AI’S IMPACT ON THE

RADIO ACCESS NETWORK

Published September 2024

Brought to you by Sponsored by

AI’S IMPACT ON THE RAN REPORT

INTRODUCTION

Ray Le Maistre
Editorial Director,
TelecomTV

Welcome to the latest DSP Leaders Report, which focuses on one of the key areas of
mobile network development – the use of artificial intelligence (AI) in the radio access
network (RAN) and its impact.

There’s no doubt that the topic of how AI could and should be used by network
operators has been high on the list of industry discussions for many years already, even
before the ramp up of interest in network automation that began about a decade ago,
fuelled in turn by the development of software-defined networking (SDN) and network
function virtualisation (NFV) tools.

But it has also been bubbling up to the top of the strategic technologies list in recent
years with the deployment of 5G, the development of open network systems (including
Open RAN) and the increasing cloudification of communications network platforms. And
then, of course, generative AI (GenAI) became the topic de jour in late 2022 and all of a
sudden it was hard to have a conversation in telecom (or, indeed, just about any industry
vertical) without AI cropping up.

But GenAI doesn’t feature in this report: Rather, the focus here and in the survey we
made available to the mobile network operator (MNO) community, is all about the use
of AI tools in the management of the RAN, where the industry is right now in terms of
those developments and how the MNOs see it evolving.

www.telecomtv.com 2
AI’S IMPACT ON THE RAN REPORT

METHODOLOGY

We published a survey for network operator staff in July 2024. The questions were
devised solely by the TelecomTV editorial team. We received responses from 58
individuals who work at MNOs across Asia, EMEA (Europe, the Middle East and Afri-
ca) and the Americas: As the chart, below, shows, 58% of our respondents are from the
EMEA region, 26% from Asia, and 16% from the Americas.

15%

EMEA region
Asia
58% The Americas
26%

Our respondents represented more than 40 operator groups, from multinational

operators with companies in multiple major markets to Tier 2/Tier 3 operators in small
countries, so the industry spread was quite wide within our set of responses. The
respondents ranged from the very top level of decision-makers through to network
engineering staff: 20% of the respondents are CXO level (CTO/CTIO/CEO/CSO etc).

58% Of respondents work at MNOs in Europe, the Middle East

and Africa

26% Of respondents work at MNOs in Asia

16% Of respondents work at MNOs in the Americas

www.telecomtv.com 3
AI’S IMPACT ON THE RAN REPORT

AI IS HOT, BUT IS IT ALSO COOL?

To start, we wanted to know whether the MNO community feels it’s key for AI to play a
role in the RAN. So we asked: How important is it for mobile network operators to use
AI tools to optimise their radio access network (RAN) operations?

2%

5%

45%
48%

Absolutely critical
Quite important
Not that important
Not important at all
Not sure

It will come as no surprise, I’m sure, to find that the response is overwhelmingly
positive, with 45% believing it is absolutely critical and another 48% believing it is
quite important. Perhaps the biggest surprise is that the score isn’t higher for the
‘absolutely critical’ response. It’s also worth noting that 5% of respondents believe it’s
not that important, so not everyone’s on board the AI bandwagon… though at least
no-one thought it was ‘not important at all’.

www.telecomtv.com 4
AI’S IMPACT ON THE RAN REPORT

NO TIME LIKE THE PRESENT...

As everyone is constantly reminding everyone else, AI isn’t new in general and certainly is
not new to the telecom sector, including mobile operators, as tests, trials and deployments
have been underway for years. But we wanted to know to what extent AI use is already
pervasive in the MNO community, so we asked: When will your company use AI tools to
help manage its RAN?

AI is already being used to

52%
help manage the RAN

2025 19%

2026 8%

2027 or later 7%

Never 0%

Not sure 14%

0 10 20 30 40 50 60

Percentage of respondents

It might have been interesting to get a small group of telecom industry folks to take
a guess at the result of this question beforehand, because it’s quite difficult to find
anyone at an event such as Mobile World Congress (MWC) who doesn’t say “Oh,
we’re already using AI in our network”. With that in mind, I expected an almost clean
sweep with the response to this particular question.

But, as you can see, just 52% of our respondents say their company is already using AI
to help manage their radio access network, while another 19% say this will happen in
2025. Together, those numbers account for the majority of our respondents, but it’s
interesting to note that there are some who believe such deployments are still at least
two to three years away (if not longer), while 14% of our respondents aren’t sure about
the status/deployment timeline of using AI to manage the RAN.

www.telecomtv.com 5
AI’S IMPACT ON THE RAN REPORT

GETTING TO GRIPS WITH THE IMPACT OF AI

When the hype around a particular technology is strong, there is often a tendency
among MNOs to want to be seen as a leader, to be in the vanguard, sometimes without
actually thinking hard enough about why they’re deploying it: There are plenty of tales
about this already in the 5G era. But when it comes to AI, caution is the watchword –
this is something that no operator can afford to get wrong.

