IMT-2030 (6G) Use Cases and Architectural Framework

IMT-2030 is not just a higher-throughput version of IMT-2020. ITU-R M.2160 expands the 5G service model from three communication scenarios into six usage scenarios and fifteen capability dimensions. The technical shift is from a radio access network optimized mainly for bit transport to a unified Compute, Sensing, and Connect platform where radio, edge compute, positioning, sensing and AI execution share timing, spectrum and control-plane state.

Executive Brief: From IMT-2020 to IMT-2030

IMT-2020 organized 5G around eMBB, URLLC and mMTC. That model worked because the primary engineering problem was still communication: more bits per hertz, lower user-plane latency, higher device density and a more flexible RAN/core split. IMT-2030 keeps those needs but changes the system boundary. The network is expected to carry traffic, sense the radio environment, support distributed AI workloads, expose compute resources close to users and interwork with non-terrestrial access when terrestrial coverage is weak or absent.

The main engineering consequence is that the radio interface can no longer be optimized in isolation. A 0.1-1 ms air-interface target is useful only if queueing, fronthaul, edge compute, policy enforcement and application control loops fit inside the same service budget. A 1-10 cm positioning target is useful only if timing, synchronization, multipath modeling and sensor fusion are reliable at deployment scale. A 50-200 Gbit/s peak data rate is useful only where channel bandwidth, beam management, RF power, thermal design and blockage conditions support it.

Architectural constraint: IMT-2030 KPIs are not additive. A single site will not simultaneously deliver 200 Gbit/s peak rate, 0.1 ms latency, 10^8 devices/km2, 1 cm positioning and wide-area coverage. Practical design selects a KPI envelope per scenario, band, cell radius, mobility state and compute location.

The Six Core Usage Scenarios

ITU-R M.2160 splits IMT-2030 into three evolved 5G pillars and three native 6G pillars. The evolved group extends the 5G triangle. The native group introduces functions that previous IMT generations did not treat as core radio-system capabilities.

Evolved 5G Pillars

1. Immersive Communication: Evolution of eMBB

Technical context and objectives

Immersive communication extends eMBB from high-rate video delivery to synchronized multi-stream media and interaction. The relevant IMT-2030 capability envelope includes 50, 100 and 200 Gbit/s peak data rate examples, 300-500 Mbit/s user-experienced data rate examples, 30-50 Mbit/s/m2 area traffic capacity, higher spectrum efficiency than IMT-2020 and low enough latency to maintain interaction between physical and rendered objects.

The problem is not just throughput. A holographic or high-resolution XR session needs bounded jitter, synchronized audio/video/haptic streams, stable uplink capacity for spatial capture, edge rendering close to the user and fast recovery from beam blockage. For dense venues, capacity must be engineered per square meter, not only per cell.

Concrete enterprise use case

A multinational engineering firm deploys multi-user holographic telepresence across design centers. Each room has volumetric capture rigs, spatial audio, haptic work surfaces and a local edge GPU cluster. Users see life-size remote participants and manipulate a shared CAD assembly. The radio system provides high-rate uplink from room sensors, high-rate downlink for rendered views and local breakout to an edge application server in the same metro region. Time synchronization is distributed to capture devices, RAN nodes and rendering servers so that audio, body pose, object state and display frames remain aligned.

Engineering bottlenecks

• Sub-THz and mmWave coverage: 50-200 Gbit/s requires very wide channels and high-order spatial multiplexing. Above 92 GHz, ITU notes high attenuation, limited transmit power and blockage as major operating challenges. Indoor hot-spots and short-range backhaul are plausible early targets; wide-area mobility is not.
• Uplink asymmetry: Immersive capture can invert the normal consumer traffic model. Multiple cameras, depth sensors and haptic devices can require sustained uplink capacity that is difficult under TDD patterns optimized for downlink.
• Beam management: Head movement, body blockage and room reflection changes can cause fast channel variation. Beam recovery must be faster than the application's frame budget.
• Compute-offload budget: Edge rendering must include radio transmission, fronthaul/backhaul, queueing, GPU inference/rendering and return path. Moving rendering to a regional cloud may break interaction latency even when the air interface is fast.

Physics constraint: Sub-THz bandwidth improves peak rate and sensing resolution, but the link budget loses margin through molecular absorption, rain loss, blockage, RF front-end noise figure, power amplifier efficiency and beam-pointing error. The deployment unit becomes a room, corridor, factory cell or hot-spot zone, not a conventional macro sector.

