ETSI TR 104 099 V1.1.1 (2026-08)

TECHNICAL REPORT
Reconfigurable Radio Systems (RRS);
Feasibility study on the role of AI/ML techniques
for spectrum sharing
2 ETSI TR 104 099 V1.1.1 (2026-08)

Reference
DTR/RRS-0154
Keywords
dynamic spectrum sharing, native AI
ETSI
650 Route des Lucioles
F-06921 Sophia Antipolis Cedex - FRANCE

Tel.: +33 4 92 94 42 00  Fax: +33 4 93 65 47 16

Siret N° 348 623 562 00017 - APE 7112B
Association à but non lucratif enregistrée à la
Sous-Préfecture de Grasse (06) N° w061004871

Important notice
The present document can be downloaded from the
ETSI Search & Browse Standards application.
The present document may be made available in electronic versions and/or in print. The content of any electronic and/or
print versions of the present document shall not be modified without the prior written authorization of ETSI. In case of any
existing or perceived difference in contents between such versions and/or in print, the prevailing version of an ETSI
deliverable is the one made publicly available in PDF format on ETSI deliver repository.
Users should be aware that the present document may be revised or have its status changed,
this information is available in the Milestones listing.
If you find errors in the present document, please send your comments to
the relevant service listed under Committee Support Staff.
If you find a security vulnerability in the present document, please report it through our
Coordinated Vulnerability Disclosure (CVD) program.
Notice of disclaimer & limitation of liability
The information provided in the present deliverable is directed solely to professionals who have the appropriate degree of
experience to understand and interpret its content in accordance with generally accepted engineering or
other professional standard and applicable regulations.
No recommendation as to products and services or vendors is made or should be implied.
No representation or warranty is made that this deliverable is technically accurate or sufficient or conforms to any law
and/or governmental rule and/or regulation and further, no representation or warranty is made of merchantability or fitness
for any particular purpose or against infringement of intellectual property rights.
In no event shall ETSI be held liable for loss of profits or any other incidental or consequential damages.

Any software contained in this deliverable is provided "AS IS" with no warranties, express or implied, including but not
limited to, the warranties of merchantability, fitness for a particular purpose and non-infringement of intellectual property
rights and ETSI shall not be held liable in any event for any damages whatsoever (including, without limitation, damages
for loss of profits, business interruption, loss of information, or any other pecuniary loss) arising out of or related to the use
of or inability to use the software.
Copyright Notification
No part of this document may be reproduced in any form, by any means and in any media, without the prior written
authorization of ETSI and except as expressly permitted below.
By way of exception and when the document is a normative deliverable (European Standard (EN),
Technical Specification (TS), Group Specification (GS) or ETSI Standard (ES)), ETSI authorizes to reproduce
and incorporate into products, services and technical documentation only those extracts (e.g. templates) that are strictly
necessary for the technical implementation of the normative deliverable, to ensure compliance with the latter.
Nothing in this notice shall be construed as limiting any mandatory exceptions to copyright provided by applicable law.

© ETSI 2026.
All rights reserved.
