As the slogan of the first 6G 3GPP RAN1 meeting (RAN1#122) declared:

Only time will tell whether the next generation will follow the success of the “even” releases (2G, 4G/LTE) or the more turbulent path of the “odd” releases (3G, 5G). At this stage, it is far too early to make reliable predictions. Nevertheless, it is exciting to witness how 6G is beginning to take shape, starting from its very first technical meeting.

It should be noted, however, that the foundations for 6G were laid long before RAN1#122. Significant research has already been carried out in projects such as HEXA-X, HEXA-X II, 6G Flagship (Finland), EU Horizon projects like 6G-SANDBOX, 6G BRAINS, and REINDEER, 6G-SHINE in China, etc. Within 3GPP itself, companies began aligning their 6G directions officially in March 2025 at the first 6G workshop.

Before RAN1 began its dedicated 6G study, the RAN plenary initiated a high-level investigation into 6G scenarios and use cases (RP-243327). The results of that effort will be captured in TR 38.914, serving as the 6G counterpart to the Rel-14 TR 38.913 for 5G NR.

The scope of the 6G Study Item (SID) is defined in RP-251881 (“New SID: Study on 6G Radio”), and the associated work plan is described in R1-2506303.

In future posts, I plan to collect and summarize all RAN1 agreements related to the 6G study on a dedicated page.

6G at RAN1#122 meeting

The first 6G discussions in RAN1 took place at RAN1#122, held from August 22–26, 2025, in Bangalore, India. The agenda grouped 6G work items into dedicated topics (see full agenda in R1-2505100). In addition to the key 6G topics summarised in the sections below, some of the most central topics will only begin in later meetings, starting with RAN1#124 in February 2026. The workflow for each agenda item (AI) followed a common structure. Feature Leads (FLs) were appointed to summarize submitted technical documents (TDocs), lead offline discussions, and prepare draft agreements. During online sessions, selected TDocs were presented (chosen via voting; see Excel list), allowing companies to highlight their key proposals and trigger further debate.

Offline discussions gave companies space to refine the FLs’ draft agreements. These were then brought back to the online sessions, where they were further discussed, revised, and ultimately captured in the official chair notes.

At this early stage, the agreements should not be seen as final decisions. Instead, they serve as guidance and orientation points for structuring future discussions.

11.1 Overview

The FL summary is: R1-2506618 The first group of agreements is closely related to the question on How to design a single RAT to serve diverse devices, channel bandwidth. What device types will be introduced in 6G, and what will be their common feature set and distinguishing add-on features (modules) dedicated to specific use cases / device types. For the device types, no specific list was agreed but the examples may be Fixed Wireless Access (FWA), smartphones, low complexity smartphones, wearables, sensors, etc.

6G must improve both downlink and uplink coverage compared to 5G, especially for cell-edge and deep-indoor scenarios. The targets need to reflect the diversity of devices (e.g., smartphones, IoT sensors, wearables, FWA) and use cases (from massive IoT to XR). By setting unified coverage benchmarks such as maximum coupling loss (MCL), the system can ensure a consistent user experience across different device categories and deployment scenarios.

Synchronization signals are critical for initial access and mobility. The study agreement below asks to determine which factors—such as minimum channel bandwidth, numerology, low-tier device capabilities, or detection reliability—should shape the design of the 6G synchronization signal (similar to NR’s SSB). Periodicity also needs careful design to balance fast cell detection and power efficiency.

Multi-RAT Spectrum Sharing (MRSS) allows NR and future 6G carriers to coexist on the same spectrum. This agreement highlights the need to capture insights from LTE-NR Dynamic Spectrum Sharing (DSS), where overhead and scheduling conflicts affected efficiency.

The Bandwidth Part (BWP) framework in NR was introduced to reduce UE power consumption by allowing devices to operate on narrower portions of a wide carrier. However, complexity in configuration and signaling was a challenge. This agreement ensures 6G study captures both the benefits and drawbacks of NR’s BWP framework.

This agreement aims to capture what worked and what challenges arose in 5G NR when combining spectrum across different bands and carriers (carrier aggregation, dual connectivity, supplementary UL (SUL)).

The agreement reflects the need to define the minimum bandwidth that even the most basic 6G devices (e.g., IoT sensors, wearables, low-cost smartphones) must support without aggregation. Options from 3 MHz to 20 MHz are being discussed, with flexibility left open for UL vs. DL asymmetry, duplexing differences, or subcarrier spacing.

This agreement builds on 5G NR’s experience with duplexing (how UL and DL share spectrum in time/frequency). For 6G, RAN1 will examine both traditional approaches (e.g., FDD, TDD) and newer concepts like sub-band full duplex (SBFD) and dynamic duplexing. The study will clarify feasibility, interference management, and device complexity impacts. It also leaves the door open for other modes, ensuring flexibility as new requirements (e.g., uplink-heavy AI/immersive use cases) emerge.

