Chapter 2. Need of 5G and IMT Service Requirements
Updated: 3 days ago
Chapter 2. Need of 5G and IMT Service Requirements
2.1 What You will learn Here
Figure 2.1: The six building blocks of this chapter, ordered from requirement to deployment.
What you will be able to explain by the end
How a mobile generation moves from an ITU requirement to a live operator network.
All eight IMT-2020 capability dimensions and what each one actually measures.
Why eMBB, URLLC and mMTC exist as separate categories rather than one specification.
The difference between a standards target and an operational KPI, and why confusing them causes bad decisions.
Which category each enterprise and consumer use case actually depends on.
Remember Requirements come first, capabilities turn them into numbers, categories group the numbers into network behaviours. Keep that order and the chapter stays simple. |
2.2 Who Actually Sets the 5G Requirements
Figure 2.2: The four stage chain from ITU vision to live network.
A common misconception is that 3GPP invented the 5G targets. It did not. The chain has four stages, and each one does a different job.
Stage one: the ITU states what must be achieved
The ITU Radiocommunication Sector published the IMT-2020 vision in Recommendation ITU-R M.2083, which introduced the three usage scenarios. It then published the minimum technical performance requirements in Report ITU-R M.2410, which is where the specific numbers live. Neither document explains how to build anything.
Specification reference: ITU-R Recommendation M.2083, IMT Vision for 2020 and beyond
Stage two: 3GPP specifies how
3GPP then wrote the specifications that meet those requirements. 5G New Radio and the 5G System both arrived in Release 15, with the system architecture in TS 23.501. Every design choice in that specification is an engineering answer to a number the ITU published first.
Stages three and four: vendors build, operators live with it
Vendors build radio units, core software and devices against the 3GPP specifications and are tested for conformance. Operators then deploy that equipment and are measured every day on live key performance indicators, which are a different thing entirely from the ITU targets. Section 2.6 returns to that point.
Common Misconception The twenty gigabit peak rate is often quoted as a 3GPP figure. It is an ITU requirement from Report ITU-R M.2410. 3GPP designed New Radio to be capable of meeting it. |
2.3 The Eight IMT-2020 Capability Targets
Figure 2.3: The eight measurable capability dimensions defined in Report ITU-R M.2410.
These eight dimensions turn the vision into engineering. Treat every figure as a specification target measured under defined conditions, not as something you will see on a live cell.Q
Capability | IMT-2020 target | What it actually measures |
Peak data rate | 20 Gbps downlink, 10 Gbps uplink | Theoretical maximum under ideal conditions |
User experienced data rate | 100 Mbps downlink, 50 Mbps uplink | The realistic rate a subscriber should see |
User plane latency | 4 ms for eMBB, 1 ms for URLLC | One way delay over the radio interface |
Connection density | 1 million devices per square km | Devices supported in a dense area |
Mobility | Up to 500 km per hour | Speed at which the link stays usable |
Peak spectral efficiency | 30 bit/s/Hz downlink, 15 uplink | Bits carried per hertz of spectrum |
Area traffic capacity | 10 Mbit/s per square metre | Total throughput in an indoor hotspot |
Reliability | 99.999 percent | 32 byte packet delivered within 1 ms |
Why no network meets all eight at once
Peak data rate wants wide bandwidth, high order modulation and large transport blocks. One millisecond latency wants short transmission slots and minimal processing. Connection density wants cheap radios that sleep for hours. These pull the design in different directions.
This tension is the entire reason 5G splits into service categories. Instead of one compromise specification, the standard defines three behaviour profiles and lets the operator choose which one a given slice delivers.
Important When someone quotes a 5G figure, ask which of the eight dimensions it belongs to and under which category. A peak rate and a URLLC latency figure describe two different networks. |
2.4 Three Service Categories, One Network
Figure 2.4: The three IMT-2020 usage scenarios and what each one prioritises.
Recommendation ITU-R M.2083 groups everything 5G must do into three usage scenarios. Learn the names properly, because the rest of this course keeps returning to them.
Category | Full name | Prioritises | Typical traffic shape |
eMBB | Enhanced Mobile Broadband | Bandwidth and capacity | Large sustained flows, video dominated |
URLLC | Ultra Reliable Low Latency Communication | Latency and reliability | Very small packets, strict deadlines |
mMTC | Massive Machine Type Communication | Device count and battery life | Tiny infrequent reports, long sleep periods |
The important idea is that these are not three networks. They are three behaviours the same infrastructure must produce on demand. A single enterprise customer often needs two of them at once, which a shared best effort network cannot provide.