ETSI
3 ETSI TR 104 099 V1.1.1 (2026-08)
Contents
Intellectual Property Rights . 4
Foreword . 4
Modal verbs terminology . 4
Executive summary . 4
Introduction . 5
1 Scope . 6
2 References . 6
2.1 Normative references . 6
2.2 Informative references . 6
3 Definition of terms, symbols and abbreviations . 7
3.1 Terms . 7
3.2 Symbols . 7
3.3 Abbreviations . 7
4 Survey of key administrations . 8
4.1 Introduction . 8
4.2 United States . 9
4.3 United Kingdom . 9
4.4 European Union . 10
4.5 Australia . 10
5 AI/ML for spectrum sharing . 10
5.0 Introduction . 10
5.1 Spectrum awareness and spectrum sensing . 11
5.1.1 Spectrum awareness . 11
5.1.2 Spectrum sensing . 11
5.2 Dynamic spectrum allocation . 12
5.3 Interference impact reduction . 12
6 Security aspects of AI/ML spectrum sharing . 13
6.1 Introduction . 13
6.2 Adversarial attacks on ML models . 14
6.2.1 Data poisoning . 14
6.2.2 Spectrum poisoning . 14
6.2.3 Model extraction and inference . 14
6.3 Privacy risks in AI/ML enhanced spectrum sharing systems . 14
6.3.0 General . 14
6.3.1 PU location and usage inference . 15
6.3.2 SU identity and mobility tracing . 15
6.4 Mitigation strategies . 15
6.4.1 Adversarial training . 15
6.4.2 Certified defences . 15
6.4.3 Outlier and anomaly detection . 15
6.4.4 Multi-agent validation. 15
6.5 Privacy-preserving ML . 15
6.5.1 Differential Privacy (DP) . 15
6.5.2 Secure Multiparty Computation (SMC) and homomorphic encryption . 16
6.6 Summary . 16
7 Conclusion . 16
History . 17

ETSI
4 ETSI TR 104 099 V1.1.1 (2026-08)
Intellectual Property Rights
Essential patents
IPRs essential or potentially essential to normative deliverables (European Standard (EN), Technical Specification (TS),
Group Specification (GS) or ETSI Standard (ES)) may have been declared to ETSI. The declarations pertaining to these
essential IPRs, if any, are publicly available for ETSI members and non-members, and can be found in
ETSI SR 000 314: "Intellectual Property Rights (IPRs); Essential, or potentially Essential, IPRs notified to ETSI in
respect of ETSI standards", which is available from the ETSI Secretariat. Latest updates are available on the
ETSI IPR online database.
Pursuant to the ETSI Directives including the ETSI IPR Policy, no investigation regarding the essentiality of IPRs,
including IPR searches, has been carried out by ETSI. No guarantee can be given as to the existence of other IPRs not
referenced in ETSI SR 000 314 (or the updates on the ETSI Web server) which are, or may be, or may become,
essential to the present document.
Trademarks
The present document may include trademarks and/or tradenames which are asserted and/or registered by their owners.
ETSI claims no ownership of these except for any which are indicated as being the property of ETSI, and conveys no
right to use or reproduce any trademark and/or tradename. Mention of those trademarks in the present document does
not constitute an endorsement by ETSI of products, services or organizations associated with those trademarks.
DECT™, PLUGTESTS™, UMTS™ and the ETSI logo are trademarks of ETSI registered for the benefit of its
Members. 3GPP™, LTE™ and 5G™ logo are trademarks of ETSI registered for the benefit of its Members and of the
3GPP Organizational Partners. oneM2M™ logo is a trademark of ETSI registered for the benefit of its Members and of ®
the oneM2M Partners. GSM and the GSM logo are trademarks registered and owned by the GSM Association.
Foreword
This Technical Report (TR) has been produced by ETSI Technical Committee Reconfigurable Radio Systems (RRS).
Modal verbs terminology
In the present document "should", "should not", "may", "need not", "will", "will not", "can" and "cannot" are to be
interpreted as described in clause 3.2 of the ETSI Drafting Rules (Verbal forms for the expression of provisions).
"must" and "must not" are NOT allowed in ETSI deliverables except when used in direct citation.
Executive summary
The present document examines the feasibility and role of AI/ML techniques for Spectrum Sharing (SS). It analyses
current regulatory signals and research trends from major regions (US, UK, EU, Australia, and selected Asia-Pacific
markets), distils the state of the art in AI-enhanced spectrum awareness/sensing, dynamic allocation, and interference
impact reduction, and identifies security and privacy risks together proposing possible mitigations appropriate for
AI-enabled Spectrum Sharing.
