knowledge

REPORT

in autonomous
networks
SPONSORED BY:

Author:
Dana A. Cooperson,
Contributing Analyst

Editor:
Ian Kemp,
Managing Editor

March 2022
inform.tmforum.org
contents

03 04 09
the big picture section 1: section 2:
what is intent and why key intent concepts and
does it matter? building blocks

15 18 20
section 3: section 4: section 5:
challenges to CSPs are making where is intent-driven
implementing intent progress automation heading?

24 26
section 6: additional
make it happen features &
– strategies for resources
embracing and
implementing intent

2 inform.tmforum.org
the big picture
The concept of intent in autonomous networks is still very new and often Autonomous networks, and using
misunderstood. The idea of intent-based networking has been kicking around intent to drive them, are new
telecoms since at least 2017, but serious industry conversations of intent as a approaches in telecoms and there
driver of telecoms operations automation are only now progressing. are few if any robust commercial

T
examples. In this report we use TM
he drivers are clear. add complexity. This complexity Forum Catalyst proofs of concept,
Business, service and is outstripping CSPs’ ability to and extrapolation from successful
network operations manage, control and assure automation implementations and
are becoming increasingly customer experience, service promising new solutions, to show
complex as services become operations, cost and network what can be accomplished.
more heterogeneous, dynamic, performance using today’s tools
distributed, and broadened to and approaches. The focus is on the market
include multiple cloud assets implications of intent in autonomous
As a consequence, manual and networks, not on the technology
and third-party developer and
static programmatic and rules- that underpins it. It explores the
other partner solutions. 5G in its
based automation must give way concepts and building blocks of
standalone incarnation brings
to more model- and knowledge- intent and highlights resources to go
the promise of new revenue
driven approaches that are based deeper into the subject.
streams for communications
on the intents behind business,
service providers (CSPs) and Read this report to understand:
service, and resource requirements
more differentiated services
and constraints. By adopting those What intent is in the context
that better meet customers’
approaches, services can adapt of telecoms networking and
specific needs, but it will also
and evolve more autonomously as operations automation
network conditions, business goals
Why intent is important
and customer requirements shift
over time. The key concepts and building
blocks of intent in autonomous
networks
Meeting customer needs
Manual and static Drivers of, and obstacles to,
The ability to drive autonomous
programmatic the implementation of intent-
networks and CSP operations
driven automation
and rules-based with intent is built on a simple yet
powerful concept: By capturing How CSPs can implement
automation intent-driven automation, and
customers’ needs, and acting to
must give way to satisfy them without exposing or what might help accelerate
more model- and mandating any implementation adoption
knowledge-driven details, customers will have a Recommendations for CSPs
better experience and CSPs will to plan the journey to intent-
approaches have more flexibility in how they driven network automation.
respond to meet those needs.

3 inform.tmforum.org
section 1:

what is intent and

why does it matter?
TM Forum’s member group, the Autonomous Networking Project (ANP), in the Intent in Autonomous Networks
Introductory Guide defines intent as: “The formal specification of all expectations including requirements, goals and
constraints given to a technical system.” An intent articulates what the intent owner expects the system to provide or
do, but not how to do it. Decoupling the “what” from the “how” improves implementation flexibility and ultimately
allows autonomous systems to fulfill the intent based on data and knowledge.

That TM Forum definition is loop autonomy without intent.

compatible with the IETF’s Intent-based management
definition and illustrates the is baked into the AN journey
practical approach of the ANP at least as early as Level 3
team: wherever possible, it (conditional autonomous
is building on and extending networks) and becomes essential Intent is the
previous standards work on at Levels 4 and 5 – highly and driving force
intent, embracing applicable fully autonomous networks,
best practices from outside respectively (see graphic on p.5).
of autonomous
telecoms. The ANP has been Read our recent report
networks (AN).
working to document the Autonomous networks: exploring Automation of
properties and principles of intent,
how to express the intent, and
the evolution from level 0 to level networks and
5 to find out more about the
its impacts (such as changes related operations
steps CSPs must take to achieve
to rules and responsibilities), autonomous network operations. cannot evolve
and it has developed a set of
and mature to
technical recommendations for
implementation. full closed-loop
Intent is the driving force of autonomy without
autonomous networks (AN). intent.
Automation of networks and
related operations cannot evolve
and mature to full closed-

4 inform.tmforum.org
section 1: what is intent and why does it matter?

Autonomous networks maturity model Key intent concepts

Each level of the maturity model has a set of characteristics describing Several simple yet powerful
the evolutionary stage of the CSP’s journey from fully manual to fully concepts are explicit in or implied
autonomous operations: by the intent definition:
Intents have “owners” and
“handlers”.
An intent articulates what
the intent owner expects
the system to provide or do,
but not how to do it; intent
mandates no implementations
or policies.
A typical requirement
expressed as an intent is what
Fully autonomous network is to be delivered (a service,

5
The system has closed-loop automation capabilities an application and so on) and
across multiple services, multiple domains – including with what properties (such as
partners’ domains – and the entire lifecycle. cost and uptime).
The expression of the intent
Highly autonomous network as passed to the handler must

4
In a more complicated, cross-domain environment, the be both comprehensive and
system enables decision-making based on predictive unambiguous for the machine
analysis or active closed-loop management of service- to properly implement it.
driven and customer experience-driven networks.
An intent refocuses the
interaction between human
Conditional autonomous network (the intent owner) and

3
The system senses real-me environmental changes
machine (intent handler) on
and in certain network domains will optimize and
the business needs; intent is
adjust itself to the external environment to enable
the “currency” of exchange
intent-based, closed-loop management.
between human and machine,
and it can also be the
Partial autonomous network

2
currency between a machine
The system enables closed-loop operations owner and handler.
and maintenance for specific units based on AI Decoupling the “what”
modelling under certain external environments. from the “how” improves
operations flexibility. It enables
Assisted operations and maintenance negotiation between owner

1 The system executes a specific, repetitive subtask and handler to optimize the
based on pre-configuration in order to increase machine response and allows
execution efficiency. the machine to respond
dynamically over time to
changing conditions without
Manual operations and maintenance mandating a return to the

0 The system delivers assisted monitoring

capabilities, but all dynamic tasks must be
executed manually.
owner to confirm permission
to make changes that bring
operations in line with the
intent.
Source: TM Forum, 2021

5 inform.tmforum.org
section 1: what is intent and why does it matter?

