations used UDP without transport layer impact and that TCP modelling assumptions themselves strongly affect simulation results. They oppose developing new IoT traffic models and instead require reuse of the IMT-2020 mMTC traffic model from TR 37.910. For AI/ML traffic, they present categorized characteristics including two-level packet importance dependency, packet delay budget of 10-15ms for real-time services, and packet success rate definitions considering packet dependency.
Key proposals
- Proposal 1 (Sec 2): For 6GR evaluation, RAN1 to model the UE antenna as follows for carrier frequencies other than 30GHz, with the note that the antenna element/location of T is a subset of the element/locations for R.
- Proposal 2 (Sec 3.2.2): Regarding the study on extending FTP-3 to consider mixed packet sizes, X=2 packet sizes are proposed; Y=X packet sizes and Z=X packet arrival rates/mean inter-arrival time are simulated for each UE, with up to Y packets arriving simultaneously, and the inter-relation for S_i and T_i depending on service QoS characteristics.
- Proposal 3 (Sec 3.3): Modelling the impact of bidirectional traffic flows caused by TCP ACK needs to be well justified first.
- Proposal 4 (Sec 3.4): mMTC traffic model from IMT-2020 (TR 37.910) shall be reused for 6G IoT evaluation.
- Section 3.1 view: AI traffic should be categorized into real-time and non-real-time services with specific characteristics including error tolerance, importance differentiation, data arrival patterns, packet delay budget (10-15ms for real-time), and packet success rate requirements.
- Section 3.2.1 view: Alt2 (Y=X packet sizes per UE) is preferred over Alt1 (Y=1 packet size per UE) for FTP-3 extension due to more uniform traffic distribution and robust UPT performance across different packet size ratios.
- Section 3.1 view: Packet importance should be modelled with a two-level importance model where failure of high-importance packets causes associated low-importance packets to be treated as failures.