6G is expected to unify terrestrial networks (TN) and non-terrestrial networks (NTN, e.g., satellite, High-Altitude Platform Station (HAPS)) into a single design. This agreement directs study on how NTN-specific challenges (large delay, Doppler, coverage variation) affect 6G air interface design. It also leverages experience from NR-based NTN and IoT NTN deployments to avoid fragmentation and ensure smooth operation across TN/NTN, supporting global coverage and ubiquitous service delivery.

11.2 Evaluation assumptions

The FL summary is: R1-2506581 This agreement sets the foundation for evaluation work in 6G by aligning with TR 38.914 as the baseline reference for deployment scenarios. It ensures that general assumptions—such as antenna models, spectrum settings, and link budgets—are harmonized in AI 11.2, providing consistency across different studies and AIs. At the same time, it leaves flexibility by allowing topic-specific assumptions (e.g., waveform or MIMO) to be treated separately and deferring key design parameters like subcarrier spacing to their dedicated agenda. This structure helps organize work efficiently and avoids overlap across discussions.

Conclusion

• Template in R1-2506582 is to be used for collecting inputs from companies.
  • Additional NTN or TN assumptions, if any, or any necessary change of the parameters, are to be incorporated into the updated one of R1-2506582.

The conclusion provides a practical framework for gathering and maintaining inputs across companies. Using a single template (R1-2506582) guarantees uniformity in how assumptions are reported, making them easier to consolidate. A post-meeting discussion will follow in between the meeting to collect companies inputs on the common evaluation assumptions.

This agreement highlights the importance of selecting and evolving traffic models to ensure realistic evaluation of 6G performance. Established LTE and NR traffic models (e.g., FTP models from TR 36.814/36.872, VoIP and instant messaging from TR 36.814 and TR 38.840, XR from TR 38.838) provide a solid baseline and allow continuity with previous generations. However, 6G is expected to support far more diverse applications — including AI/ML-driven services, immersive XR/Metaverse-style experiences, and highly interactive communications — which current models cannot fully capture. Therefore, new models are proposed to account for mixed packet sizes, variable arrival times, stricter packet delay budgets, and bidirectional dependencies such as UL acknowledgments affecting DL throughput. By combining legacy models for comparability and new ones for emerging use cases, evaluations will reflect both backward consistency and forward-looking realism.

11.3 Waveform and frame structure for 6GR air interface

11.3.1 Waveform

The FL summary is: R1-2506595

This was the first agreement of 6G study in RAN1. It reflects continuity with 5G NR: CP-OFDM and DFT-s-OFDM have proven performance, ecosystem maturity, and hardware support. DFT-s-OFDM in particular offers lower peak-to-average power ratio (PAPR), which is important for uplink coverage and UE power efficiency. At the same time, the door remains open for enhanced versions (e.g., spectrum shaping, filtering) or even new OFDM variants, provided they demonstrate clear benefits such as better coverage, energy efficiency, or spectral efficiency.

For the downlink, CP-OFDM remains the primary choice thanks to its flexibility and MIMO-friendliness. However, 3GPP is allowing further exploration: DFT-s-OFDM could be added in DL for scenarios like coverage enhancement (e.g., IoT, NTN) where lower PAPR brings benefits. Any additional waveform proposal must be concrete—clearly tied to a use case, show how it multiplexes with CP-OFDM, and demonstrate practical gains while maintaining compatibility with multi-RAT spectrum sharing (MRSS) and efficient multi-user scheduling.

This note signals that the bar for introducing new waveforms is set high. Proponents must go beyond theoretical advantages and provide quantified evidence—e.g., coverage gains in dB, efficiency improvements, or energy savings—while also analyzing the trade-offs in complexity, coexistence with CP-OFDM/DFT-s-OFDM, and specification impact. This ensures that only well-justified enhancements or additions are considered, preventing fragmentation of the air interface and keeping the 6G design focused and practical.

In general, a number of alternative or enhanced waveforms were proposed in the TDocs, such as:
For Uplink (UL):

• Enhanced DFT-s-OFDM (e.g., with frequency domain spectrum shaping / FDSS, spectrum extension, or multi-layer extensions) – improves coverage and reduces PAPR.
• Filtered / windowed OFDM variants (e.g., f-OFDM, W-OFDM) – for better out-of-band emission (OOBE) control.
• CE (Constant Envelope)-OFDM, FM (Filtered Multitone)-OFDM, BS (Band-Swep)-OFDM – advanced OFDM-based candidates to address specific KPIs like energy efficiency or Doppler robustness.
• Low-PAPR techniques (precoding, spectrum shaping, etc.) – targeted to NTN or cell-edge scenarios.