changes over time as their needs disambiguates intent from other

change. But its automation is rules- drivers of automated action.
and policy-based rather than fully Intents, once interpreted and
intent driven as would be needed expressed so a machine can
for autonomous operations.
Intent is the understand them, may need to
At a higher level of abstraction, be decomposed into sub-intents
‘engine’ that intent could be expressed as: that can be subjected to rules/
powers the “I need to connect these four policies, RPA, AI/ML, intelligent
vehicle of locations for the purpose of x,
y and z” – for example, running
business process management
(iBPM), orchestration and so
automation specific applications, supporting on. In that sense, intent is the
into motion some number of people and “engine” that powers the vehicle
running a 24x7 operation. This of automation into motion and
and gives it level of abstraction gets closer to gives it velocity.
velocity. allowing the customer to express
the business requirements of the Why does intent matter?
service it wants to buy.
The motivations and drivers
At an even higher layer of for moving to intent-based
Intent can be expressed at various abstraction, and expressed
levels of abstraction. At a first automation approaches include:
more in natural language terms
level, for example, a customer that must be understood and Networks and network
might express specific service processed by a machine, a operations are
characteristics they desire customer might say: “I want growing increasingly
(e.g., a specific SLA parameter, access to my test network from complex. Always-connected,
or guaranteed bandwidth or 8am to 5pm on Monday to Friday, always-on communications,
latency), which the system and I only want this group of and the extension of ever more
would then translate into device people, but not other groups, sophisticated networks and
configurations or other actions. to have access.” Or, an operator applications to buildings, vehicles,
The customer doesn’t need to might express the need to realize edge locations, devices and
worry about the details, but does a certain financial return when multiple clouds, is outstripping
need some understanding of providing a specific service type our ability to manage such
networking. to a customer. Interpreting this sprawling end-to-end
In telecoms, this level of intent- business intent would require connections flexibly, dynamically
driven (also referred to as “intent- analyzing a lot of different data. and cost-effectively enough to
based”) networking has been In very general terms, the level of meet customer and application
embraced and implemented abstraction required in expressing requirements. Humans must
narrowly within the enterprise intent rises from resources to teach machines how to take over
services domain to date, for services to business intents. many day-to-day operations now
example with software-defined done primarily manually or with
wide area networking (SD-WAN) programmatic, as opposed to
Not just a new name knowledge-based, automation.
and network as a service (NaaS).
The Vodafone Ready Network, The concepts of intent, policy/ The complexity of
as an example, is a “portfolio of rules and orchestration are customer services and,
on-demand networking products complementary; all are needed therefore, customer
for the digital business”. It’s a to drive business, service and service order processes, is
NaaS offering that provides fast network operations automation. rising with network complexity.
provisioning (zero-touch) and If operations is ever to become 5G is at the root of many of the
enterprise self-control, “driven by autonomous, policy and increasing service opportunities
business policies and powered by orchestration, robotics process as well as increasing complexities.
analytics”. Introduced in 2018, like automation (RPA), natural 5G promises to support private
all NaaS implementations it lets language understanding (NLU), network services as well as new
customers pick service options AI/machine learning (ML), and services such as ultra-low latency
– for example, for an SD-WAN other automation technology for vehicle to anything (V2X)
service with specific bandwidth actions must flow from intent; the and industrial control. Making the
and security features – and make notion of “expectations” is what most of new 5G opportunities

6 inform.tmforum.org
section 1: what is intent and why does it matter?

end visibility of their network. value is low latency or some

Customers want to define, other quality over time. This
manage and prioritize their approach would be a big shift for
services as per their needs, and CSPs, which have spent decades
selling network technology, and
Customers they want it done fast.” This is
one example of a growing trend customers, who have bought it.
want to define, toward more customer visibility
manage and and control. The evolution to true Flexible services
intent-driven networking would
prioritize their allow Windstream to go further In research for this report one

services as per towards realizing its AN vision. large European-based CSP

noted the following scenario:
their needs, The variety and A customer might want to
complexity of
and they want customer engagement
securely connect its offices to
each other and to multiple public
it done fast.” touchpoints are rising. As cloud providers. The customer
services are becoming more would only have to consider its
complex and enterprises and specific goals and requirements,
consumers want more control, not the technical details. The
and successfully supporting them
and as the ways and systems CSP’s autonomous system could
will require a more autonomous
through which CSPs engage conclude through its knowledge
service operations approach
with their customers grow, and data assessment that a
than is possible today. CSPs
the need for automation rises. secure SD-WAN overlay service
want more flexibility in service
Driving engagement through the supported by its own network
implementation; through intent-
customer’s intent promises to underlay and those of its partners
driven autonomous systems
streamline interactions. best meet the customer’s SLA
they can better meet the wishes
of their customers while also Network resources and the CSP’s costs.
optimizing profitability. should scale and adjust “This would be a global business
flexibly based on need, as is the offering,” says the operator. “We
Enterprises want more
case with public cloud compute, may underpin a local business
control over their
storage, and networking customer’s service by different
services and how they
and streaming services. operators’ underlays, but that
support their business
Enterprises have become would be transparent to the
goals. Intent allows enterprises
used to assuming the public customer. The customer provides
in diverse industry verticals to
cloud can provide infinitely- the topology of their sites,
express what they want and
scaled capacity. Consumers including cloud gateways, and
need in their own language,
likewise expect streaming and we design the service as needed.
not the language of networking
other over-the-top services to Essentially, they get a private
technology. A 2020 TM Forum
adjust automatically based on network, without ever using
case study details the goals
network and device conditions. the public internet. Our system
and benefits of Windstream’s
Telecoms network resources could make autonomous changes
Project Kailash, which created
need to be as ubiquitous and to the underlay as needed to
a programmable, sliceable and
flexible. Resources should be maintain the contracted SLA and
scalable autonomous network
available wherever the customer cost targets without first getting
(AN) controlled by “automated,
is – for example, connecting permission from the customer,
intent-based policies” to improve
automatically via either Wi-Fi because such flexibility would be
efficiency of the activation
or cell service – based on what baked into the contract.”
lifecycle. Bhaskar Madari, Senior
is available and best satisfies
Vice President for Delivery at Driving autonomous networks
the intent. CSPs need to
Windstream’s implementation with intent won’t eliminate the
design services accordingly. All
partner, Prodapt, noted: “There increasing complexity, but it can
customers may not need five-9s
is a growing demand from “encapsulate” it in autonomous
availability given that they often
customers to control what domains that can then be
have multiple ways to connect
service they want, by when federated. Existing approaches
at a given time, so why incur the
they want and how they want, – for example, policy- and rules-
cost of giving them more than
along with [having] end-to- based – fall short, as they are
they need. Perhaps what they

7 inform.tmforum.org
section 1: what is intent and why does it matter?