Key Takeaway The requirement to deliver all three behaviours from one infrastructure is precisely the argument for network slicing, which Chapter 13 covers in full. |
2.5 Inside the Three Categories
Figure 2.5: The three deployment layers that make up enhanced Mobile Broadband.
eMBB, the category that pays the bills
Operators deployed eMBB first because it monetises immediately against existing subscribers. It deploys in three layers. Sub-6 gigahertz, which 3GPP calls FR1, provides the coverage layer with channels around 100 megahertz. Mid band with massive MIMO and beamforming provides urban capacity. Millimetre wave, called FR2, provides hotspot capacity with carriers up to 400 megahertz, but with very short range and poor building penetration.
Typical applications are high resolution video, fixed wireless access as a fibre substitute, immersive media, and simply keeping service usable at crowded venues.
Figure 2.6: The URLLC latency budget, stage by stage.
URLLC, where late equals wrong
URLLC is the category that reshapes architecture. Walk the budget and you see how little room exists. Device processing, radio transmission, transport and user plane processing all consume time before the application even responds.
Two definitions matter here and learners confuse them constantly. The one millisecond figure is a user plane latency target measured one way over the radio interface, not an end to end application figure. Reliability is defined precisely as 99.999 percent for a 32 byte packet delivered within 1 millisecond.
Notice the packet size. A robot control message is a few dozen bytes, so bandwidth is never the challenge. The challenge is that a late message is a failed message, which is why URLLC forces the User Plane Function out to the edge.
Did You Know? Distance is physics. Light travels roughly 200 kilometres per millisecond in fibre, and that is before any equipment adds processing delay. A central user plane anchor simply cannot serve a one millisecond budget. |
Figure 2.7: mMTC signalling behaviour compared with a smartphone.
mMTC, a million devices that barely speak
mMTC optimises for the opposite of eMBB. Device count, module cost and battery life dominate, and throughput is almost irrelevant. Battery targets around ten years are achieved through power saving mode and extended discontinuous reception, so the device sleeps almost all of the time.
The engineering challenge sits in the control plane rather than the user plane. A smartphone holds long sessions and hands over frequently. A smart meter sends a few bytes twice a day. The problem is millions of registrations arriving in the same window, which is a signalling problem, not a throughput problem.
Important Most machine type traffic today runs on LTE-M and NB-IoT, which 3GPP continues to evolve inside the 5G framework. Reduced Capability, known as RedCap, arrived in Release 17 for devices between those and full 5G. |
2.6 From ITU Targets to Operational KPIs
Figure 2.8: The six KPI families an operator watches on a live network.
Standards give you targets. Operations give you counters. These are not the same thing, and quoting one when you mean the other leads to bad expectations and worse contracts.
KPI family | What is measured | Typical target | The question it answers |
Accessibility | Registration success rate | Above 99.5 percent | Can the device get on the network |
Retainability | Session drop rate | Below 0.5 percent | Does the session survive |
Integrity | User experienced throughput | Above 100 Mbps | Is the service good enough |
Latency | Round trip time | Service dependent | Does it feel instant |
Mobility | Handover success rate | Above 99 percent | Does it survive movement |
Availability | Cell and slice uptime | Up to 99.999 percent | Is it there when needed |
In 5G, availability becomes commercially interesting. An enterprise slice can carry a contractual uptime figure, which means a KPI stops being an internal engineering metric and becomes a term in an agreement.
Common Misconception An ITU target is a laboratory maximum under ideal conditions. A KPI is what your customer experiences on a Tuesday evening under load. Never quote one when you mean the other. |
2.7 Enterprise Use Cases and the Category They Need
Figure 2.9: Six enterprise use cases mapped to the service category each one depends on.
This is where the theory becomes a purchase order. Each industry brings a requirement, and that requirement maps to a category and then to an architecture decision.
Sector | What they deploy | Category needed |
Smart manufacturing | Motion control, machine vision, guided vehicles | URLLC with a dedicated slice and on site user plane |
Connected healthcare | Remote monitoring, connected imaging | URLLC for monitoring, eMBB for imaging |
Ports and logistics | Remote crane control, autonomous yard vehicles | URLLC, where a delayed stop command is a safety event |
Utilities and metering | Grid telemetry, millions of smart meters | mMTC, dominated by scale and battery life |
Transport and automotive | Vehicle to everything, fleet telemetry | URLLC for safety, mMTC for telemetry |
Live media production | Remote cameras, contribution feeds | eMBB with sustained uplink |
Look at how often one customer needs two categories. A hospital needs reliability for monitoring and bandwidth for imaging. A vehicle fleet needs URLLC for safety messaging and mMTC for telemetry. That combination is exactly what a slice delivers.
