In development · Rel-20 studies → Rel-21 normative work
6G is taking shape through explicit choices about what to reuse from 5G, what to extend and what complexity not to carry forward. The current picture is a study baseline with unresolved branches, not a finished specification.
An adopted study result can guide design without being a normative 6G requirement. Maturity is shown separately from the meeting outcome.
Release 20 establishes requirements and technical direction. Release 21 is expected to carry the normative 6G specifications.
Current baseline
Values are only comparable when their definitions and evaluation scenarios match.
| Parameter | Current value | Definition and maturity | Evidence |
|---|---|---|---|
| Theoretical peak data rate | 36 / 18 Gbit/s Adopted | Downlink / uplink requirement in the current 6G scenarios report. Approved study requirement | 3GPP 38.914 |
| Connection density | 10⁶ devices/km² Adopted | Massive IoT requirement; this is device density, not a typical number of people. Approved study requirement | 3GPP 38.914 |
| HRLLC latency target | 1 ms Adopted | A service-specific radio target under defined assumptions, not end-to-end latency. Approved study requirement | 3GPP 38.914 |
| Basic waveform | CP-OFDM · CP-OFDM or DFT-s-OFDM Adopted | Downlink · uplink. The basic waveform remains an evolution of 5G NR. Study agreement | RP-261078 |
| 5th-percentile spectral efficiency | 0.675 / 0.45 bit/s/Hz Adopted | Downlink / uplink in the dense-urban evaluation scenario. Approved study requirement | RP-261496 |
| Basic / extended coverage | 143 / 153 dB MaxCL Adopted | Basic mobile broadband / extended massive-IoT coverage framework. Approved study requirement | RP-261496 |
| Horizontal positioning | 0.75 m indoor · 6 m urban macro (90% of UEs) Adopted | 90th-percentile horizontal error; the internal vertical target is 3 metres. Approved study requirement | RP-261496 |
CP-OFDM remains the downlink baseline. The uplink supports CP-OFDM and DFT-s-OFDM, with rank-2 uplink MIMO for DFT-s-OFDM.
Current movement: In October RAN1 dropped a dedicated low-PAPR enhancement for uplink 16QAM after evaluations showed net losses, and decided not to define transmit-filter coefficients for steep-filter low-PAPR schemes; RAN4 characterises the filters instead.
LDPC remains the data-channel baseline and Polar remains the control-channel baseline within the established 5G range.
Current movement: A possible third LDPC base graph is being restricted to data rates beyond the ordinary 5G range; no final inclusion decision exists. For control payloads beyond the 5G range, uplink control information above 1,706 bits will use at most six Polar segments; the downlink mother-code length is chosen at RAN1#127.
Each slot with synchronization blocks has two candidate SSB positions, and four-step random access is the baseline procedure. Two-step, RACH-less and one-step access stay under study as additional procedures that need a justification.
Current movement: The SSB length (4, 5 or 6 symbols) and its default period for initial cell search (five options from 20 to 160 ms) were deferred to RAN1#127 (16–20 Nov 2026). The period sets how long a base station can stay silent, the core of 6G network energy saving.
The study continues with a disaggregated CU/DU structure and explicit support for RAN-controlled data collection and authorized data access.
Current movement: Flow control, context handling and coordination across the split remain study topics.
6G standalone, sharing 5G spectrum through multi-RAT spectrum sharing (MRSS) with handover between 5G and 6G, is the base case. In October SA2 made single registration the interworking baseline: a phone is registered with either the 5G or the 6G core at a time.
Current movement: RAN#113 postponed any additional migration option to RAN#114 (7–10 Dec 2026). A 14-company coalition led by Samsung, Qualcomm and NTT DOCOMO wants a 6G-anchored dual connectivity; Nokia and Ericsson reject every dual-connectivity and dual-stack variant; CMCC, Orange and satellite operators want dual registration, kept alive in a simplified form that the core architects must confirm by RAN#114.
Interim agreements: one NAS protocol stack and one serving access and mobility function (xAMF) per phone; network functions stay the standardized building blocks, with separate registration (NRF) and routing (SCP) functions that may be collocated.
Current movement: SA2 did not choose between three overall architectures that differ in where AI sits, and moved the decision to SA2#178 (16–20 Nov 2026). In informal polls, 46 companies to 7 wanted essential connectivity to work without AI; AI agents as a core part of the architecture drew 16 yes and 32 no.
Combining a sub-7 GHz and a millimetre-wave carrier from base stations at different sites, over backhaul with up to 10 ms delay, was found feasible in the physical-layer and protocol studies.
Current movement: RAN#113 asked RAN2 to study a multi-vendor interoperable solution with two independent schedulers. Whether it builds on carrier aggregation or dual connectivity is not decided.
The study limits the number of formal device tiers and includes mobile broadband, fixed wireless access and massive IoT.
Current movement: In June the low tier got a one-antenna minimum and a 20 MHz radio-bandwidth baseline. In September the RAN plenary suspended text on 6G device types in its study report until the radio study's own device-type discussion concludes.
6G voice reuses the existing IMS architecture and procedures over a guaranteed-bit-rate QoS flow, with 6G as the IP access network. No new Diameter reference points between IMS and the 6G core.
Current movement: Voice continuity between 6G and 5G relies on keeping the IP address. Whether to adopt a voice fallback at all remains for further study.
September 2026 plenaries and October 2026 working groups
SA2 noted a consolidated summary of three candidate architectures and polled companies informally, but did not choose. The SA plenary wants the decision no later than SA2#178 in November.
SA2 approved interim principles for the 6G core: a single NAS protocol stack and one serving xAMF per phone, and single registration with either the 5G or the 6G core for interworking. Dual registration remains for further study.
RAN1 fixed two candidate SSB positions per slot and four-step random access as the baseline, and moved the SSB length and its default period to RAN1#127.
Evaluations showed net losses for spectrum shaping with 16QAM, so RAN1 does not pursue a dedicated low-PAPR enhancement for it.
Three camps split over dual connectivity and dual registration. RAN#113 kept 6G standalone with spectrum sharing as the base case, required shared NR/6G control-channel resources, and postponed any extra option to December.
RAN2 studies how to combine an FR1 and an FR2 carrier from different base stations over backhaul with up to 10 ms delay, with a multi-vendor interoperable solution as the goal.
The scenarios report now carries scenario-specific minimum values rather than only broad ambitions.
CP-OFDM stays in the downlink and CP-OFDM or DFT-s-OFDM stays in the uplink. More elaborate uplink variants need further evidence.
LDPC and Polar remain the baseline. A new LDPC graph survives only as a conditional option for higher data rates.
The meeting received opposing proposals and did not settle whether a narrowed shaping study should continue.
Read the editorial account in 3GPP Quarterly — September 2026.
20 6G decisions from 13 meetings in the September 2026 quarterly, plus the interim agreements of RAN1#126-bis and SA2#177 in October. Working-group agreements are study results that later meetings can revise, not plenary decisions. Each published claim points to a specification, TDoc or the quarterly evidence set. Maturity and meeting outcome are recorded separately.