network technology, either through and responsibility framework

a static catalog, an automated can be extended beyond
ordering portal, or a heavily people to autonomous systems.
customized and human resource- Establishing a matrix of rules and
The more you can intensive sales process. However, responsibilities in the context of
abstract the better the customers don’t really care what autonomous systems is likely to be
technologies the CSP uses as long a complex undertaking, given that
with intent. The as they get the outcome they desire. a consulted party in one instance
less the customer During research for this report one (for example an intent handler)
can be a responsible party in
must specify, the CSP said: “Today, we design the
service, show the customer the another instance (for example
fewer the technical design, and the customer accepts it an intent owner to another
details that must and signs off. If we want to change handler). RACI in intent-driven
the network, even to make it better, operations must be considered
be exposed, the
we have to go back to the customer within the intent object lifecycle:
more the CSP has and ask permission.” Service detection, investigation, definition,
a choice. ordering and contracts will need distribution and operation.
to be approached from an intent
perspective to change this dynamic. Greater abstraction
too rigid and hard to change as
Once the intent owner’s In the Windstream case study
conditions (network conditions
requirements and goals are Art Nichols, Vice President of
or customers’ needs) change.
captured, the system needs first Architecture and Technology, said:
Intent-driven operations,
to determine whether or not the “The industry has done a disservice
supported by a “learning system”
intent can be met. If not, some by exposing this complexity
that uses both reinforcement
negotiation may be required to of how networking works to
learning and transfer learning –
reach an agreement on what customers.” It’s the industry’s job
it can use generalized learning
intent can be met. Only then can to make ordering and managing
from one task or use case to train
the system implement the intent services simple for the customers.
another one, and it can make a
through actions. Once the intent
sequence of decisions in complex Further, one CSP in an interview
is implemented, monitoring is
environments to achieve a goal, for this report noted: “The more
required to ensure the intent
respectively – can provide the you can abstract the better with
continues to be met (see more
needed flexibility over time and intent. The less the customer must
on intent management in section
enable increased autonomy of specify, the fewer the technical
2). Intent guides the autonomous
the operations systems. CSPs details that must be exposed, the
network to self-provision, self-
can’t develop their way out of more the CSP has a choice. As the
heal and self-configure to provide
the rigidity of current approaches system becomes more intelligent
the right bandwidth, latency or
programmatically; a knowledge- with machine-based automation,
other expressed customer needs;
based approach is needed. it can understand how things are
as long as the CSP meets these
requirements, it should be free working and can change topology
Changing the status quo to give the customer what they and other elements to ensure the
need and no more. service is working optimally.”
Intent-driven automation implies
changes to the commercial Traditional roles and responsibility Adaptable and dynamic intent-
and service order discussions frameworks apply to people; with based learning systems are
between CSPs and their intent-driven automation, such mandatory for CSPs that want
customers and partners and how frameworks must be extended to to achieve higher levels of
the CSP operates and manages the systems that will be deciding operational autonomy. If the
the network and related support best actions – autonomously or destination is autonomous
systems. These shifts also imply in concert with the intent owner – operations, then intent provides
changes in who does what and and implementing them. The TM the fuel and the signposts for
who is responsible for what. Forum AN Project team proposes that journey.
Intent requires a whole different how the so-called “RACI” (who is In the next section, we’ll dig into
approach to providing services to Responsible, who is Accountable, some of the key concepts behind
customers. Today, CSPs largely sell who must be Consulted, and intent in autonomous networks.
who must be Informed) rules

8 inform.tmforum.org
section 2:

key intent concepts

and building blocks
Intent concepts, management
building blocks and modeling Autonomous Network Project intent resources
and interface technologies can
get complicated fast. This section Much more depth on the TM IG1253B Intent Extension Models
of the report looks deeper into Forum Autonomous Network Extensions to the common
the intent concepts discussed in Project’s intent proposal, model for expressing intents
section 1 to give context to why recommendations, and that might be relevant in specific
intent is so fundamental to the progress on specifying APIs, domains or contexts.
autonomous network journey. interfaces, models and other IG1253C Intent Life Cycle
critical components required to Management and Interface
It is important to consider the implement intent in AN, can be Details of lifecycle management
following properties of intent: found in these documents: including roles and
I ntents declare requirements, IG1253 Intent in Autonomous responsibilities related to intent
goals and constraints. They Networks management.
express both outcomes desired The ANP’s overall proposal IG1253D Intent Manager
and outcomes to be avoided. for implementing intent in Capability Profiles
They can require that the system autonomous networks. Schema for registering and
meets specific performance
IG1253A Intent Common Model discovering intent manager
KPIs or metrics, or they can
The generic model-based capabilities and scope.
request that the system meets
approach for human-machine
more abstract conditions, such
and machine-machine expression
as a financial return.
and interpretation of intent.
I ntents do not invoke specific
actions. They leave actions
open and let the system met. If it can no longer meet
I ntents are persistent and
determine the right course the stated intent, the handler
lifecycle managed. Only the
of action, or if the outcome will attempt remediation and
intent owner can cancel an
desired is even possible. alert the owner.
intent. Until such time, the
I ntents are composable and system will continue to check Intents are infrastructure

additive. As such, intents may that the intent is being met; agnostic and portable.
conflict with each other or if not, the system will try to Intents are the “currency
may need to be prioritized. repair the handling of the of exchange” between user
intent so that conditions are

9 inform.tmforum.org
section 2: what is intent and why does it matter?

and handler and the basis tiers of the operations hierarchy:

of communication between business, service and resource.
autonomous domains Some of the most important
and between layers of intents to express as operators
the AN framework. Intent move towards autonomous What can’t
portability may cross vendor operations include:
or operator domains, so a A business intent could
be observed,
common and standardized include SLA negotiations measured and
interface is critical. and agreements/contracts. controlled
Intents are measured and
 Once these business
grounded in data. What can’t imperatives are expressed over time
be observed, measured and in a contract, the operator can’t be viable
controlled over time can’t be
viable intents with which to
and its autonomous
systems should be free to
intents.
drive the autonomous system. change implementations
Operation must be verifiable automatically over time
the user has, and what data
under conditions that might without needing to stop and
security or sovereignty
change dynamically. The ask for permission.
constraints apply.
data needed to manage the At the service level, intents
intent must be accurate and A resource intent might
include service delivery
accessible. specify network function
requirements such as what
instance placement
Examples of intents that a user service or catalog item
requirements or latency
or intent owner might need to is to be delivered, what
requirements for an edge
express encompass the three performance expectations
application.

where intent fits into

autonomous networks

0 1 2 3 4 5

Level 0 Level 1 Level 2 Level 3 Level 4 Level 5

Manual Assisted Partial Conditional High Full
Operation & Operation & Autonomous Autonomous Autonomous Autonomous
Maintenance Maintenance Network Network Network Network

<<Intent Owner>>
Intent Experience
expectations/requirements e.g. SLA compliance

Intent Management <<Intent Handler>>

Control Loop Management

Decision
Awareness Analysis Execution Control loop
making

TM Forum, 2022

10 inform.tmforum.org
section 2: what is intent and why does it matter?

intent management function

intent report

Knowledge Decision Actuation

update Intent management function (handler) action

the basic building block of intent-based operation

Other
Intent
(not intent-based)
interface
Interfaces
report intent

TM Forum, 2022

Intent management based closed loop management

of automation is likely to work
The basic intent building block is
its way into Level 3 conditional
the relationship between intent
autonomous network operations,
owners and intent handlers
The basic intent that is facilitated through intent
but it will be a prerequisite for
Level 4 highly autonomous
building block is management processes.
operations, albeit under human
the relationship In the infographic on p.10, we can supervision where still required.
see a generic interaction whereby
between the owner, which can be either a
By Level 5, awareness, analysis
and decision execution of intent
intent owners human or a machine, passes its are fully autonomous in all
and intent expectations and requirements scenarios.
to the intent handler, which
handlers that is always a machine. Through
Going a level deeper, the
infographic above illustrates
is facilitated the control loop, the handler
how the intent handler in one
through intent becomes aware of the intent,
instance can also be an intent
analyzes it, makes decisions
management based on its knowledge base,
owner to another handler. The
intent management function/
processes. and executes the intent. It then
handler is responsible for a
reports back to the owner on
specific operational domain; its
the agreed-upon terms, such
actions, therefore, might include
Quantitative intents can be as frequency of reporting. The
passing intent to another handler
expressed in absolute values, handler constantly assesses
responsible for another domain. If it
ranges, or minimum or maximum conditions to ensure the intent’s
performs the action, it acts through
performance requirements. persistent state. Such intent-

11 inform.tmforum.org
section 2: what is intent and why does it matter?