For Downlink (DL):

• DL DFT-s-OFDM – proposed by multiple companies for better coverage in IoT or NTN cases.
• Filtered OFDM (F-OFDM) – reduces OOBE and supports forward compatibility with mixed services
• Other OFDM-based enhancements (e.g., applying windowing or precoding to CP-OFDM) – to improve robustness in high-mobility or wideband deployments.
• OTFS (Orthogonal Time Frequency Space, incl. Zak-OTFS) – positioned by some as more robust to high-Doppler and potentially beneficial for ISAC (communication + sensing).
• Orthogonal Sequence Division Multiplexing (OSDM) – studied as another alternative structure for DL.

Almost all alternatives are still OFDM-based (to keep hardware reuse, MRSS compatibility, and migration with 5G), but with enhancements targeting coverage, PAPR, spectral efficiency, or sensing integration. Non-OFDM families (e.g., OTFS) are not excluded but face a high justification burden.

11.3.2 Frame structure

The final version of the feature lead summary is R1-2506604, but it is currently unavailable. The latest version is R1-2506603.

This agreement establishes the baseline subcarrier spacing (SCS) values to be studied for different frequency ranges in 6G. The intention is to maintain backward compatibility with 5G NR (e.g., 15 kHz for FDD, 30 kHz for TDD) while also exploring higher SCS options (e.g., 60 kHz, 120 kHz) for mid- and high-frequency bands to balance coverage, latency, and phase noise resilience. Importantly, the agreement also highlights that numerology for sensing use cases is deferred to the dedicated sensing study item, acknowledging that sensing may require different symbol durations or cyclic prefixes than communication.

The agreement indicates a preference for simplification by supporting only the normal cyclic prefix (NCP), which is already widely deployed in 5G NR. However, the “FFS” (For Further Study) note leaves open the possibility of other cyclic prefix options (e.g., extended or flexible CP) if future evaluations—particularly for integrated sensing and communications (ISAC)—show a need for longer guard intervals to handle large delay spreads or sensing echoes.

11.4 Channel coding and modulation for 6GR interface

11.4.1 Channel coding

The FL summary for control channel coding is: R1-2506611.
The FL summary summary for data channel coding is: R1-2506590

This guidance clarifies that evaluations for control channel coding in 6G should focus on error rate metrics such as BLER (Block Error Rate) and FAR (False Alarm Rate), while also allowing analysis of complexity and decoding latency. The intention is to make comparisons consistent and transparent across proposals. Importantly, companies must show how their extensions differ from 5G NR control channel codes (polar codes and short-block codes), and justify why new methods are needed. In practice, this means that unless an extension offers measurable gains—e.g., lower error probability in extreme reliability cases, or lower complexity—it may not be considered worthwhile.

Here, the emphasis is on BLER results as the primary performance measure, complemented by throughput, complexity, and decoding latency analysis. Proponent companies need to provide detailed assumptions (scenarios, HARQ design, modulation orders, block sizes, etc.), ensuring that results are comparable. As with control channels, the baseline is 5G NR channel coding (LDPC for data), so any proposed extension must show clear benefits such as supporting much higher data rates, larger payloads, or improved reliability with an acceptable complexity/performance trade-off. This reflects industry consensus that 6G should evolve from 5G’s LDPC/Polar foundation, rather than replacing it with entirely new schemes.

In summary, most companies want to keep LDPC (data) and Polar (control) as the baseline. Extensions to LDPC (new base graphs, higher parallelism, larger lifting sizes) are the most likely evolution. Extensions to Polar may be considered if larger control payloads are proven necessary. Completely new schemes are not seriously on the table at this stage.

11.4.2 Modulation, joint channel coding and modulation

The FL summary is: R1-2506579

This agreement essentially means that 6G starts with the proven 5G NR modulation formats (QAM family plus π/2-BPSK for uplink coverage). These provide a balance of spectral efficiency (higher-order QAM for high SNR scenarios) and robustness (lower orders for coverage and control). While 1024QAM is set as the downlink baseline, uplink is capped at 256QAM due to UE power and RF hardware constraints. The “FFS” clause leaves the door open for 4096QAM or other novel modulations if they show compelling gains in throughput, coverage, or power efficiency. Another open item is whether enhancements like constellation shaping (probabilistic or geometric) or low-PAPR variants should be introduced, as they can improve energy efficiency and robustness. An interesting aspect is whether learning-based modulation, i.e., AI/ML-generated constellations adapted to channel conditions will be considered as a viable alternative.

Joint Modulation and Coding (JMC) is not part of the current agreement, but several companies proposed it as a longer-term study item. JMC aims to go beyond the traditional 5G Bit-Interleaved Coded Modulation (BICM) framework by jointly designing coding and modulation (e.g., multilevel coding, trellis-coded modulation, probabilistic shaping + coding, or iterative detection/decoding). The potential benefits are better spectral efficiency, unequal error protection, and improved performance for high-order QAM, but this comes at the cost of higher complexity and implementation challenges.