2.8 Consumer Use Cases, Where the Traffic Lives
Figure 2.10: Six consumer services and the category each one leans on.
Enterprise services carry the premium pricing, but consumer services still generate the volume. Video streaming remains the single largest traffic source, growing through higher resolutions and more screens per household.
Fixed wireless access deserves attention. It substitutes for fibre where civil works are impractical and has become one of the strongest 5G revenue lines globally, using pure eMBB capability.
The interesting case is cloud gaming
For cloud gaming the throughput is easy and the responsiveness is hard. A speed test looks excellent while the user complains about lag. This pushes the user plane toward the edge even for a consumer service, which is the same architectural conclusion URLLC reached from a very different starting point.
Key Takeaway Wearables and home sensors behave like mMTC even inside a consumer package. The category describes the traffic behaviour, not the customer type. |
2.9 LTE Capabilities Compared With 5G
Figure 2.11: The same capability dimensions, measured against LTE-Advanced and IMT-2020.
Use this as a revision sheet. Every figure is a specification target under defined conditions, not a live measurement.
Capability | 4G LTE-Advanced | 5G IMT-2020 |
Peak data rate | 1 Gbps downlink | 20 Gbps downlink |
User experienced rate | Around 10 Mbps | 100 Mbps downlink |
User plane latency | Around 10 ms | 4 ms eMBB, 1 ms URLLC |
Connection density | Around 100,000 per square km | 1 million per square km |
Mobility | Up to 350 km per hour | Up to 500 km per hour |
Peak spectral efficiency | Around 15 bit/s/Hz | 30 bit/s/Hz downlink |
Service model | One best effort network | Slices per service category |
The last row matters most. LTE offered one best effort network to everyone. 5G offers slices tuned per service category. That row, not the throughput figures, is the real generational change, and it is the thread that runs through the rest of this course.
Remember If you remember one comparison from this chapter, make it the service model row. Everything else is an increment. That row is a change of kind. |
2.10 Key Takeaways
Figure 2.12: The five points to carry forward.
The ITU sets the bar and 3GPP builds to it. The IMT-2020 targets come from Report ITU-R M.2410.
There are eight capability dimensions, and no single deployment maximises all of them at once.
The three categories describe three behaviours. eMBB for bandwidth, URLLC for reliability, mMTC for scale.
Standards targets and operational KPIs are different things and should never be quoted interchangeably.
One infrastructure must deliver all three behaviours on demand, which is precisely why network slicing exists.
Chapter 3 takes the next step. It covers the 3GPP release timeline and the Non Standalone and Standalone deployment options that every operator had to choose between.
2.11 Frequently Asked Questions
Is 20 Gbps something my phone will ever reach?
No. That figure is a peak data rate under ideal laboratory conditions with maximum bandwidth, the best possible signal and no other users. The number to watch is the user experienced data rate, which IMT-2020 sets at 100 megabits per second downlink. That is the figure that describes a real subscriber.
What is the difference between M.2083 and M.2410?
M.2083 is a Recommendation that sets out the IMT-2020 vision, including the three usage scenarios. M.2410 is a Report that defines the minimum technical performance requirements, which is where the eight measurable targets live. Vision first, numbers second.
Why does 5G need three categories instead of one specification?
Because the targets conflict. Peak data rate wants wide bandwidth and large transport blocks, while one millisecond latency wants short slots and minimal processing, and massive device counts want cheap radios that sleep. One compromise specification would serve none of them well.
Is eMBB just faster 4G?
In capability terms it is the natural continuation of what LTE already did well, and that is why operators launched with it first. What differs is the scale of the increase and the fact that eMBB now shares infrastructure with two other categories that have completely different requirements.
Is the 1 ms URLLC figure end to end?
No, and this is the most common misreading in the chapter. It is a user plane latency target measured one way over the radio interface. Transport, core processing and the application response all sit outside that figure, which is why real end to end budgets are considerably larger.
What exactly does 99.999 percent reliability mean here?
It is not a general uptime figure. Report ITU-R M.2410 defines it as the probability of successfully delivering a 32 byte layer 2 packet within 1 millisecond. Both the packet size and the deadline are part of the definition, so quoting the percentage alone is meaningless.
Why is connection density measured per square kilometre?
Because the constraint is local, not national. The difficulty is serving an enormous number of devices inside the coverage of a small number of cells, where they compete for the same random access resources and signalling capacity. A national device count would hide that problem.
Does mMTC actually run on 5G New Radio today?