Detection
Identify the need to define,
change or remove intent to
set requirements, goals
and constraints. React
to handling success
or failure

Operation Investigation
Operate according to Verify what intent is feasible
the given intent. Report
status and success back
to the owner
lifecycle of an and can be provided by the
underlying handlers

intent object

Distribution Definition
Identify the intent Decide what to require
handlers according to the from the underlying intent
targeted management handlers and create,
domains and inform them modify or remove intent
about the new intent accordingly
configuration

TM Forum, 2022

conventional interfaces, such as an intent interpreter function and all, required contextual
open APIs that effect configuration must interpret it to remove any information.
and setup of a particular network ambiguity and formalize its Beneficiaries of an intent and its
element or function. expression. TM Forum’s ANP handling could include the CSP’s
If the intent comes directly strongly suggests that a single customers (internal and external);
from a user (for example via point of intent interpretation is the service provider shareholders,
a customer portal), it may be logically located close to the legislators and regulators; and
expressed in a natural or domain- intent originator so that the potentially other groups or
specific language. In this case, interpreter can capture any, individuals. It is unlikely that all

12 inform.tmforum.org
section 2: what is intent and why does it matter?

these beneficiaries (for example Customers may express their communicate throughout the
shareholders and legislators) goals, needs and requirements intent object’s lifecycle. It enables
will also be intent owners, but through intent that then needs the communication of intent
they could certainly need to to be broken down into a set of to the handler, the exploration
be informed through intent federated “sub-intents” that are and negotiation of intent
management reporting. Note that domain and subsystem specific. requests based on feasibility
although an intent can only have Each domain or subsystem and available options between
one owner, it can have more than would typically have an intent handler and owner, and the
one beneficiary. management function that is reporting of intent state back
specialized in the tasks it is to the owner. It also enables
Spanning CSP operations responsible for. The scope of an the registration and discovery
intent manager’s responsibility, of intent handlers’ capabilities
Intent and its handling will therefore, will match clearly within the system. The interface
be distributed across a CSP’s defined autonomous domains or is polymorphic, which means that
business, service and resource subsystems with clear borders. a single API implementation is
operations, across architectural very adaptable across different
layers and throughout domains and sub-domains
autonomous domains as Intent interfaces
and as user requirements and
illustrated in the graphic below. The intent interface (or intent system capabilities change. No
API) is the interface through domain-specific APIs need to
which owners and handlers be developed: the interface is

intent management mapping

to CSP operations stack
• Customer management
• Partner management
Customer Contract and order
portal management • Product management
• Marketing and sales
Business intent manager
Business management
Business Operations • Revenue management
portal (charging and billing)

Operations Service intent manager • Service design

portal • Service management
Service (including catalog, inventory,
Operations order management,
Intent managers specific to operation sub-domains activation and assurance)
• Service orchestration
(including lifecycle
management)

• Network orchestration
Resource
• VNF/CNF management
Operations
• Network and element
management
AD1 Intent Manager AD2 Intent Manager ADn Intent Manager
• Domain controllers

AD = autonomous domain = Intent interfaces TM Forum, 2022

13 inform.tmforum.org
section 2: what is intent and why does it matter?

domain agnostic. The intent The creation of an ontology

 it’s very different from other
API adapts to domain-specific is central to the modeling telco-focused standards
requirements through intent of intent-driven automation. approaches, which are
model federation. An ontology provides a dedicated to domain-specific
The intent modeling approach is framework representing intent approaches.
different from non-intent modeling shareable and reusable I ntent models and interfaces
in several important aspects: knowledge across a domain. are built around intent
Ontologies go beyond objects and knowledge;
I ntent models are divided taxonomies and other
into a common (or “root”) they require a different
knowledge representations approach than standard
model and extension models. to allow the expression
The common model defines information models and
of relationships and links APIs. Typical telecoms API
all the domain and use case between concepts. In the
independent vocabulary and development approaches
intent-driven telecoms world, are built on Unified Modeling
semantics required to support ontology allows expressivity
intent management, including Language (UML). UML works
of intent. The ontology well for static APIs and use
intent reports. Extension interface is fundamental to
models, on the other hand, by (human) developers, and
expressing intent and to will continue to be used in
extend the intent common connecting the general to
model to domain- and use many telecoms applications.
the specific so that each The intent API, however,
case-specific vocabulary time there is a new service
and semantics. The beauty requires the evolution to
concept or type (for example, something more dynamic and
of this two-tier approach network slicing) to add to
includes simpler and more machine-usable and machine-
the system, you don’t have to adaptable. TM Forum’s ANP
robust implementations of create something customized;
intent, and putting domain- suggests RDF (Resource
you can just extend what’s Description Framework) and
specific extensions in the already in place. The
hands of subject matter its related software stack
process and approach are because these tools are built
experts (such as 3GPP the same regardless of the
members for RAN or mobile for knowledge-based systems:
intent reference point, be they allow for machine
core domain intents). The it business or operational
federation of the common reasoning and logic inference.
support system (BSS/OSS), Although such approaches
and extension models ensures radio access network (RAN)
the adaptability of intent are relatively new to telecoms,
management, SD-WAN they are widely used in other
expression and reporting. connectivity, multi-domain machine knowledge domains.
network slicing, or some other
specific network or service In the next section, we’ll discuss
domain. This is a key part of industry challenges toward
TM Forum’s approach, and implementing intent.

14 inform.tmforum.org
section 3:

challenges to
implementing intent
One large European-based operator noted in an interview for this report that recognizing the intent, and trying to
understand what the customer or network planner wants and formally capturing that requirement, was a bigger
challenge than determining whether or not the intent can be handled and, if so, acting on it – at least at the resource
level, where the CSP has been doing a lot of research and development. Work to formally model intents is underway,
as discussed in section 2, but will take time to socialize and implement.

O
nce intents are properly Marrying intent to action
modeled and expressed,
Of course, there are other tricky
sending them to an intent
cognitive tasks for autonomous
handler to use its knowledge to
systems to master. Tayeb Ben
decide on a course of action and
Meriem, Senior Standardizations
What are lacking
perform that action could result
Manager at Orange, said in an in the industry
in a cascade of intents or API
interview for this report that what today are methods
calls to other systems. It could
also generate tickets to other
are lacking in the industry today and tools to verify
“are methods and tools to verify
systems to dispatch personnel
that intent has been achieved
that intent has
to a site to deploy a switch and
satisfactorily and, if not, to provide been achieved
connect it to a network access
an explanation as to why the satisfactorily and,
interface, and then issue an
API call to configure a virtual
intent is not realizable and report if not, to provide
alternatives to the owner.” He
LAN on the new switch. The
noted that this “why” explanation
an explanation as
cognitive system must solve the
can be complex for tools to create to why the intent is
orchestration challenge of how
and even for people to fully grasp not realizable and
to sequence the actions logically
and physically in time.
and verify. There can be multiple report alternatives
dependencies that are not simple
to interpret and understand. There
to the owner.”
can be many operational layers

