Communication Technology

 

 

 

Future of Tele Communication

 

Predicting telecom is easy to do badly. Every generation arrives with a list of applications that mostly do not happen, and then succeeds or fails on something duller that was settled years earlier. So this page separates what is already agreed from what is still only a proposal. The agreed part is smaller than it looks, and it is the part worth knowing.

What actually drives a new generation ?

A new generation is not a marketing decision, whatever the launch events suggest. Each one has answered a specific bottleneck in the generation before it, and that bottleneck was usually visible several years ahead. So the useful question is not what 6G will be called. It is which limit of 5G matters most today.

Let's review the pattern quickly, because it repeats. 1G ran out of capacity, so 2G went digital. 2G could not carry data usefully, so 3G built a packet core. 3G could not deliver the throughput that smartphones suddenly demanded, so 4G rebuilt the air interface around OFDM and IP. 4G could not serve enormous device counts and very low latency at the same time, so 5G split the problem into eMBB, URLLC and mMTC.

Each step solved the previous bottleneck and created a new one. That is the honest way to read the sequence, and it is also why a generation number tells you very little on its own.

So what limits 5G today? Three answers appear repeatedly, and peak throughput is not one of them. Coverage at mid-band and above stays expensive, because propagation does not improve when you buy more spectrum. Energy per bit has become a cost problem rather than an engineering curiosity, since a dense network keeps its radios powered most of the day. And the network still tells the application almost nothing about itself, so the application cannot adapt to it.

  • Each generation answers the previous one's bottleneck : find the limit that costs the most today and you have a reasonable guess at what the next release will work on. Marketing arrives later.
  • Peak rate is the least interesting number : it is the easiest figure to quote and the one that changes least about how a network is built or paid for.
  • The current limits are coverage, energy and awareness : all three are cost problems as much as engineering ones, which is why they shape the next generation more than any application does.

What does ITU mean by IMT-2030 ?

Before 3GPP writes a single specification, ITU-R sets the target. It did that for 5G with the IMT-2020 recommendation, and it did the same for 6G in November 2023 with the IMT-2030 framework. That document is where the six usage scenarios come from, and most 6G presentations are drawn from its diagram.

The change is easiest to see side by side. IMT-2020 had three corners. IMT-2030 has six, and three of those are the earlier three widened and renamed, while the other three did not exist before. Figure 1 puts the two next to each other.

IMT-2020 (5G) IMT-2030 (6G) eMBB URLLC mMTC Immersive Communication Hyper Reliable and Low-Latency Comm. Massive Communication Ubiquitous Connectivity AI and Communication Integrated Sensing and Communication extends an IMT-2020 scenario new in IMT-2030

Figure 1. IMT-2020 defined three usage scenarios and IMT-2030 defines six. Three of the six are the earlier ones widened and renamed. Two of the three genuinely new ones ask the network to do something other than carry data.

IMT-2030 usage scenario

New or extended

What it covers

Immersive Communication

extends eMBB

XR, holographic type communication, and high rate video in both directions rather than mostly downlink.

Hyper Reliable and Low-Latency Communication

extends URLLC

Industrial control, medical applications and safety systems, with tighter bounds than 5G set.

Massive Communication

extends mMTC

Very large numbers of low cost, low power devices, including ones with no battery to speak of.

Ubiquitous Connectivity

new

Coverage where there is none today, mainly through non-terrestrial networks.

AI and Communication

new

The network as a distributed compute and inference resource, not only as a pipe between endpoints.

Integrated Sensing and Communication

new

Using the radio signal to detect and locate objects as well as to carry data.

The framework also names overarching aspects that apply across all six, sustainability and security among them. Those are easy to skip and they are the parts with real design consequences, because a requirement that applies to every scenario constrains every choice made underneath it.

  • Three corners became six : IMT-2030 did not replace the 5G triangle. It widened each of the three corners and added three more beside them.
  • Two of the new three are not about carrying data : AI and Communication asks the network to compute, and Integrated Sensing and Communication asks it to measure the world. Neither is a throughput requirement.
  • The framework sets targets, not designs : ITU-R says what the system has to achieve. 3GPP decides how, and the two documents are often years apart.

Where is 6G in the 3GPP schedule ?

A reader tracking 3GPP contributions needs the release numbers rather than the generation name, because the documents are filed by release. 6G does not appear under a heading called 6G. It appears as study items in Release 20 and as normative work in Release 21.

Release 18 was the first release branded 5G-Advanced, and Releases 19 and 20 continue that line. The 6G study began inside Release 20, which is where the channel model work, the requirements and the early radio studies sit. Release 21 is planned as the first release to carry normative 6G specifications.

That numbering matters in a practical way. If you are searching for 6G TDocs, look in the Release 20 study item folders rather than for anything labelled 6G. Expect the early documents to be RAN1 and RAN2 studies rather than specifications. The first specifications will read the way the early Release 15 NR documents did: incomplete, heavily bracketed, and revised at every meeting.

