6G — The Complete Technical Guide

We are barely five years into the 5G era, and the next generation is already taking shape. Not as a marketing buzzword, but as a concrete engineering programme spanning 3GPP study items, ITU-R recommendations, and multi-billion-euro research consortia. 6G is not simply “faster 5G.” It is a fundamental reimagining of what a wireless network is, what it can sense, what it can compute, and how intelligence flows through every layer of the stack. This article is the definitive technical walkthrough — from the 3GPP Release 21 timeline all the way down to the sub-terahertz waveforms that will carry your data in 2031.

The Road to 6G: 3GPP & ITU Timeline

Unlike 5G, which emerged primarily from 3GPP Release 15 in 2018, the 6G journey is a carefully orchestrated dual-track effort between the ITU (setting the vision and performance requirements) and 3GPP (writing the implementable specifications).

ITU-R IMT-2030 Framework

In November 2023, the ITU published Recommendation ITU-R M.2160-0, defining the IMT-2030 framework. This is the 6G equivalent of what IMT-2020 was for 5G. It establishes six usage scenarios (expanded from 5G’s three), fifteen capability dimensions, and the performance envelope that any candidate technology must meet. The technology submission window opens in 2027, with ITU evaluation concluding around 2030.

3GPP Release 20 & 21

3GPP has adopted a unique dual-track structure for Release 20 (July 2025 – June 2027). One track completes 5G-Advanced evolution with 126 Work Items. The other track, Rel-20_6G, is dedicated entirely to 6G feasibility studies — no normative specs yet, only Technical Reports like TR 22.870 (6G use cases, approved at TSG SA#111) and TR 38.914 (6G scenarios and requirements, v0.4.0 agreed at RAN#111 in March 2026).

Release 21, targeted to begin around Q1 2027, will produce the first normative 6G specifications. ASN.1/OpenAPI freeze is projected by March 2029 — just in time for IMT-2030 technology submission to the ITU.

IMT-2030 KPIs: 5G vs 6G Head-to-Head

The ITU’s IMT-2030 framework redefines the performance envelope across fifteen dimensions. But there is a critical distinction engineers must understand: the ITU sets aspirational capability targets, while 3GPP TR 38.914 defines the concrete minimum technical performance requirements that the actual 6G radio technology must meet. The table below shows both.

Six Usage Scenarios (Expanded from Three)

5G defined three pillars: eMBB, URLLC, and mMTC. IMT-2030 expands to six, reflecting the convergence of communication with AI, sensing, and computing:

14 Deployment Scenarios (TR 38.914 Section 4)

TR 38.914 defines fourteen concrete deployment scenarios — each with specific carrier frequencies, inter-site distances, antenna configurations, user distributions, and mobility profiles. These are not theoretical — they are the exact scenarios against which 6G radio technology will be evaluated.

Architecture Requirements (TR 38.914 Section 5.2)

TR 38.914 mandates specific architecture principles that will shape 6G RAN design. These are not aspirational — they are requirements the 6G system shall fulfill:

New Services Defined (TR 38.914 Section 5.4)

Beyond traditional mobile broadband, TR 38.914 mandates support for these service categories:

The 6G Spectrum Landscape

6G will not operate in a single band — it will orchestrate a symphony of frequencies spanning from below 1 GHz to above 275 GHz. TR 38.914 confirms that 6GR shall support frequencies between 410 MHz and 52.6 GHz for terrestrial networks (including all existing NR operating bands), with the 7 GHz band (FR3) appearing in 12 of 14 deployment scenarios. Sub-THz bands above 52.6 GHz are studied for future extensions.

Upper Mid-Band (FR3: 7–24 GHz)

WRC-23 identified 6.425–7.125 GHz and 10–10.5 GHz for IMT globally. Additional bands under study for WRC-27 include 7.125–8.4 GHz and 14.8–15.35 GHz. This “sweet spot” offers better propagation than mmWave while providing significantly more bandwidth than sub-6 GHz. Ericsson and Qualcomm have already validated 6G physical-layer prototypes in the 6–8 GHz cmWave range.

Sub-Terahertz (100–300 GHz)

The true frontier. WRC-27 Agenda Item 1.7 evaluates candidate sub-THz bands: 102–109.5 GHz, 151.5–164 GHz, 167–174.8 GHz, 209–226 GHz, and 252–275 GHz. These frequencies can provide up to 100 GHz of contiguous bandwidth (compared to 1 GHz maximum in 5G mmWave), enabling the 1 Tbps peak rate target. But atmospheric absorption, particularly water vapor peaks at 183 GHz and 325 GHz, limits range to short-distance, high-density use cases.