2. Hyper-Reliable and Low-Latency Communication: Evolution of URLLC

Technical context and objectives

HRLLC extends URLLC toward stricter failure probabilities and shorter control loops. IMT-2030 gives an air-interface latency research target of 0.1-1 ms and reliability from 1-10^-5 to 1-10^-7. IMT-2020 URLLC established a 1 ms user-plane latency reference and a 1-10^-5 reliability requirement for small packets under specified test conditions. IMT-2030 pushes that envelope for time-synchronized operations where a missed packet can stop a production line, destabilize a robot cell or create a safety event.

Concrete enterprise use case

A semiconductor fab uses wireless closed-loop control for mobile manipulators, inspection robots and automated material handling. The controller sends motion commands at deterministic intervals. Robots return encoder, force, vision and safety-state data. The private 6G system reserves configured grants for uplink telemetry, uses packet duplication across two TRxPs, anchors the user plane at an on-premises UPF and keeps control applications in a local edge cluster. The scheduler gives safety traffic higher priority than video diagnostics and uses admission control to prevent overload.

Engineering bottlenecks

• End-to-end timing: A 0.1 ms air-interface target leaves almost no budget for transport and compute. The DU, CU-UP, UPF and control application may need to be campus-local.
• Reliability measurement: 10^-7 packet failure probability cannot be proven with a short drive test. Operators need statistical acceptance plans, synthetic load, fault injection and continuous monitoring by traffic class.
• Interference determinism: Unlicensed or shared spectrum can be unacceptable for hard control unless coexistence and priority mechanisms are enforceable.
• Redundancy cost: Multi-TRxP transmission, packet duplication, dual connectivity and redundant edge nodes improve reliability but consume spectrum, power and compute headroom.

Timing constraint: Light in fiber travels roughly 1 km in 5 microseconds one way. A 20 km fiber path consumes about 100 microseconds before queueing, protocol processing or application execution. HRLLC control loops therefore drive DU/CU-UP/UPF and application placement toward the factory, port, hospital or campus edge.

3. Massive Communication: Evolution of mMTC

Technical context and objectives

Massive communication extends mMTC from low-rate sensor access to much broader device classes. IMT-2030 gives a connection-density target range of 10^6-10^8 devices/km2. The usage scenario includes batteryless or long-life devices, intermittent sensors, logistics tags, industrial meters, agriculture nodes, city infrastructure and mobile asset trackers. The key capabilities are connection density, low power consumption, extended coverage, security, reliability and operational cost per device.

Concrete enterprise use case

A national logistics operator instruments containers, pallets, cold-chain packages, yard vehicles and warehouse shelves. Outdoor macro coverage handles wide-area tracking. Private indoor cells handle dense warehouse inventory. Passive or energy-harvesting tags wake only when scanned or scheduled. The network aggregates tiny uplink messages, deduplicates repeated reports at the edge and exposes event streams to the warehouse management system. Devices with temperature excursions or security alarms are moved from low-duty-cycle reporting to a higher-priority bearer.

Engineering bottlenecks

• Random access overload: Device density targets are meaningless if attach and access procedures collapse during synchronized wake-up events after power restoration, shift changes or disasters.
• Battery and firmware lifecycle: A ten-year sensor life requires strict control of paging, measurement reporting, retransmission behavior, security update size and roaming behavior.
• Identity and key management: Millions of low-cost devices create an operational security problem. Credential rotation, compromised device quarantine and supply-chain provenance become network functions.
• Coverage versus capacity: Deep indoor and rural coverage often requires lower bands and narrow bandwidth; extreme density in factories may require dense small cells and local aggregation.

Native 6G Pillars

4. Integrated Sensing and Communication (ISAC)

Technical context and objectives

ISAC is a new IMT-2030 usage scenario. The radio interface is expected to support range, velocity and angle estimation, object detection, presence detection, localization, imaging and mapping while also providing communication. M.2160 does not set one universal sensing-resolution number; it identifies measurement dimensions such as accuracy, resolution, detection rate and false alarm rate. Positioning, a separate IMT-2030 capability, has a research target of 1-10 cm.

Sensing performance depends strongly on bandwidth, carrier frequency, aperture, waveform, SNR, coherent processing interval and environment. A simple range-resolution approximation is c/(2B): about 15 cm with 1 GHz effective bandwidth and about 1.5 cm with 10 GHz. Angle resolution depends on array aperture and beamwidth, not only on bandwidth.

Concrete enterprise use case

A container port deploys ISAC-capable private 6G cells across cranes, truck lanes and storage blocks. The network tracks vehicles, pedestrians, containers and crane hooks even when some objects have no active modem. Connected autonomous yard trucks receive communication service and positioning assistance; unconnected workers and obstacles are detected through radio sensing. The port operations platform fuses ISAC outputs with cameras, lidar and TOS data to enforce speed limits, geofences and anti-collision rules.