The present document analyses policy trajectories across jurisdictions, and contemplates AI-assisted spectrum
management (e.g. US National Spectrum Strategy and its 2024 Implementation Plan, Ofcom's hybrid sharing vision for
upper 6 GHz), which make sharing viable in bands that are impractical to clear for exclusive assignment ([i.1] and [i.2]).
Technically, AI/ML improves detection accuracy, prediction, and real-time decision making, enabling improved
protection of incumbents and a more efficient utilization of the shared spectrum. However, AI also widens the attack
surface (data poisoning, spectrum spoofing, model extraction) and raises privacy concerns, underscoring the need for
adversarial robust models, cryptographic protections, certification, and auditing.
ETSI
5 ETSI TR 104 099 V1.1.1 (2026-08)
Key conclusions:
i) AI/ML is a material enabler of Spectrum Sharing under diverse coexistence regimes;
ii) regulators increasingly plan for AI-augmented management;
iii) standardization should cover architectures, metrics, datasets, assurance, and conformance for AI-enabled SS;
and
iv) security/privacy requirements are essential from the outset.
Introduction
The industry is at cross-roads in respect of spectrum needs and its availability. Continuous growth of capacity demand,
especially in view of upcoming 6G deployments, highlights the need for additional spectrum. However, the more
desirable part of available spectrum is fully saturated with very little possibility of exclusive usage. Thus, regulators are
exploring new avenues for making the most efficient use of spectrum, including the possibilities of spectrum sharing.
The scarcity of desirable spectrum for 6G and the rising complexity of coexistence with non-3GPP incumbents
(e.g. radars, satellites) make dynamic, automated sharing attractive. AI/ML offers concrete capabilities for enhancing
sensing, prediction, optimization, and control that are impractical in fully algorithmic architectures. The US National
Spectrum Strategy (NSS) and the 2024 Implementation Plan explicitly pursue technology-enabled sharing and band
studies [i.3].
In Europe, Ofcom's 6 GHz hybrid sharing concept and BEREC's analyses of AI in telecoms point to future regulatory
models where sharing-by-design is embedded in equipment standards. The EU AI Act will also shape deployment of
AI in network functions, including spectrum functions [i.2].
The present document captures an early perspective of the possible impacts of AI on spectrum sharing in the
foreseeable future.
Artificial Intelligence (AI) techniques are expected to impact the capabilities of transmitters and receivers of virtually
all telecommunication services, and AI is also likely to have impact on similar communication services that use
spectrum and the ways in which devices access and share spectrum.
AI thus can potentially manage transmissions in particular frequencies, in particular directions and at particular time-
windows, thus reducing interference towards other users of the same spectrum. AI can also potentially allow reception
of certain levels of interference by reducing its impact on the desired signal. Additionally, AI has the potential to
improve interference estimates, spectrum occupation forecast and manages automatically and transparently dynamic
spectrum access and sharing. All of the above may allow less conservative deployment scenarios to protect incumbents.
These enhanced capabilities may allow regulators to regulate and manage spectrum in ways that were unimaginable
even a few years ago.
ETSI
6 ETSI TR 104 099 V1.1.1 (2026-08)
1 Scope
The present document studies technical approaches for automated spectrum access to support dynamic, temporary, and
flexible spectrum sharing. In light of recent advancement on AI/ML and their potentiality, the aim of the present
document is evaluating the possibility to include similar mechanism within a Spectrum Sharing framework. In
particular, it analyses the possible advantages of such techniques, and their relative risks from both a technical and
regulatory point of view.
The present document will begin by analysing the outlook for AI/ML based solutions for spectrum sharing among the
different regions of the world.
2 References
2.1 Normative references
Normative references are not applicable in the present document.
2.2 Informative references
References are either specific (identified by date of publication and/or edition number or version number) or
non-specific. For specific references, only the cited version applies. For non-specific references, the latest version of the
referenced document (including any amendments) applies.
NOTE: While any hyperlinks included in this clause were valid at the time of publication, ETSI cannot guarantee
their long-term validity.