15 inform.tmforum.org
section 3: challenges to implementing intent

involved too, with the business layer available tools, the CSP said it in that document. This current
dependent on the lower service could define intents and check process is too rigid and time-
and resource layers, for example. compliance against them, but go consuming, and the policies are too
Another CSP said none of the no further: The CSP could not hard to change. “We want to make
available tools it has tested can operationalize example situations policy management an important
truly close the loop between such as “last night at 12:00 ‘X’ capability overall, not just for us,
intent and action to the extent happened, so the system needs but for our customers to be able to
it wanted. Data center tools are to apply ‘Y’ action automatically.” manage policy directly,” he noted.
more advanced than tools for They could check manually if the
traditional networks, as you would network meets the intent now, but Practical limitations
expect given the relative maturity not trigger actual data analysis
some time in the future; they As is true when CSPs embark on
of intent-driven automation in any major change to operations
IT as compared to telecoms. In could not automate operations in
real time in a closed loop. strategy they must grapple
addition, the CSP notes that data with breaking down both
centers have repeating patterns This CSP wants to move away organizational and technical
of connectivity and operate in a from current processes, whereby silos. Their customers must
controlled environment, hence it needs to create a document embrace the changes as well. The
intent-driven operations are less outlining customer experience or box below shows some of the
complex and easier to implement security policies and then get a challenges they face.
than in telecoms networks. Using developer to code based on what’s

In the shift to intent-driven automation CSPs must:

vendor-specific implementation, Only through seeing the system

it just needs to communicate suggest the right actions
with the vendor network or time after time will telecoms
functional element through an personnel grow comfortable
 rapple with significant
G open, non-proprietary API, which enough to eliminate go/no go
inherent complexity, such as could be non-intent-based. manual decision points. One CSP
how to incorporate resources acknowledged that although
and interfaces that are outside his company has a strategy
the intent-based components to implement AN, its network
of the system. The best way engineers are understandably risk
to encapsulate domains and averse and protective. The notion
subsystems and their existing  hink through the implications
T of going “all in” is too much,
non-intent-based APIs and of the TM Forum Autonomous so the company is focusing on
approaches is to leave them Network Project’s proposal sets of automation changes and
untouched (and unbroken). on intent-driven operation, working to close the loop where
Intent-based APIs can be including the impact of intent-driven they have high certainty as to
implemented at the integration/ approaches on regulatory, legal, how to fix problems that do arise
demarcation points between security and governance imperatives and where the repercussions of a
autonomous domains while to which they may be subject. failure to fix a problem are minor.
leaving the non-intent-based
systems within those domains
to execute appropriately. This
approach is analogous to the
way that vendor-agnostic cross-
domain network orchestrators Learn to trust increasingly more Shift mindsets.
work with vendor-specific automated operations on the Beyond simply learning how to
domain controllers to create way to autonomous operations. trust autonomous systems, there
end-to-end services: the network Delegating responsibility, are more fundamental issues
orchestrator does not need to especially to machines, makes regarding whether organizations
worry about the details of the many people uncomfortable. are ready to leap into the unknown.

16 inform.tmforum.org
section 3: challenges to implementing intent

One vendor interviewed for use new modeling and interface

this report noted, for example, development approaches, such
that it helped a CSP implement as the RDF-based software stack
automated operations capabilities mentioned in section 2, that are
that learned as they went along, built for dynamic, knowledge- Engage more frequently with
and which could automatically based systems. One CSP noted enterprise executives, not just
address 95% of the CSP’s network that it has been building its network engineers, during the
issues. The CSP’s network internal expertise, for example, sales process. One CSP noted
operations teams, however, on software-defined networking that rather than engage with a
would not allow the system into over the past several years; the customer’s network engineers, it
production because they felt service enhancement opportunity would be better to engage with its
they could neither understand this approach has brought is now executives; only then could they
nor control it. Embracing change clear. Even so, he says: “In the fully understand the customer’s
will take top-down support and network itself we are relatively business intent rather than just their
encouragement, education, training risk-averse, so we are building lower-level intents. When a customer
and the creation of opportunities up our internal capabilities wants to set up new office or retail
for ownership of the change. incrementally, cautiously and in an locations, for example, the CSP and
agile manner.” its systems need to understand the
essence of what they are trying to
do, not just have them pick from a
service catalog.

Grow expertise. Knowledge

graphs, ontologies and other Ensure they have the right data
fundamental concepts of available when and where it’s
making intent-driven operation needed. Aaron Boasman-Patel,
a reality have been embraced TM Forum’s Vice President, AI & Consider how to implement
by hyperscalers for some Customer Experience, and the AN greater customer self-diagnosis
years; these concepts are a project lead, notes that “data is of service issues. It would
fundamental part of Kubernetes- the oxygen required to implement truly be a bonus for CSPs and
based cloud computing and intent properly.” Few CSPs today their customers if intent-driven
autonomous system operation can confidently say they have a automation could also expand
in data centers. Because these data strategy in place that will let customers’ ability to self-
concepts are new to CSPs, them easily unlock the door to diagnose and automate repair of
personnel will need to learn to intent-driven operations. service problems.

A learning curve many stakeholders have an interest What about the legal or regulatory
in understanding how the system requirements that enterprises
As CSPs look to build intent-driven
operates, so transparency and in different industry verticals or
systems, they must make sure
clarity are critical for accountability geographies might be subject to;
they have humans at the right
and for establishing trust. how might this new way of obtaining
touchpoints, have established
Of course, it is not just CSPs that services affect those concerns?
the right boundaries around
have an intent learning curve; their CSPs and their customers will need
autonomous systems, and have the
customers must learn to order and to work through these ramifications
right transparent processes in place
manage their communications to get to the ultimate promise of
to grow trust and for governance,
services and engage with service the improved experience, flexibility
conflict resolution and so on.
providers in a very different way and simplicity that intent-driven
Naturally, algorithm developers/
and to trust autonomous systems. processes should bring to customers.
owners won’t want to give direct
access to the algorithm themselves, Customers are not used to In the next section, we’ll look
because that’s where the real communicating with CSPs in terms at industry progress toward
value and intellectual property is. of their intents: Do they even know implementing intent to drive
However, as noted in section 2, how to communicate their intents? autonomous networks.

17 inform.tmforum.org
section 4:

CSPs are making

progress
One example of a product that shows what is possible with an AI/machine learning-driven approach, although not
explicitly an intent-driven one, is the Anomaly Detection Service, jointly developed by Vodafone and Nokia and
launched in July 2021. The service is delivered as-a-service by Nokia to Vodafone on Google Cloud Platform.

T
he capability establishes By moving from its on-premises many times” approach, and a
the root cause for a radio Big Data Platform to the Google- 25-30% operational efficiency
network problem by hosted analytics approach, improvement in anomaly
separating normal and unusual Vodafone has created a “data detection. The system as
behavior, and determines if the ocean” that fixes the inaccuracies launched supported network
unusual behavior is aberrant, that propagated by virtue of its planning and optimization; it
based on unsupervised learning separate operating companies will ultimately support the full
models. If the system verifies a maintaining their own data lakes. operations lifecycle and additional
problem, it suggests a fix to the The as-a-service solution also lets AI/ML-driven automation
humans in charge, who can decide Vodafone shift Capex to Opex applications. According to the
whether to sanction the action. spending and scale spending as it press release, “Vodafone expects
Vodafone is working on closing uses capacity. that around 80 percent of all
loops between detection and As reported in an Analysys its anomalous mobile network
corrective action where the risks of Mason case study, Vodafone issues and capacity demands
letting the system take control are saw a 60-70% reduction in will be automatically detected
relatively low and where it has high effort through implementation and addressed using Anomaly
confidence in root cause analysis of a “develop once, deploy Detection Service.”
in determining the fix.