Mostly not. The majority of machine type traffic runs on LTE-M and NB-IoT, which 3GPP continues to evolve inside the 5G framework and which satisfy the mMTC requirements. RedCap, introduced in Release 17, targets the middle ground between those technologies and full 5G devices.
How does 5G achieve lower latency on the radio interface?
Mainly through flexible numerology and shorter transmission slots. Larger subcarrier spacing shortens the slot duration, so a transmission opportunity arrives sooner. Mini slot scheduling allows a transmission to start without waiting for a slot boundary. Chapter 5 covers the mechanism in detail.
Can one slice deliver both eMBB and URLLC?
It can be configured to try, but the result is usually a compromise that satisfies neither well. The normal approach is separate slices with different QoS profiles and different user plane placement, since URLLC typically needs an edge UPF while eMBB does not.
Why is area traffic capacity expressed per square metre?
Because it describes the dense indoor hotspot scenario, where the design question is how much total throughput a small floor area can sustain. Expressing it per square metre makes it directly comparable across venue sizes and independent of the cell radius.
What does mobility interruption time of zero milliseconds mean in practice?
It is a requirement that a handover should not produce a gap in user plane transmission. In practice it is approached using techniques such as dual active protocol stack handover, where the device maintains a connection to both cells briefly rather than releasing one before adding the other.
Scenario-Based Questions
A factory asks for guaranteed one millisecond response. What do you check first?
Check whether they mean radio interface latency or application response time, because those differ substantially. Then check where the user plane will terminate, since a central UPF makes the budget unachievable regardless of radio performance. Only then discuss slicing and QoS configuration.
A utility wants to connect 800,000 meters across a city. Which capability governs the design?
Connection density and signalling capacity, not throughput. Total data volume from 800,000 meters reporting twice a day is trivial. The design risk is the registration and random access load, particularly after a power event when large numbers reconnect at once.
A broadcaster wants to replace fibre for live camera feeds. Which category and why?
This is eMBB, but with an unusual emphasis. The demanding direction is uplink rather than downlink, and jitter matters as much as raw throughput. Scheduled or dedicated capacity is often required, since best effort uplink at a busy venue will not hold a contribution feed.
Troubleshooting Based Questions
Users report poor 5G experience although peak throughput tests pass. Where do you look?
Peak throughput tests measure the wrong dimension. Look at user experienced data rate under load, cell edge performance, and latency. A cell that hits high peak rates for a single idle user can still deliver poor experienced rates once loaded.
An enterprise slice misses its availability KPI but no alarms fired. What now?
Check what the availability figure was defined against. Availability can be measured per cell, per slice, or per service, and those give different results. A slice can be technically up while a specific network function instance serving it was degraded, which cell level counters will not show.
Architecture-Based Questions
Where do the service categories actually appear in the architecture?
They appear as slice and QoS configuration rather than as separate hardware. A slice is identified by an S-NSSAI whose slice service type indicates eMBB, URLLC or mMTC. Within a session, treatment is applied through QoS flows identified by a 5QI. Chapters 10 and 13 cover both mechanisms.
Does supporting URLLC require changes outside the radio?
Yes, and that is the point of the category. It requires the User Plane Function to be placed close to the user, deterministic transport between the radio and that user plane, and redundancy so that a single failure does not break the reliability target. The radio alone cannot deliver URLLC.
How do the eight capabilities map onto the three categories?
Not evenly, and that is deliberate. eMBB is measured mainly on peak rate, user experienced rate, spectral efficiency and area traffic capacity. URLLC is measured on user plane latency and reliability. mMTC is measured on connection density, with battery life and cost as commercial constraints rather than ITU targets.
Standards Referenced in This Chapter
Document | Subject |
ITU-R M.2083 | IMT Vision for 2020 and beyond, defining eMBB, URLLC and mMTC |
ITU-R M.2410 | Minimum requirements related to technical performance for IMT-2020 |
3GPP TR 38.913 | Study on scenarios and requirements for next generation access technologies |
3GPP TS 23.501 | System architecture for the 5G System, including slicing and QoS |
3GPP TS 22.261 | Service requirements for the 5G system, stage 1 |
3GPP Release 17 | Introduces Reduced Capability, RedCap, for mid tier devices |
The full 3GPP specification archive is available at 3gpp.org specifications. Always check the release version that matches your deployment.
End of Chapter 2. Need of 5G and IMT Service Requirements
5G Network Architecture Masterclass
------------------------
Learn Telecom. Crack Interviews. Build Careers.
support@telecomhunt.com | (c) 2026 Telecom Hunt, All rights reserved.
Comments