Engineering bottlenecks

• Waveform tradeoff: A waveform optimized for communication throughput may not provide the ambiguity function, bandwidth, sidelobe behavior or coherent processing time needed for sensing.
• False alarms and missed detections: Sensing KPIs must define probability of detection, false alarm rate, range/velocity/angle resolution and update periodicity. A safety system cannot rely on a vague object-detection claim.
• RF coexistence: ISAC transmissions can resemble radar-like operation. Coexistence with incumbent radar, passive sensing services, fixed links and existing mobile networks requires regulatory and interference engineering.
• Privacy and data governance: Sensing can detect unconnected objects and human motion. Enterprise deployments need retention, anonymization, lawful-use controls and clear separation between safety data and surveillance data.

5. Artificial Intelligence and Communication (AIAC)

Technical context and objectives

AIAC covers two directions. First, the network supports AI workloads through distributed data processing, distributed learning, model sharing, inference, training support and compute orchestration. Second, AI functions are used inside the communication system for channel estimation, beam selection, mobility prediction, energy control, anomaly detection and traffic engineering. M.2160 names these AI-related capabilities but does not assign one TOPS, FLOPS or model-size threshold.

The useful engineering metrics are therefore local: inference latency per function, model size, update frequency, training data volume, feature freshness, accelerator utilization, power per inference, rollback time, explainability for operations and accuracy under radio drift.

Concrete enterprise use case

A railway operator runs AI-assisted predictive control for high-speed train connectivity. Trackside RAN nodes, train modems and edge servers share channel, location and blockage features. A model predicts beam transitions, serving-cell changes and link degradation along tunnels, cuttings and stations. The model runs inside or near the DU for time-critical beam decisions, while a non-real-time training pipeline updates regional models using logged RF data. The core exposes train service class, route schedule and SLA policy to the edge orchestrator.

Engineering bottlenecks

• Where inference runs: Near-real-time RIC control loops are not fast enough for every PHY/MAC decision. Sub-slot beam or link adaptation models may need to execute in the DU, RU or device.
• Model lifecycle: A model trained on one city, band, antenna layout or season may fail elsewhere. Operators need versioning, monitoring, rollback and drift detection.
• Data movement cost: Uploading raw I/Q, CSI or sensing data to a central cloud can consume more transport capacity than the user service. Feature extraction must often happen at the edge.
• Operational accountability: A black-box scheduling decision that drops emergency, robotic or medical traffic is unacceptable. Policies must constrain AI actions and expose audit records.

6. Ubiquitous Connectivity: NTN and Coverage Continuity

Technical context and objectives

Ubiquitous connectivity aims to connect uncovered or sparsely covered areas and provide service continuity through interworking with other systems. M.2160 explicitly includes interworking with non-terrestrial networks such as satellite systems, high-altitude platform stations as IMT base stations, unmanned aircraft systems and non-IMT terrestrial access. The objective is not to make every satellite link match an urban small cell. The objective is continuity of service with predictable capability by access type.

Concrete enterprise use case

A mining company operates autonomous haulage across a remote site. Terrestrial private 6G covers pits, crushers and maintenance yards. LEO satellite access covers haul roads, exploration areas and emergency response zones beyond terrestrial coverage. Vehicles maintain a single service identity through the 5G/6G core. When terrestrial RSRP and predicted path quality fall below threshold, the modem prepares NTN access before losing the macro cell. Safety telemetry and command channels continue through the satellite path, while high-rate video waits for terrestrial coverage or local edge upload.

Engineering bottlenecks

• Doppler and delay: LEO satellites create time-varying Doppler and propagation delay; GEO links add much larger delay. Timing advance, frequency compensation and HARQ behavior must be adapted by orbit and payload type.
• Session handover: A seamless user session requires make-before-break policy, common identity handling, suitable UPF anchoring and application awareness of changing RTT and throughput.
• Link budget: Handheld or vehicle terminals face antenna gain, power and obstruction limits. Service classes must map to what the satellite link can actually support.
• Regulatory and roaming complexity: Spectrum rights, landing rights, lawful intercept, emergency service routing and data residency vary across countries and satellite operators.

Technical Capabilities Matrix

The table below tracks the fifteen IMT-2030 capability dimensions and contrasts them with the IMT-2020 baseline. Values are from ITU-R M.2160 for IMT-2030, ITU-R M.2083 for the IMT-2020 vision and ITU-R M.2410 where the 5G minimum requirement is more specific.

Capability constraint: Sensing resolution, AI-compute capacity and sustainability are not single universal numbers in M.2160. They become testable only when the deployment defines bandwidth, aperture, sensing target class, model workload, accelerator location, carbon boundary and service availability target.