18 inform.tmforum.org
section 4: CSPs are making progress

Intent-driven autonomous networks for smart mobility

TM Forum Catalyst proofs of opportunities enabled by 5G in the topology to a toll road’s geo
concept are proving to be a good vehicle-to-everything (V2X) smart and traffic data; self-configuring
avenue for industry players to mobility context. infrastructure resources based on
collaborate on implementing The team developed an Intent traffic data; self-healing to recover
intent-driven automation Management API specification from network issues; and self-
applications and to test the for expressing business intent reconciling of credits given by
TM Forum ANP’s intent proposals specific to their smart mobility the toll road operator to vehicle
in real-world use cases. applications. The applications owners that participate in a flow
In one Catalyst project members considered the intents of four control scheme.
set out to accelerate the expression different intent owners – the
of intent and its related automation toll road operator, the vehicle Watch the video to find out
at the business layer. It addressed user, the CSP and the vehicle more about the Catalyst Project.
two use cases: road traffic manufacturer – in the context
management using coordinated of a complex multi-partner
driving of autonomous vehicles, ecosystem. Using control loops
and the related environmental at various layers it demonstrated
management of road traffic at multiple dimensions of AN “self-X”
a municipal level. Its ultimate operation: self-onboarding by toll
business goal was to harness road operators, vehicle OEMs and
intent-driven autonomy to optimize vehicle owners; self-provisioning
both cost and monetization based on mapping network

5G accelerates smart on the expressed business intent. create slices for enterprises. The
aviation The Catalyst’s architecture mapped implementation included assurance
well to TM Forum’s Open Digital of the differentiated services and
In another Catalyst project
Architecture domains and used dynamic charging based on a
members built on a previous
a number of the Forum’s Open user’s experience. Through an
phase addressing the potential for
APIs. For the customer loyalty ordering portal, the customer could
using 5G in the aviation industry.
use case, the airport or airline choose slice service details, options
They tested automation based
can provide VIP passengers with and performance requirements,
on two business intent-driven
a better 5G communications generate a price and affirm the
use cases in the event that a
experience by adjusting or slice order. Closed-loop automation
5G network slice hits capacity:
enhancing their service. For the maintained SLAs and network
strengthening customer loyalty
connectivity revenue optimization resource optimization, using AI-
in the face of flight delays;
use case, an airport’s revenue and based predictive assurance to
and optimizing revenues from
profitability can be optimized trigger remediation as needed.
connectivity services that the
by automatic rebalancing of 5G Finally, the enterprise controlled
airport provides to users.
traffic between a slice supporting lifecycle management of the 5G
This phase of the Catalyst focused slice and charging policies through
revenue-generating services and
on allowing the enterprise to its expressed business intent.
one supporting airport operations
control the lifecycle management
based on network traffic analysis. Watch the video to find out
of 5G slices and related charging
Both use cases allow the enterprise more about the Catalyst Project.
policies based on business intent
to control its services based on its
captured through a customer
expressed business goals.
portal. It included closing the loop
between service orchestration, The Catalyst team implemented
service assurance, and network an automated model-driven
automation and orchestration solution that exposed 5G network
processes to guarantee SLAs and slice-as-a-service offerings and
optimize network resources based generated service orders to

19 inform.tmforum.org
section 5:

Where is intent-driven
automation heading?
TM Forum’s IG1253 proposal suite of documents lays the groundwork for standardizing the implementation of intent
in autonomous networks. Through the Autonomous Networks Project (ANP) team’s collaboration, the Forum will
soon issue the first release of the intent-related open API implementation. It will also continue to refine the intent
common model, add to the initial set of domain-specific extension models that vendors and operators can use, and
incorporate industry comments in the implementation documentation.

A
NP is preparing IG1253E native operations. Most experts to be standardized. “Open
to document intent in agreed that the opportunity APIs are the ‘three pin plug for
real-life use cases such for and impact of intent-based standardization’,” says TM Forum’s
as service order automation. networking are significantly CTO, George Glass: with open
Domain-specific models, such as better where a programmable, APIs, a CSP can abstract, expose
for slicing, are under development remotely-configurable and and control the capabilities
by complementary standards cloud-based virtualized network irrespective of the vendor.
development organizations. is in place. Network-as-a-service Experts also raised the issue of
(NaaS), intelligent infrastructure, what is likely to be the harder
When to implement? orchestration software and policy- industry challenge: expressing
based action are all precursors to intents and breaking them
One interesting discussion point intent-based networking.
that arose during interviews into sub-intents to send to
for this report was whether It’s difficult to implement intent- specific handlers for action,
it was too early to consider based networking without the or acting autonomously on
implementing intent-driven ability to automatically spin up/ the intent. The consensus was
automation because the industry down network capabilities, and that reaching highly and fully
is still in the early stages of interconnectivity, management autonomous networks would
the move to virtual and cloud- and control interfaces need be a real challenge. “Trying to

20 inform.tmforum.org
section 5: where is intent-driven automation heading?

get machines to make cognitive use intent as a mechanism to

decisions will make intent look build control loops and operate
easy…but training the machine systems. He says: “We expect
learning will only be as good as very similar needs with respect
the requirements and data sets to autonomous systems.” He
input,” says Glass at TM Forum. adds: “There are established
Trying to get Implementing intent-based technologies such as ontology-
machines to make autonomous operations in an based knowledge graphs and
application or domain here or machine reasoning that are not
cognitive decisions
there, as in Levels 2 and 3, will adopted by telecoms standards
will make intent be one thing, but federating and systems yet, although they
look easy…but autonomous domains to allow are mature and well understood
a broad scope of autonomous and used elsewhere.” The
training the machine
operations will likely be a long, ANP proposal adopts these
learning will only incremental process. approaches because they will aid
be as good as the Jörg Niemöller, Expert of implementation.
requirements and Analytics and Customer “An autonomous system
Experience at Ericsson and must be able to consider the
data sets input.”
IG1253 guide lead and author, consequences of its actions
suggests studying examples of with respect to their utility,”
how systems outside telecoms says Niemöller. “It also needs

intent expertise-building Aspirational autonomous

operations example:
and aspirations streaming service intents
and conflict resolution
Set a customer experience (CX)
threshold for streaming in country X.
If experience goes under the stated
threshold, automatically buy more
Pure bandwidth from provider Y
research
As needed, be able to easily change
the stated threshold for a specific
Standards event (such as the Olympics)
activity Set a commercial requirement, such
as “ensure a profit of Z%” for the
streaming service in country X
If there is a conflict or collision
between the CX policy and the
Practical commercial policy, or there is a
3-pronged higher probability that something
implementation
approach will break if one action is taken
rather than another one, there needs
to be a process for resolving that
conflict as seamlessly as possible
Grow and evolve the implementation
over time to cover more complex
situations

TM Forum, 2022

21 inform.tmforum.org
section 5: where is intent-driven automation heading?