Be careful with dates. The commonly quoted target is commercial 6G service around 2030, and that number comes from the ITU timeline rather than from any operator commitment. 5G slipped against its own early schedule, and there is no particular reason to expect 6G to run more punctually.

  • Search by release, not by generation : the FTP tree is organised by release and working group. A search for 6G finds presentations, and a search for the Release 20 study items finds the actual work.
  • Release 20 is study and Release 21 is normative : a study item produces a technical report and no requirements. Nothing in a TR obliges anyone to build anything.
  • The 2030 date is a target, not a commitment : treat any deployment forecast attached to it as a forecast, especially one that names a quarter.

Which technical directions are actually being worked on ?

Several directions have enough work behind them to be worth reading about now, and a much longer list does not. The five below all have study items, contributions and measurement campaigns attached to them, which is the practical test of whether a topic has become real yet.

AI and the air interface

Machine learning has been inside base station schedulers for years, so the new part is not AI itself. The new part is putting learned components into the air interface, where both ends have to agree on them. That agreement is the hard problem, because the UE model and the gNB model have to stay compatible across vendors and across software updates. 3GPP studied this in Release 18 for CSI feedback, beam management and positioning, and the difficulty found there was less about accuracy than about how you specify, test and version a model.

Integrated Sensing and Communication

Radio waves reflect off objects, and a receiver that already estimates the channel has most of what a radar needs. ISAC uses that: the same waveform that carries data also reports what it reflected from. The attraction is that the infrastructure already exists. The difficulty is that sensing wants wide bandwidth and line of sight, communication wants coverage and capacity, and one waveform has to serve both. Privacy is a design input here rather than an afterthought, because a network that can locate objects can locate people.

Spectrum in the 7 to 24 GHz range

Millimetre wave delivered the bandwidth 5G promised and not the coverage, so the industry looked for a middle ground. The result is the 7 to 24 GHz range, usually called upper mid-band or FR3. It offers far more bandwidth than the sub-6 GHz bands and it propagates far better than 24 GHz and above. The difficulty is that the range is already occupied. Satellite services, fixed links and government users sit across much of it, so the engineering question arrives attached to a regulatory one.

Non-terrestrial networks

Satellite connectivity used to be a separate industry with separate handsets. Release 17 changed that by bringing NTN into the 3GPP specifications, so the same NR and IoT protocols run over a satellite link. Low earth orbit constellations keep the latency tolerable, and direct-to-handset service has moved from demonstration to commercial offering. For 6G this stops being an extra: the Ubiquitous Connectivity scenario assumes satellite is part of the network rather than a fallback beside it.

Energy as a design constraint

Network energy was a running cost for most of 4G and it has become a design input. A dense 5G network spends a large share of its energy on radios that are transmitting nothing, because a cell has to stay reachable even when idle. Release 18 added network energy saving features for exactly that, cell discontinuous transmission among them. For 6G the target is stronger. Energy per bit has to fall faster than traffic rises, or the network cannot be afforded or powered.

  • A topic is real when it has a study item : that is the cheapest filter available. A conference keynote costs nothing, and a study item means companies have committed engineers to it.
  • Two of these change what the network is for : ISAC and AI both ask the network to do something other than move bits, and that is a larger change than any new waveform.
  • Spectrum decisions outlive technical ones : a band allocation is settled between governments and lasts decades. The radio design around it can be revised in a release.

What is the commercial problem behind all of this ?

One constraint shapes every answer above, and it is not a technical one. Traffic on mobile networks has grown for two decades, and revenue has not grown with it. Until that gap stops widening, it limits what any generation is allowed to cost.

The mechanism is simple enough. An operator sells a flat monthly price, so revenue scales with the number of subscribers. Traffic scales with what those subscribers watch, and video demand has no obvious limit. So the cost of carrying a byte has to fall roughly as fast as the number of bytes rises. Falling that fast keeps the economics level, and it does not improve them. That is why cost per bit, rather than peak rate, decides whether a technology gets deployed.

Several responses have been tried, and they all work on the same side of the equation. Network sharing spreads the cost of the radio network across operators. Open RAN aims to separate hardware from software so that the supplier market widens. Cloud and virtualised deployments move functions onto general purpose hardware. Each of these reduces cost, and none of them creates new revenue.

The revenue side is where the record is poor. Network slicing, edge computing and exposed network APIs have each been offered as the answer, and none has yet produced revenue at the scale the investment needs. This is worth remembering when you read a 6G forecast. The technical claims in one are usually well grounded, and the business case attached to them usually is not.

  • Cost per bit decides deployment : a technology that halves cost per bit will be built even if it adds no feature. One that adds a feature and raises cost usually will not.
  • The cost side has answers and the revenue side does not : sharing, Open RAN and virtualisation are all real. Nothing on the revenue side has worked at the same scale yet.
  • Read the technology and the business case separately : they are written by different people and they age at different rates. The engineering in a 6G paper usually outlives the market forecast beside it.

YouTube

Reference