The 6G Base Station: Architecture Revolution

The 5G gNB as we know it — a centralized unit managing a fixed cell with boundaries — will undergo a fundamental transformation. TR 38.914 defines BS antenna configurations up to 2,304 elements at 7 GHz and 4,096 elements at 30 GHz (compared to 5G’s typical 64T64R). The 6G “base station” concept dissolves into a distributed, AI-native, sensing-capable fabric of access points, intelligent surfaces, and computing nodes.

From CU/DU/RU to Fluid Disaggregation

5G introduced the CU-DU-RU functional split. 6G pushes this further into fully virtualized, containerized, cloud-native microservices. Every RAN function becomes a software workload that can be placed anywhere — at the edge, in the cloud, or distributed across multiple sites. The O-RAN Alliance’s next-generation research group (nGRG) is actively defining this architecture, including communications-computing convergence where RAN nodes provide compute resources alongside connectivity.

Cell-Free Massive MIMO

This is perhaps the most radical shift. Instead of a UE connecting to “the nearest cell,” it is simultaneously served by many distributed access points (APs), all coordinated by a central processing unit. There are no cell edges. There is no cell-edge performance degradation. Every user receives macro-diversity from multiple directions, with the AP cluster dynamically adapting to their location.

The coordination overhead is significant — requiring high-capacity fronthaul and sophisticated joint precoding algorithms — but the payoff is a uniform quality-of-service experience that today’s cell-centric networks simply cannot achieve. Research shows cell-free architectures can deliver 10× improvement in 95th-percentile user throughput compared to conventional massive MIMO.

Reconfigurable Intelligent Surfaces (RIS)

RIS consists of arrays of programmable meta-atoms (sub-wavelength elements) that dynamically control the reflection, refraction, and scattering of electromagnetic waves. Think of them as “smart mirrors” for radio signals. They are nearly passive (consuming milliwatts, not watts), require no RF chains, and can be embedded into building facades, windows, and even clothing.

In a 6G network, RIS turns the propagation environment from a passive obstacle into an active, programmable resource. A signal blocked by a building can be intelligently reflected around it. Indoor coverage can be extended without deploying additional access points. Joint RIS-MIMO optimization is an active research frontier.

Joint Communication and Sensing (JCAS/ISAC)

The 6G base station doubles as a radar. By analyzing the reflections of its own transmitted signals, it can detect objects, track movement, measure distances, and build real-time 3D maps of the environment — all while simultaneously carrying user data. This is Integrated Sensing and Communication (ISAC), and it is already a 3GPP Release 20 study item.

Use cases include gesture recognition for AR interaction, autonomous vehicle navigation assistance, environmental monitoring, intrusion detection, and feeding real-time sensing data into network digital twins. Ericsson demonstrated working ISAC prototypes at MWC 2025.

The 6G Physical Layer: Beyond OFDM

5G NR’s physical layer is built on CP-OFDM and DFT-s-OFDM — waveforms that excel in multipath channels at moderate Doppler spreads. But 6G’s operating conditions — sub-THz frequencies, 1,000 km/h mobility, satellite links with extreme Doppler — push well beyond OFDM’s comfort zone. A new generation of waveforms, coding schemes, and antenna architectures is emerging.

OTFS: Orthogonal Time Frequency Space

OTFS modulates signals in the delay-Doppler domain rather than the time-frequency domain used by OFDM. Each symbol is spread across the entire time-frequency grid via a 2D Fourier-like transform, making every symbol experience the full diversity of the channel. This makes OTFS exceptionally robust against time-varying multipath — exactly the conditions encountered by high-speed trains, LEO satellites, and V2X communication.

Where OFDM sees a rapidly changing channel that requires frequent pilot overhead, OTFS sees a quasi-static delay-Doppler representation that remains stable for much longer. This translates to lower pilot overhead, better spectral efficiency at high mobility, and more reliable detection.

AFDM: Affine Frequency Division Multiplexing

AFDM is a natural generalization of OFDM with two tunable parameters that control the “twistedness” of the affine time-frequency domain. It achieves optimal diversity in doubly-dispersive channels while maintaining strong backward compatibility with existing 4G/5G OFDM infrastructure. Companies within 3GPP have been actively proposing AFDM since October 2025, and it is a tutorial topic at IEEE WCNC 2026.