to evaluate advantages versus to be poor for the domain or matured over time. Once that
disadvantages and judge the risk application the CSP wants is in place, the CSP can move
involved.” In short, the system to automate, it will be more up to higher levels of intent.
must understand and act, not just practical to start elsewhere, Start at the business layers

execute pre-defined policies, in where the data has already and work down. If risk
order to be truly autonomous. been scrubbed and where aversion to system autonomy
Kevin McDonnell, Huawei’s reliable telemetry-based is a significant concern among
Senior Director, Intelligent interfaces and other quality network operations personnel,
Automation, and one of the data sources are in place to building and expanding
IG1253 authors, notes that from support dynamic operation. autonomous systems in the
Huawei’s perspective “intent  tart with a persona-based
S business or service layer
isn’t premature, it’s overdue.” set of service packages that may be perceived as less
Huawei believes it will take three can be expressed as intents. risky by staff because intent
to five years for the industry to NaaS has been implemented as a driver of IT system
get to Level 4 of AN evolution, less like a completely flexible autonomy is a more mature
but intent can and should be set of service building blocks concept. Indeed, automation
applied at Levels 2 and 3 – partial and more like a finite menu technologies are already more
and conditional autonomous of service options from which established in IT domains, and
networks, respectively – to help a user can choose. Intent systems that rely on them and
speed things up. Therefore, with could be implemented as a their use can be extended to
industry standards in place, set of service packages – for more extensive closed-loop
Huawei believes that intent could example, for homeworkers, operations. More abstract
reach an initial commercial stage gamers, families and so on – expressions of intent will be
in two to three years. to bound the almost infinite required at the business and
The graphic on p.21 represents set of options and make the service levels than at the
one CSP’s approach towards sales and ordering process resource level. IT and network
building intent expertise as well easier for both buyers and people can be cross-trained,
as an aspirational autonomous sellers. as CSPs have done in other
operations goal. Start bottom-up at the
 contexts, to learn from each
resource level and within other and move the whole
a specific domain or operations organization
Taking the first steps forward.
organization. Virtually all
There is no single recipe that the people interviewed for Follow the top-to-bottom

provides a step-wise approach this report suggested this tool availability. Tayeb Ben
to intent implementation that approach, although it does Meriem of Orange suggests
fits all CSPs’ business situations. assume that the NetOps that since good tool chains
However, there are some solid organization is ready to move for translating and validating
approaches that many CSPs beyond its understandable intent are only slowly coming
should consider: risk aversion and fear of lack to market, they are likely
I mplement first where data of control. To translate intent to be available first where
are available, abundant and into action the system must the application process
“clean”. Good, believable understand what someone’s implementation complexity
and available data are trying to do, decide on and is relatively low and the
needed to drive autonomous implement an action, and application sits within a
action based on intent. monitor for change over time specific domain (e.g., the 5G
Either the CSP has clean to decide if further action is mobile core network, or 5G
data collection, storage and required. Starting closer to radio access network), and a
analytic capabilities before the resources strikes many single CSP’s purview. This is a
implementing intent-based people as an easier approach: variation on the “implement
automation, or it must take the lower-level intent is more in a particular domain or
the time to source and scrub straightforward to implement, organization” strategy in that
the right data as a preliminary essential to make the “good tools” imply an end-to-
step. If data hygiene is judged automation work, and can be end, top-to-bottom closed-

22 inform.tmforum.org
section 5: where is intent-driven automation heading?

loop capability that spans Finally, suppliers can do the

business, service and resource work to prepare their solutions
operations rather than just now so that backend workflows,
an intra-layer implementation orchestration and other
approach. processes are in place to drive
automation. ServiceNow, for
Extend an earlier zero-touch
 There is no example, is doing this in the order
automation project to add
intent- and AI-driven closed- single recipe that management domain. While
loop operation. This strategy provides a step- doing the background work to
generate orders, it is working to
may be the most pragmatic wise approach understand intent as expressed
of all. The Vodafone anomaly
detection example discussed in to intent in customer interactions – such
section 4, and the Windstream implementation as with a virtual agent or other
customer interaction – so
Project Kailash case study that fits all it can support intent-based
and Vodafone Ready Network
implementation discussed in CSPs’ business ordering in the future. If the
section 1, offer examples of situations. service interaction is driven
capabilities and service offerings by a virtual agent, the entire
that could be extended into engagement could be touchless
closed-loop autonomous when customers are ready for
operations over time. Another helpful approach for implementation.
vendors is to add intent – or at So where will intent ultimately
CSPs will likely take several of these least analytics- and AI/ML-driven
approaches to implementing intent make the biggest, most positive
operation – explicitly to their new business impact? Ben Meriem
to drive autonomous systems. product roadmaps. For example,
Whatever path or paths a CSP of Orange expects it will be
Ericsson announced a Dynamic wherever there are touchpoints
decides to take, however, it would Network Slice Selection Solution
be unwise to expect a “rip and for service instantiation and
in January 2022. It allows delivery to the customer, both
replace” strategy. customers to set up access to external and internal to the CSPs’
one or more 5G network slices own systems. This could include:
Moving forward based on application – such as at customer-CSP interaction
Funding regional or other standard mobile broadband, points, such as business
research standardization efforts, gaming or secure enterprise portal interactions and virtual
proofs of concept and pilots are connectivity. The slice requests agent-assisted interactions; at
valuable avenues of collaboration. today are handled in a rules- customer-CSP network/private
CSPs will prefer vendors that are based, programmatic way, but cloud demarcation points;
aligned with them in terms of when asked by an analyst to and at CSP and public cloud
open APIs in particular, because comment on driving the solution demarcation points. In other
such interfaces can encapsulate with intent, Ericsson noted words, at points at which assets
the complexity at the lower levels that there is no reason it could could be monetized.
in brownfield situations where not integrate analytics-based
automation for SLA enforcement In the final section, we give
intent-based interfaces and easily guidance to help CSPs implement
programmable infrastructure may and other operations in the
future. Indeed, the vendor their visions of intent in
be absent. Supplier alignment autonomous networks.
with the broader TM Forum highlighted intent-based network
ANP approach will also ease operations as one of the three
integration and will therefore be AI R&D thrusts for its managed
valuable to many CSPs. services organization.

23 inform.tmforum.org
section 6:

make it happen –
strategies for embracing
and implementing intent
All communications service providers (CSPs) aspiring to implement autonomous networks will need to embrace
intent. Implementation can be done incrementally to build trust and expertise and to break down organizational
resistance. CSPs will have to educate their customers as well so that their business intent can be properly captured
and managed through its lifecycle. Here we outline some overarching steps they can take.

Fragmentation of efforts to conditions that might change.

implement intent will just draw The importance of data cannot
out the timeline. The more CSPs be overestimated. Data strategy
stay informed and get involved implementation is a precursor of
with influencing the velocity of intent in autonomous networks.
standards innovation the better,
because it will help suppliers
prioritize their R&D investments.