The following referenced documents may be useful in implementing an ETSI deliverable or add to the reader's
understanding, but are not required for conformance to the present document.
[i.1] NTIA (March 2024): "National Spectrum Strategy Implementation Plan".
[i.2] Ofcom: "Consultation: Hybrid sharing: enabling both licensed mobile and Wi-Fi users to access
the upper 6 GHz band".
[i.3] NTIA (November 2023): "National Spectrum Strategy".
[i.4] Autumn Statement 2023 - GOV.UK.
[i.5] Everything you need to know about Spectrum Sandboxes | UKTIN.
[i.6] "Deep Learning for Spectrum Sensing in Cognitive Radio Networks: A Survey", by S. Sharma, et
al. (2020).
[i.7] "AI-Powered Spectrum Sensing for Cognitive Radio Networks: A Survey", by A. Al-Dweik, et al.
(2021).
[i.8] "Predictive Spectrum Allocation for Cognitive Radio Networks using Machine Learning", by
TM
M. A. Khan, et al., IEEE Transactions on Cognitive Communications and Networking.
[i.9] "Deep Reinforcement Learning for Dynamic Spectrum Allocation in Cognitive Radio Networks",
TM
by A. Al-Shuwaili, et al., IEEE Transactions on Cognitive Communications and Networking,
vol. 5, no. 1, pp. 123-135, 2019.
[i.10] "A Deep Reinforcement Learning Approach for Dynamic Spectrum Allocation in 5G Networks",
TM
by M. Chen, et al., IEEE Transactions on Wireless Communications, vol. 19, no. 10,
pp. 6698-6712, 2020.
ETSI
7 ETSI TR 104 099 V1.1.1 (2026-08)
3 Definition of terms, symbols and abbreviations
3.1 Terms
For the purposes of the present document, the following terms apply:
Artificial Intelligence (AI): field of computer science which enables systems/machines to simulate human intelligence
in processes
Machine Learning (ML): subfield of AI which allows- systems/machines to imitate intelligent human behaviour
through learning (identification of patterns)
3.2 Symbols
Void.
3.3 Abbreviations
For the purposes of the present document, the following abbreviations apply:
AI Artificial Intelligence
AoA Angle of Arrival
API Application Program interface
CBRS Citizen Broadband Radio Service
CSMA Carrier Sense Multiple Access
DNN Deep Neural Network
DoD Department of Defence
DP Differential Privacy
DSIT Department for Science, Innovation and Technology
ESC Environment Sensing Capability
EU European Union
FCC Federal Communications Commission
GPT Generative Pre-trained Transformers
IMT International Mobile Telecommunication
LBT Listen Before Talk
LSTM Long Short-Term Memory
MIMO Multiple Input Multiple Output
ML Machine Learning
NOI Notice Of Inquiry
NSS National Spectrum Strategy
PCA Principal Component Analysis
PU Primary User
QoS Quality of Service
RL Reinforcement Learning
RNN Recurrent Neural Network
RSSI Received Signal Strength Indicator
SAS Spectrum Access System
SINR Signal to Interference plu Noise Ratio
SS Spectrum Sharing
SU Secondary User
SVDD Support Vector Data Descriptor
SVM Support Vector Machine
TAC Technological Advisory Council
WRC World Radiocommucation Conference
ETSI
8 ETSI TR 104 099 V1.1.1 (2026-08)
4 Survey of key administrations
4.1 Introduction
Spectrum regulations historically depended on a steady supply of the valuable resource and spectrum availability for
users has so far been mostly stable during the period the regulations have been in force. However, as demand for
capacity grows and desirable spectrum remains a scarce resource, every effort is underway to make the most efficient
use of it. In this context, Artificial Intelligence (AI) and Machine Learning (ML) have become very important. Below
are the basic definitions for some context:
Artificial Intelligence: The field of computer science which enables systems/machines to simulate human intelligence
in processes.
Machine Learning: The subfield of AI which allows systems/machines to imitate intelligent human behaviour through
learning (identification of patterns).