The key advantage: AFDM can be implemented with minimal modifications to current OFDM processing chains, making it an attractive evolutionary path for operators who want next-generation performance without a complete hardware overhaul.

Holographic MIMO

While 5G uses massive MIMO arrays with 64 or 128 antenna elements at half-wavelength spacing, 6G envisions holographic MIMO — ultra-thin, extremely large surfaces with sub-wavelength-spaced antennas. This approaches a continuous electromagnetic aperture, enabling holographic beamforming with spatial resolution that can separate users mere centimeters apart.

At sub-THz frequencies where wavelengths shrink to sub-millimeter, fitting 1,024+ antenna elements into a compact form factor becomes practical. The result is ultra-massive MIMO with unprecedented multiplexing and beamforming gains — the physics needed to deliver 1 Tbps.

Semantic Communications

For 80 years, wireless communication has followed Shannon’s framework: transmit every bit accurately, regardless of meaning. Semantic communication breaks this paradigm. Instead of transmitting raw bits, the system extracts the meaning of the message at the transmitter and reconstructs it at the receiver using shared AI models.

Consider transmitting a video of a factory floor. Classical communication sends every pixel. Semantic communication understands that the receiver needs to know “machine #7 is overheating” and transmits only that semantic representation — potentially 100× fewer bits. The EU’s 6G-GOALS project is building the framework for this, with three pillars: AI-empowered semantic data representation, timing-aware semantic communication for distributed reasoning, and sustainability via semantic-empowered RAN.

Orbital Angular Momentum (OAM)

OAM exploits the orbital angular momentum property of electromagnetic waves to create orthogonal spatial modes. Each OAM mode (topological charge) provides an independent data channel on the same frequency — a new dimension of multiplexing beyond spatial, time, and frequency. While challenges remain with beam divergence at long distances, OAM is highly promising for short-range, ultra-high-capacity point-to-point links such as backhaul and data center interconnects.

AI/ML at the Physical Layer

3GPP Release 18 initiated the first formal study (TR 38.843) on AI/ML for the NR air interface, focusing on three use cases: CSI feedback enhancement, beam management, and positioning. 6G takes this from “optimization add-on” to native intelligence:

The 6G Core Network: Beyond SBA

The 5G Core introduced the Service-Based Architecture (SBA) — network functions exposed as services with HTTP/2-based APIs. It was revolutionary. But 6G demands a core that is not just service-based but intent-based, AI-native, and compute-converged. TR 38.914 explicitly requires that the 6G RAN design “enable lower CAPEX/OPEX with respect to current networks” and “enable lower energy consumption to achieve sustainability” (Section 5.2). The network becomes a wide-area intelligent fabric.

Intent-Based Networking

Instead of configuring network functions parameter by parameter (as in 5G), operators express high-level business intents: “Ensure 99.999% reliability for the factory automation slice with 100μs latency budget.” An AI-native intent engine translates this into concrete network configuration, resource allocation, slice composition, and policy enforcement — automatically, continuously, and without human intervention.

Network Digital Twins

The Network Digital Twin (NDT) provides a real-time virtual replica of the entire network — RAN, transport, core, edge/cloud, NTN, and applications. Before deploying any change, operators test it on the digital twin first. AI models are trained in the simulated environment. Predictive maintenance anticipates failures before they happen. The O-RAN nGRG is actively researching NDT integration with 6G architecture.

Compute-Network Convergence

This is a fundamental shift identified by the HEXA-X consortium: 6G evolves from a communication-centric platform to a communication-computing-data-centric platform. Network nodes are not just routers and switches — they are compute nodes. AI inference runs at the edge, on-device, and in-network. Task offloading and distributed computing become native network services, replacing the centralized cloud dependency that characterizes today’s architecture.

Native Non-Terrestrial Networks

In 5G (Release 17+), NTN support was added as an overlay — essentially adapting NR to work over satellite links with increased timing advance and Doppler pre-compensation. In 6G, NTN is native. LEO constellations, HAPS, and UAVs are first-class network nodes with unified session management, seamless handover between terrestrial and non-terrestrial segments, and integrated resource scheduling. The target: ubiquitous global coverage including oceans, polar regions, and airspace.