With AI, it seems there’s still a great deal left to figure out in terms of what such tools
can do in the context of network operations and what the impact of any deployment
might be. So we asked: Is it too early to properly understand how AI can impact RAN
operations?

2%

24%

74%

Yes
No
Not sure

The response is very clear, with a significant 74% of respondents saying it is not too
early to understand the impact of AI on the RAN, while only 24% believe more insight
is needed, and only 2% are unsure. Confidence levels are clearly very high, no doubt
fuelled by the learnings from those existing deployments.

www.telecomtv.com 6
UNVEILING THE POTENTIAL OF
AI IN TELECOMMUNICATIONS:
THE OPEN RAN REVOLUTION

Empowering Open RAN

Telecom Networks with AI

Wind River Studio delivers an integrated, distributed cloud platform unifying

infrastructure, orchestration, analytics, and AI capabilities so operators can
deploy and manage their intelligent 5G edge networks globally.

Wind River Studio is the proven solution for Open RAN. It is used in production,
at scale, for vRAN and Open RAN deployments today.

To learn more, visit windriver.com

AI’S IMPACT ON THE RAN REPORT

IMPACT ASSESSMENT

So we already know there are numerous AI-in-the-RAN deployments and more planned
soon, but the extent and operational impact of these deployments will be variable.
Ultimately, MNOs want AI to improve efficiencies and positively impact the experience
of their customers in a number of different ways, so we asked: By when do you think AI
will positively impact RAN operations for most mobile network operators?

It already is 24%

In 2025 22%

In 2026-27 38%

In 2028-29 14%

2030 or later 2%

Never 0%

0 10 20 30 40
Percentage of respondents

As the results show, the spread is quite even: 24% of our respondents believe there is
already a positive impact from the use of AI in the RAN, but the majority (60%) expect
that positive impact to be recognised in the next three years – 22% in 2025 and 38%
in 2026-27. Just 16% of our respondents expect the positive impact to kick in during
2028 or later, so it seems like we can expect to hear about lots of positive use case
studies in the next few years!

www.telecomtv.com 8
AI’S IMPACT ON THE RAN REPORT

DEDICATED RESOURCES

One of the many major challenges associated with Ai is what is increasingly known
as the ‘knowledge gap’: Telcos, along with everyone else, are keen to hire staff with
experience of using AI tools and developing AI-friendly strategies. There’s also the
question of whether telcos should have a centralised AI team or develop teams with
a specific focus or areas of expertise.

So we asked: Should mobile network operators form specific teams to focus on the
development and management of AI tools for the RAN?

7%

14%

79%

Yes
No
Not sure

This is something our respondents are particularly keen on, with a whopping 79%
saying that MNOs should build AI research and development teams specifically to
focus on the needs of the RAN and only 14% saying that such a team is not
necessary. (The more important follow-up question, of course, is whether MNOs will
ever get the green light from the executive leadership team to invest in such a team,
but that’s a question for another day…)

www.telecomtv.com 9
AI’S IMPACT ON THE RAN REPORT

SOURCING THE OPTIMUM TECHNOLOGY

In-house development teams can be one source of the AI tools and applications
needed to optimise the RAN, but such technology can be obtained elsewhere. So we
asked: From where should mobile network operators source the AI tools they will use
for RAN optimisation?

Internal development teams 26%

RAN network equipment vendors 27.5%

Other vendors* 27.5%

Not sure 19%

0 5 10 15 20 25 30
Percentage of respondents

*For example: Software developers, cloud software developers

Our MNO respondents were split on this one, as you can see. And although the
operators want an internal team focused on the development and management of AI
tools for the RAN, many of them clearly think that team should be working with tools
sourced from outside their four walls, as only about a quarter of them think an
internal team is the best place to source their AI-for-the-RAN tools, while some kind
of external vendor is the popular choice for most.

www.telecomtv.com 10
AI’S IMPACT ON THE RAN REPORT

AI IN THE RAN – WHAT IS IT GOOD FOR?

We included an ‘open’ question in our survey, asking our respondents to provide their own
responses, which they all did!

We asked: In what ways can mobile network operators benefit from the deployment of
AI tools to optimise the RAN (eg. energy efficiency, improved spectrum resource
management)?