AI/ML techniques are now used widely in radio technologies including in AI-native air interfaces, for various purposes
including channel estimates, scheduling and resource allocation, interference cancellations, load balancing/traffic
steering, etc.
Two recent AI/ML technology trends may significantly impact the regime of spectrum management and regulation.
First, there has been a revival of traditional AI over the past decade, driven by an explosion of availability of different
types of data, which in turn, was enabled by tremendous growth in storage and computational technologies. This AI
boom has been further brought to the forefront of technology discussions by the maturity of generative AI in the last
couple of years as reflected in advancements in Generative Pre-trained Transformers (GPT) in applications such as
® ®
OpenAI's ChatGPT or Google's Bard/Gemini. These fundamental models are being extended for numerous
customized models to suit numerous purposes in virtually any application in any industry.
Secondly, with 6G in the horizon, the search for new spectrum to adopt the new mobile generation and to accommodate
the continued explosion of mobile data has started all over again. However, the required amount of exclusively licensed
spectrum in desired frequencies is virtually non-existent as the entire range is occupied by critical applications in most
developed countries. Hence there is a renewed interest in spectrum sharing - especially vertically between mobile
operators and non-3GPP-based incumbents, e.g. radars, microwave radios, satellites etc. This notion of spectrum
sharing is opening new horizons for using AI, which in turn, provides administrations new tools for making spectrum
available.
Regulators in the US, UK and Europe are envisioning that AI will be a key technology to realize the possibilities of
efficient spectrum management and sharing in the 6G era. The only wide-scale implementation of spectrum sharing in
the world, CBRS band in the US, has been a rather challenging exercise to gain eventual commercial success. Newer
AI/ML technologies promise significant improvements over the current state of the art.
Analysis Mason in a recent publication discussed how "AI-techniques" may be used for spectrum management but
concluded that "given the complex nature of interference avoidance and the limited amount of experience in using
AI/ML techniques, machine learning seems unlikely to result in significant overhauling of general spectrum
management approaches. However, it is possible that dynamic forms of sharing within specific bands designated for
such purposes could be optimized through devices and networks that use machine learning".
While there is a very broad impact of AI on mobile networks (e.g. traffic management, performance enhancement,
Quality of Service (QoS) delivery, etc.) and their operations (e.g. energy consumption, maintenance, customer service,
etc.), the discussions in the present document are restricted only to the aspects that may impact spectrum
administration, management and regulations in the foreseeable future. It is conceivable that newer AI applications
will emerge that will impact spectrum usage and regulation - either directly or indirectly - and the present document will
need to be updated at appropriate time intervals.
ETSI
9 ETSI TR 104 099 V1.1.1 (2026-08)
4.2 United States
The National Telecommunication and Information Administration (NTIA) of the United States published its National
Spectrum Strategy (NSS) in November 2023, highlighting the need for using advanced technology techniques like
AI/ML to maximize the utilization of the valuable spectrum resources, and the importance of spectrum sharing,
given that desirable spectrum bands for 6G have already been allocated and in use, and there is not much room left for
spectrum clearance to allow for exclusive 6G usage.
Earlier in the year the Federal Communications Commission (FCC) issued a Notice Of Inquiry (NOI) to assess the
spectrum usage by various non-federal entities. This involves collection of a vast amounts of data and processing them,
which would have been practically impossible even a few years ago, especially without today's advancements of
AI/ML. This has been stated as:
"As the radiofrequency environment becomes more congested, leveraging technologies such as artificial intelligence
to understand spectrum usage and draw insights from large and complex datasets can help facilitate more efficient
spectrum use, including new spectrum sharing techniques and approaches to enable co-existence among users and
services."
In July 2023, the FCC and National Science Foundation (NSF) held a conference "AI for Communications Networks and
Consumers". Chairwoman Rosenworcel mentioned that 'AI can be used to improve spectrum efficiency in the i
...