Network Slicing 2.0

6G extends slicing toward sub-slicing (finer granularity within slices), enterprise-operated independent slices with shared authentication, and AI-driven dynamic slice lifecycle management — creation, modification, and termination in real-time. Cross-domain slicing spans terrestrial, NTN, and edge compute segments seamlessly.

AI-Native: Intelligence at Every Layer

If there is one phrase that defines 6G, it is AI-native. Not “AI-enhanced” or “AI-assisted” — those describe 5G Advanced. In 6G, artificial intelligence is embedded in the protocol stack from the physical layer up to the application layer. It is not an optimization tool; it is the network.

Five Levels of AI Integration in RAN

The AI-RAN Alliance (NVIDIA, Samsung, Ericsson, and others) is driving this evolution, building the hardware and software frameworks for running AI inference at RAN timescales — sub-millisecond decisions on scheduling and beamforming, using GPU-accelerated processing at the network edge.

6G Security: Zero-Trust & Post-Quantum

5G security was a significant improvement over 4G — SUPI/SUCI for subscriber privacy, 5G-AKA for mutual authentication, and network slicing isolation. But 6G faces threats that 5G was never designed for: quantum computers capable of breaking current encryption, AI-powered attacks that evolve in real-time, and a vastly expanded attack surface from billions of IoT devices, RIS elements, and satellite links.

Zero-Trust Architecture

6G adopts a zero-trust security model at its foundation. No implicit trust — every network function, device, and user authenticates continuously. Dynamic, contextual trust scoring evaluates risk before every interaction. Research has demonstrated 95–98% trust score accuracy with 22.2% energy efficiency improvement compared to traditional perimeter security.

Post-Quantum Cryptography

3GPP SA3 is studying the integration of Post-Quantum Cryptography (PQC) algorithms to ensure 6G networks resist quantum computing attacks. NIST has already standardized lattice-based (CRYSTALS-Kyber, CRYSTALS-Dilithium) and hash-based (SPHINCS+) algorithms. 6G will need to support these natively — not as a retrofit — because “harvest now, decrypt later” attacks mean that data encrypted today with RSA/ECDSA could be compromised when quantum computers mature.

AI-Enhanced Threat Detection

With AI embedded throughout the network, 6G enables real-time, distributed threat detection. Every network function can flag anomalies. Federated learning models share threat intelligence across operators without sharing raw data. Automated response can isolate compromised nodes within milliseconds — faster than any human SOC team could react.

6G Use Cases: What Becomes Possible

Every generation enables use cases that were physically impossible in the previous one. 3G enabled mobile video. 4G enabled ride-sharing and mobile streaming. 5G enabled industrial IoT and fixed wireless access. 6G unlocks capabilities that sound like science fiction today — but the physics and the standards work are already underway.

Holographic Telepresence

True holographic communication requires transmitting a volumetric light field in real-time: sub-centimeter spatial resolution, sub-degree orientation accuracy, less than 1 ms latency, and data rates approaching 1 Tbps. 5G cannot come close. 6G, with its sub-THz bandwidth, holographic MIMO, and edge computing fabric, makes it achievable for the first time.

Network-as-Sensor (Digital Twin Cities)

With ISAC, every 6G base station continuously senses its environment. Aggregate this across thousands of base stations in a city, and you get a real-time 3D digital twin — tracking traffic flow, monitoring structural integrity of buildings, detecting environmental hazards, and feeding autonomous systems with centimeter-accurate spatial awareness.

Autonomous Swarm Robotics

Collaborative robots that move, sense, and decide as a coordinated swarm require HRLLC (10⁻⁹ reliability), centimeter positioning, and distributed AI inference with microsecond coordination. Think warehouse automation at scale, autonomous construction crews, or search-and-rescue drone swarms operating in disaster zones. 6G’s native compute-network convergence makes the network itself the coordination fabric.

Ubiquitous AI as a Service

6G doesn’t just use AI — it serves AI. Distributed inference at the edge, federated learning across millions of devices, and AI model distribution become native network services. A farmer’s sensor in rural Africa, connected via LEO satellite, can access the same AI inference capabilities as a factory in Seoul — because the intelligence is embedded in the network, not locked in a distant data center.

Preparing for 6G: What to Learn Now

6G commercial networks are still five years away, but the technology decisions being made right now — in 3GPP study items, in O-RAN working groups, in AI-RAN Alliance labs — will determine what 6G looks like. Whether you are a telecom engineer, a researcher, or a student, here is what to invest your learning in today.

Key Industry Research to Follow