11.5 Energy efficiency

The final feature lead summary should be R1-2506602, but the latest available one is: R1-2506601.

TR 38.864 defines reference configurations (e.g., Cat-1 and Cat-2 BS types, FR1/FR2 setups) used in 5G for evaluating energy consumption. For 6G, reusing these as a baseline ensures consistency, but they may need updates for wider bandwidths, new numerologies, and larger antenna arrays.

The TR 38.864 BS power model includes sleep states, scaling rules, and transition times. Using this as a starting point provides continuity, but adaptations are required to reflect advances in hardware (e.g., faster sleep transitions, higher efficiency PAs) and 6G-specific deployments.

Unlike throughput or latency, energy efficiency metrics are not yet standardized. Metrics need to capture both instantaneous power savings and long-term sustainability benefits. For UEs, this may include “energy per bit” or “battery life extension” under specific workloads. For BS, system-level efficiency (e.g., Joules/bit over varying loads) is important.

5G already defined UE power models for smartphones, RedCap, and low-power devices. Extending these into 6G provides a foundation but must account for new device classes (e.g., XR headsets, sensing devices). This ensures energy efficiency is studied not only for high-end devices but also lightweight and constrained UEs.

Defining baselines ensures that EE improvements are measured against realistic, common setups rather than cherry-picked scenarios. For BS, this includes defining load levels, bandwidths, and antenna configurations. For UEs, this means agreed profiles for smartphones, IoT, and reduced-capability devices.

11.6 AI/ML in 6GR interface

The FL summary is: R1-2506456.

This agreement sets the groundwork for systematically identifying and comparing promising AI/ML use cases in 6G. The focus in the first meetings is not to finalize solutions but rather to map candidate use cases and describe them in a structured way (inputs/outputs, training assumptions, inference split, etc.), so that only the most relevant and feasible ones are selected for deeper study. The agreed reporting template ensures comparability across proposals and helps to highlight trade-offs in performance, complexity, power consumption, and deployment feasibility. In essence, this stage is about shortlisting the most promising AI/ML applications (e.g., beam prediction, AI receivers, CSI feedback, continuous learning) before detailed design and evaluation efforts begin.

The most popular / frequently highlighted AI/ML use cases by the companies for 6G radio (6GR) are:

1. AI/ML Receivers (Neural Receivers)
   • Use ML models to replace or enhance classical receiver blocks (channel estimation, equalization, detection).
   • Goal: better performance under high interference, extreme MIMO, and mobility conditions.
   • Reference symbol (DMRS) design with neural receivers – AI/ML receivers may allow more flexible or sparse DMRS patterns, reducing pilot overhead while maintaining reliable channel estimation, especially in wideband/high-antenna scenarios
2. Beam Management
   • Already studied in Rel-18/19 for 5G-A, carried forward to 6G:
     • Intra-cell beam prediction (spatial and temporal).
     • Inter-cell beam prediction (new in 6G, predicts beams across multiple cells).
     • Tx-Rx beam pair prediction (joint prediction to reduce delay in beam activation).
     • Beam prediction with continuous learning (adapts beam decisions in real time).
   • Widely seen as a front-runner for early normative work.
3. CSI Feedback Enhancements
   • CSI Prediction (UE-sided model) – studied in Rel-19.
   • CSI Compression (two-sided model) – ongoing in 5G-A Rel-20, strong candidate for 6G baseline.
   • Use AI/ML to reduce feedback overhead while improving accuracy.
   • RS design enhancements for CSI feedback – AI/ML-assisted CSI reporting may enable more adaptive or reduced-density reference signal designs, lowering overhead while preserving accurate channel state representation for both prediction and compression tasks.
4. Positioning
   • Explored in Rel-18/19 with multiple variants (UE-sided direct positioning, NG-RAN/LMF-assisted).
   • Still relevant but not seen as a first-priority 6G use case; may come later.
5. Continuous Learning (CL) Use Cases)
   • A new area for 6G: models adapt online to changing network/user conditions.
   • Applies to many functions: beam selection, mobility management, power control, MCS selection.
   • Strongly emphasized as essential for “AI-native” air interface.

Other AIs

Once the initial discussions on the topics above have matured, the study will continue in 2026 with the other core 6G topics:

• 11.7 Initial Access
• 11.8 MIMO operation
• 11.9 Physical-layer control, data scheduling, and HARQ
• 11.10 Duplexing
• 11.11 Spectrum utilization and aggregation
• 11.12 NTN
• 11.13 Other physical-layer signals, channels, and procedures
• 11.14 Sensing

© 2025 6ghow.com. This content is licensed under CC BY-NC 4.0. Attribution required.