Architectural Implementation Challenges

Spectrum Realities: sub-6 GHz, mmWave and sub-THz Must Coexist

IMT-2030 will not replace existing spectrum layers. It will stack new capability on top of deployed low-band, mid-band and mmWave assets. Sub-1 GHz and lower mid-band remain essential for coverage, mobility and indoor reach. Upper mid-band and FR3-type spectrum can provide a more balanced coverage/capacity layer if regulators make it available and incumbents can be protected. mmWave remains valuable for hot-spots, fixed wireless, private campuses and transport nodes. Bands above 92 GHz provide wide bandwidth for hot-spots, ISAC, sidelink, backhaul and fronthaul, but only where link budgets close.

The coexistence problem is operational as much as regulatory. Operators must coordinate TDD patterns to reduce cross-link interference, manage intermodulation and emissions near passive services, protect fixed links and radar systems, and decide which KPI belongs on which band. HRLLC control should not depend on a fragile blocked sub-THz link unless a lower-band redundant path exists. ISAC sensing may need different waveform occupancy than broadband user traffic. NTN spectrum adds another layer of cross-border and satellite coordination.

O-RAN, DU/CU and Core Evolution

O-RAN-style disaggregation helps vendor diversity and automation, but IMT-2030 pushes more functions into tight timing loops. Non-real-time rApps can optimize policy, planning, energy states and model training over minutes or hours. Near-real-time xApps can support mobility, interference and traffic optimization over roughly control-loop timescales. PHY/MAC decisions for HRLLC, beam selection, ISAC snapshots and some AI-assisted radio functions may need to run inside the O-DU, O-RU or UE because the control loop is shorter than the RIC path.

The CU/DU split also becomes workload-dependent. A campus HRLLC deployment may need O-DU, O-CU-UP, UPF and industrial control applications on-premises. An immersive media deployment may place rendering and media gateways in a metro edge. Massive communication may centralize more functions but keep access-control and aggregation logic close to device clusters. AIAC introduces model registries, feature stores, inference endpoints and accelerator pools as network resources that policy control must understand.

The core network must evolve from session anchoring and policy enforcement to compute-aware service continuity. It has to select UPF and edge locations based on latency, energy and workload requirements; expose network state to applications safely; coordinate slices with accelerator availability; and support identity continuity when access moves between terrestrial and NTN links.

NTN Integration and gNB-to-LEO Session Handover

Seamless terrestrial-to-NTN operation is harder than adding a satellite modem. The system must handle different cell motion models, round-trip times, Doppler, ephemeris data, beam footprints, gateway routing and regulatory domains. In a transparent payload architecture, more RAN processing stays on the ground and the satellite behaves closer to a relay. In a regenerative payload architecture, gNB functions can move into the satellite, reducing some gateway dependency but increasing payload complexity and lifecycle constraints.

For session continuity, the UE should not wait for terrestrial radio failure. The network needs predictive access selection using location, route, satellite visibility, service class and traffic state. A safety bearer may move to NTN before broadband traffic. PDCP duplication or application-layer replication may be used for critical flows during transition. The UPF anchor should be selected to avoid unnecessary tromboning, but moving anchors too often can break application assumptions. Enterprises should define NTN service classes explicitly: emergency text, low-rate telemetry, command/control fallback, voice, store-and-forward data or broadband.

Handover constraint: A LEO satellite link may preserve a session while changing RTT, Doppler, available throughput and packet-loss behavior. Applications that assume a terrestrial latency profile will fail even if IP connectivity remains up. IMT-2030 service continuity needs application-aware policy, not only radio handover.

Deployment Mechanics for Enterprise Networks

Enterprise 6G planning should begin with a KPI envelope, not a generation label. A port, hospital, factory, mine or rail corridor should classify flows into broadband media, deterministic control, sparse telemetry, sensing feeds, AI model traffic and emergency continuity. Each class needs a latency, reliability, bandwidth, positioning, sensing, security and energy target. The radio design then maps those classes to bands, TRxP density, redundancy, edge compute placement and core anchoring.

Three practical rules follow. First, do not put all service classes on the highest frequency layer; wide bandwidth does not replace coverage and reliability. Second, treat compute as part of the radio design; an edge server 50 km away can consume more latency budget than the air interface saves. Third, make sensing and AI observable; operators need counters for detection probability, false alarms, model drift, inference latency, accelerator saturation and energy per delivered service.

Primary Standards References

• ITU-R Recommendation M.2160-0 - Framework and overall objectives of the future development of IMT for 2030 and beyond.
• ITU-R Recommendation M.2083-0 - IMT Vision for 2020 and beyond.
• ITU-R Report M.2410-0 - Minimum technical performance requirements for IMT-2020 radio interfaces.

Frequently asked questions