Wireless Field Guide
Topic hub

6G and research

What 6G is at the time of writing and what it is not. The ITU-R IMT-2030 framework, the 3GPP Release 20 studies, and the research themes behind them, from sensing and RIS to AI and sub-THz links.

Updated 15 Nov 2026 · 6G, IMT-2030, Release 20, ISAC, RIS, AI-native air interface, sub-THz

Overview

At the time of writing, 6G is not a standard. It is a framework from ITU-R, a set of 3GPP studies, and a large body of research. Keeping those three apart is the most useful habit in this field. A paper that shows a gain in simulation, a lab demo at a trade show and a 3GPP technical report are different kinds of evidence, and none of them is a 6G specification yet.

The framework is ITU-R Recommendation M.2160, approved in November 2023. It sets the objectives for IMT-2030, the ITU name for the system that 6G technologies will be submitted to. M.2160 states that ITU plans to finish its initial IMT-2030 standardisation no later than 2030. In 3GPP, Release 20 has two tracks, according to a 3GPP article by the TSG SA chair. One continues 5G-Advanced. The other is dedicated to 6G studies, with normative 6G work planned for Release 21. Xingqin Lin’s overview of Release 20 (arXiv, 2025) describes both tracks in more detail.

This hub is for engineers who want to follow 6G work without mistaking a research result for a product plan.

Researchpapers, simulations, testbeds, lab demoscontributionsITU-R M.2160IMT-2030 frameworkapproved 11/20233GPP Release 206G study itemstechnical reports3GPP Release 21first normative6G specificationsITU-R IMT-2030initial standardsno later than 2030framework and studies: not specificationsnormative specifications5G-Advanced continues in parallel: Releases 18, 19 and 20
Where a 6G idea stands. Research results feed the ITU-R framework and the 3GPP studies, but only the 3GPP Release 21 specifications and the later ITU-R IMT-2030 standards are normative. 5G-Advanced continues in parallel through Release 20.

Key ideas

Usage scenarios set by ITU-R

M.2160 extends the three IMT-2020 scenarios and adds three new ones. The six are immersive communication, hyper reliable and low-latency communication, massive communication, ubiquitous connectivity, artificial intelligence and communication, and integrated sensing and communication. The capability values it gives are described as targets for research and investigation, not requirements. Use M.2160 when someone asks what 6G is meant to do, and the 3GPP study reports when they ask how.

Integrated sensing and communication

ISAC uses the same waveform, spectrum and hardware for both data and radar-like sensing. M.2160 lists range, velocity and angle estimation, object detection, localisation, imaging and mapping among the capabilities it needs. The engineering problems are concrete: full-duplex operation or separate receive arrays, waveform trade-offs between data rate and sensing resolution, and privacy. Liu and co-authors survey the field in Integrated Sensing and Communications (IEEE JSAC, 2022).

Reconfigurable intelligent surfaces

A RIS is a large panel of tunable elements that reflects an incoming wave in a controlled direction, with no power amplifier of its own. M.2160 names RIS among potential technologies. Di Renzo and co-authors explain the physics and open problems in Smart Radio Environments Empowered by Reconfigurable Intelligent Surfaces (IEEE JSAC, 2020), and Wu and Zhang give the reflecting-surface view in Towards Smart and Reconfigurable Environment (IEEE Communications Magazine, 2020). Check every RIS result for how the channel is estimated, since the surface has no receiver, and for how large the panel must be to beat a direct path.

AI in the air interface

M.2160 expects IMT-2030 to consider an AI-native air interface, using AI for functions such as symbol detection, decoding and channel estimation. 3GPP has already studied AI and machine learning for NR. The study report is TR 38.843, listed under Releases 18 and 19, and Lin gives an overview of it. The Release 19 version of TS 38.300 describes AI/ML-based beam prediction. On the research side, O’Shea and Hoydis set out the idea of learned transceivers in An Introduction to Deep Learning for the Physical Layer (2017), and Hoydis and co-authors argue for an AI-native design in Toward a 6G AI-Native Air Interface (2021). For the learning methods themselves, see the AI/ML Field Guide deep learning hub.

Sub-THz and new spectrum

Rappaport and co-authors review links above 100 GHz in Wireless Communications and Applications Above 100 GHz (IEEE Access, 2019). Very wide channels are available there. The costs are high free-space and atmospheric loss, phase noise, low power amplifier efficiency and blockage. ITU-R P.676 gives the gaseous attenuation model you need for any link budget at these frequencies. Which bands 6G will use is a question for national regulators and the World Radiocommunication Conference, covered in the spectrum and regulation hub.

Where to start

You need solid basics before reading 6G papers critically. Work through Noise, SNR and capacity, Fading channels and OFDM from scratch. Then read the 5G NR hub and the MIMO and beamforming hub, because most 6G proposals extend those. The weekly research digest tracks new papers.

Tools and platforms

Resource What for Who publishes it
ITU-R M.2160 IMT-2030 framework and objectives ITU-R
3GPP portal Release 20 study reports and work plans 3GPP
TR 38.901 Channel model used in most NR and 6G studies 3GPP
ITU-R P.676 Gaseous attenuation from 1 to 1000 GHz ITU-R

Vendor positions on 6G are covered on the Ericsson, Nokia and Qualcomm platform hubs. Test equipment makers are on the Keysight and Rohde & Schwarz hubs.

Common mistakes

  • Calling a research prototype or trade-show demo “6G”. Say what it is and which body, if any, it relates to.
  • Treating M.2160 capability ranges as requirements. They are research targets.
  • Comparing a new scheme against a weak baseline, or against a 5G system with features switched off.
  • Ignoring hardware limits in sub-THz papers: phase noise, converter power and array calibration.
  • Assuming an AI receiver that works on one channel model will work on another. Ask how it was tested.

Going deeper

Latest news on this topic

Recent papers

  • A Four-Branch Radar Target Simulator with Fractional Delay on a PlutoSDR

    F. Euchner, S. ten Brink · arXiv eess.SP · digest

    This paper turns an unmodified ADALM-Pluto into a radar target simulator for sub-6 GHz sensing experiments. FPGA logic generates four independent echoes, each with its own delay, Doppler shift and gain, and fractional-delay filters set range below one sample. Delay and carrier phase are updated together, so a moving target shows the correct frequency-dependent Doppler across an OFDM signal. Over-the-air lab tests with an ESPARGOS radar array reproduced 10 mm range steps with largest median errors of 2.6 mm in overclocked mode and 1.6 mm in standard mode. The fixed hardware latency puts the nearest emulated target at an equivalent 62.1 m (with the interface overclocked to 160 MHz, beyond its 61.44 MHz rating) or 197.7 m at standard settings, and the FPGA source and firmware are on GitHub.

  • Real-Time Low-Complexity Zak-OTFS with Superimposed Spread Pilots on GPUs

    C.-H. Tung, J. Zheng, V. Yogesh et al. · arXiv eess.SP · digest

    Zak-OTFS with spread pilots overlays the pilot on the data across the whole delay-Doppler grid, which saves resources but makes channel estimation and pilot cancellation expensive. Using the chirp structure of the pilot, the authors reduce the cross-ambiguity calculation to one 2D IFFT plus element-wise operations, over 7000 times faster than a partly vectorised direct GPU implementation at a frame size of 2053 by 31. Their full receiver pipeline with one turbo iteration fits a 1.03 ms latency budget on an NVIDIA RTX 6000 Ada GPU. Larger frames give more processing gain, so pilot power can drop and the bit error rate is close to converged after one iteration. All results use a simulated ITU Vehicular-A channel, with no radio in the loop.