The two responses that were referenced the most, indeed by many of our MNO
respondents, were:

• Improved energy efficiency

• Real-time cell optimisation and capacity management (similar to SON –

self-organising network)

Also generating multiple responses were (in descending order of popularity):

• Improved customer experience and quality of experience

• Network automation

• Network design efficiency/improved network planning

• Opex reduction (one respondent mentioned ‘radical TCO reduction’

• Improved troubleshooting

• Investment optimisation

Other comments mentioned were:

• Option to maximise all unused computational resources

• Radio resource partitioning

• Dynamic network slicing

www.telecomtv.com 11
AI’S IMPACT ON THE RAN REPORT

A few of our respondents went the extra mile and provided additional commentary –
here are three responses I think are worth highlighting:

• If you have already an optimised network with good machine learning algorithms
and a good network, AI can improve it potentially by another 5% to 10% – if you
have nothing, you will see bigger things.

• AI is still very much at the beginning of the technology curve for RAN and Open
RAN. We are still researching it, so it’s hard to tell where it will go and where it will
end up. Whatever use case we imagine, it will be something entirely different when
it actually gets deployed.

• To ensure a customer-driven mobile network, we should have proper AI tools and

models integrated to enable a 360-degree view of the network profile and
subscriber profile along with performance and energy-management enhancements.

Whatever use case we imagine, it will be something

entirely different when it actually gets deployed

www.telecomtv.com 12
AI’S IMPACT ON THE RAN REPORT

VIRTUALLY OPTIMISED

Much of the talk about the use of AI in the RAN these days goes hand in hand with
strategies related to virtual RAN (vRAN) and/or Open RAN, so we asked: Will it be
easier to deploy AI tools in virtual RAN (vRAN) architecturescompared with
traditional fully integrated RAN architectures?

19%

12%

69%

Yes
No
Not sure

The response here, with an overwhelming ‘Yes’ (69%), comes as no surprise, as the
architecture of a vRAN makes it (in theory) easier to integrate third-party AI tools.
Just 12% didn’t think this was the case.

www.telecomtv.com 13
AI’S IMPACT ON THE RAN REPORT

TAKE ME TO THE EDGE?

There’s also the debate about the pros and cons of deploying AI tools centrally or in a
distributed fashion, so we asked: Should AI tools for RAN optimisation be deployed at
the edge of the network in distributed nodes?

33%
43%

Yes, edge AI deployments will deliver

the best results
24%
No, AI tools will be just as effective if
deployed on a centralised cloud platform
Not sure

There was no consensus here and this is clearly an area where a great deal more
R&D is still to be done, as while 43% of our MNO respondents believe edge AI
deployments will deliver the best results, 24% think AI tools will be just as effective
if deployed on a centralised cloud platform, and a significant 33% are still not sure
about the best approach.

www.telecomtv.com 14
AI’S IMPACT ON THE RAN REPORT

EMERGING TECHNOLOGY

One of the hottest areas of R&D related to next-generation RAN management, and one
closely related to Open RAN developments, is the RAN intelligent controller (RIC) and its
associated near-real time and non-real time applications (the xApps and rApps).

Tests related to AI-enabled xApps on RIC platforms have been underway for some time
now, so we asked: How important will RAN intelligent controller (RIC) platforms be to
the use of AI in the RAN?

12%

7% 31%

Absolutely critical
Quite important
50% Not that important
Not important at all
Not sure

The good news for the emerging RIC sector is that MNOs think this technology is
going to play a key role, as 31% of our respondents believe the RIC will be absolutely
critical to the use of AI in the RAN, while another 50% think it will be quite important.
Only 7% think it won’t be that important but none of our respondents think it won’t
be important at all.

If you have any questions about our surveys, please do shoot me an

email at [email protected]

www.telecomtv.com 15
UPCOMING EVENTS
Our year-round programme of online and in-person events, expands on the
crucial topics of interest to the DSP Leaders community. For more information
and to sign up visit: www.telecomtv.com/events/

17-19 SEPTEMBER 2024 15-16 OCTOBER 2024

12-13 NOVEMBER 2024 5 DECEMBER 2024

INTERESTED IN PARTNERING?
Get in touch for more information on speaking and sponsorship opportunities,
email [email protected]
 For more information visit telecomtv.com
or email [email protected]

TelecomTV is produced by the team at Decisive Media.

© Decisive Media Limited 2024.

All rights reserved.
All brands and products are the trademarks
of their respective holder(s).

86-90 Paul Street

London, EC2A 4NE
United Kingdom