Collaborate, collaborate,
collaborate
Intent is a big ship to navigate;
a lot of industry thrust will be
needed to propel telecoms
towards the goal of autonomous
networks. Continued collaboration Test and try targeted
is essential between members deployments
of TM Forum’s Autonomous
Networks Project (ANP) and other Clarify, implement and In section 5 we outlined multiple
pragmatic approaches to intent
Forum project teams; between TM extend your data strategy implementation, once the needed
Forum ANP and other standards data for closed-loop operation are
development groups also working If data can’t be observed,
measured and controlled over at hand: persona-based packaging;
on intent-driven autonomous bottom-up narrow deployments;
systems; through regional R&D time, it can’t be used to drive the
autonomous systems and verify top-down narrow deployments;
efforts; and between CSPs and tools-based top-to-bottom
their suppliers working together. proper operation under dynamic
deployments; and extension of live

24 inform.tmforum.org
section 6: make it happen – strategies for embracing and implementing intent

automation instances to closed- of intent through designing and

loop operation. Many CSPs will architecting their systems to
employ multiple strategies at once evolve to increasing levels of
across different projects. The key automation and autonomy. They
is to pick projects with low risk and should also explicitly add intent to
relatively high reward so that the their professional service offerings
organization can grow its expertise and product roadmaps to signal
while learning to trust the systems its priority and stimulate customer
to do the right thing. discussions.
CSPs, too, must prepare their Anticipate and manage
soil by building staff comfort broader intent enablers
with the concepts of intent-
driven autonomous systems, and Intent will require many changes
giving them time and training to CSPs’ operations status quo.
to understand how the systems Roles and responsibilities and
operate and how they fit into customer engagement processes
making the implementations work will all need to be revamped. CSPs
to further the business goals of must anticipate and investigate
their company. Nothing can sink legal, regulatory and governance
a significant transformation effort implications. If not properly
Prepare the ground for addressed, these changes as well
such as intent-driven autonomous
planting networks more effectively as the staff mindset changes that
Any gardener knows that healthy than staff who aren’t properly CSPs must nurture will thwart
crops depend on healthy soil; it educated, trained and engaged in successful technology investments
makes no sense to sow expensive the success of its outcome, and and deployments.
seeds in poor soil. Vendors and an executive team that doesn’t
other suppliers can prepare the provide proper direction and
ground for planting the seeds encouragement.

25 inform.tmforum.org
additional features
& resources
27 | Huawei’s Best Practice on Intent-driven Networks

31 | The foundation for 5G-era automated operations: Oracle’s intent-driven automation solution

34 | ServiceNow and Accenture: Shifting the conversation of intent beyond network orchestration to
service orchestration

36 | TM Forum Open Digital Framework

37 | TM Forum research reports

38 | Meet the Research & Media team

26 inform.tmforum.org
SPONSORED FEATURE

Huawei’s Best Practice

on Intent-driven Networks
The intent-driven approach is a key characteristic of the autonomous
networks. It will make the network operation system be able to adapt
to the business objectives of the operator as well as the expectations
of customers and users. The role of intent is to communicate all these
expectations to the Autonomous Networks.

T
he Huawei Autonomous digital transformation. One key in TMF IG1251 AN Reference
Driving Network (ADN) differentiation is in the way ADN Architecture (as depicted below).
solution has adopted make use of the AI enabler for This type of modeling architecture
intent-driven network a one-stop AI development and lends itself to more flexibility
architecture through its close operation platform to manage and speed in AI development
collaboration in the TM Forum’s intent objectives incrementally that allows adaptation of
Autonomous Networks Project and easily. The concept involves more AI capabilities to any
to ensure an open, standard the ADN’s component of the of the appropriate layers in
and consistent method of iMaster NAIE which adopted order to influence higher
implementation. According to a 3-layer AI model inference quality performance of intent
the TM Forum’s IG1253, a key framework that is unique in the applications to meet objectives.
property of the intent-driven industry to offer specialized It is considered the ideal for ADN
communication mechanism is intelligent functions best executed to evolve, grow and improve its
that it is infrastructure agnostic as site intelligence at the NE intent functions.
and portable. It will enable layer, network intelligence at the
decoupling between systems autonomous domain layer, and Our guiding principle is to adhere
at different layers. It applies in the cloud intelligence platform to the iterative optimization
incremental implementation as the service and business process to introduce the intent-
of intent that show positive operations layer. The architectural driven extensibility to provide
reinforcement to future-proof design fits perfectly with the more network autonomy scenarios
the journey to full autonomous TM Forum’s model as described to help operators and enterprises
transform and upgrade to digital

27 inform.tmforum.org
SPONSORED FEATURE

intelligence in the pursuit of a and service configurations but misalignment of proposals at

self-fulfilling, self-healing, and self- also on all of the monitoring each standards organization.
optimizing autonomous networks. rules automatically and, while Huawei has been working on ADN
It provides for a much more all the time, coordinate in real and Intent APIs with customers in
practical form of operationalizing time with the AI engine to find different scenarios, where incident
intents on networks consisting of root cause and make smart management and 5G energy
both legacy and next generation as decision automatically for the full saving are two typical use cases.
in many of the operators’ network lifecycle. Based on the closed-loop
landscape. The intensification of feedback, the intent management
intent-driven capabilities can be engine can adjust the policies Incident Management
continuously upgraded through to ensure the integrity of the Use Case:
level and value evaluation of customer intent. Traditional alarms are generated
the case scenarios as networks Huawei is currently actively when the system detects an
undergo autonomous phase engaged in TM Forum to unexpected status, but not all
evolution. As networks scale and define the API Map of Intent faults in the system generate
move up the autonomous network for autonomous networks, and alarms. An incident is generated
level (ANL), some scenarios of associate the map with the based on the AI algorithms by
use cases deployment will identify autonomous network levels in the analyzing system alarms, KPI
mandatory requirements for future. At the same time, Huawei events, logs, etc. The reason
what intent needs to achieve. is participating in IETF standards why intent API passes the intent
For example, at ANL 3, the API on the work in-progress to define correlated incident based on
model abstraction mandates the intent-driven network assurance, maintenance intent instead
API mapping to convert customer and has further ramping up of alarms is that alarm types
intent to associated provisioning efforts to define the intent-driven are defined by devices based
policies but not necessarily on its network management services on their detection capabilities
monitoring rules. At ANL 4, the in 3GPP and ETSI. Collaborating and detection levels. An issue
intent engine not only translates amongst these SDOs would exists as different vendors
the intent relation of network safeguard any overlapping and have different definitions

28 inform.tmforum.org
SPONSORED FEATURE

for the alarms so that as the MTTK (Mean time to know) by at networking structure and traffic
network topology and device least 30 minutes. mode. This would result in the
types change, the definition of OSS and vendors systems being
alarms keeps changing as well. 5G Energy Saving Use Case: deeply coupled, making the
Consequently, OSS needs to interconnection difficult and the
learn new alarm rules. In the context of global warming, adaptation period too long. By
base station (eg. eNodeB) energy using the energy-saving intent
In the live-network deployment, a saving is an important component
reliable fault pattern database is interface, operators only need to
of OPEX saving. While wireless formulate basic energy-saving
needed by aggregating historical networks involve the coexistence
alarm data, performance KPIs policies based on services and
of 2G, 3G, 4G, and 5G base energy-saving requirements such
of the live network, network stations from multiple vendors,
topology, and service bearer path as requirements on KPIs, expected
different base station vendors energy saving time ranges, etc.
information by utilizing frequent provide multiple energy saving
pattern mining, hierarchical The energy-saving system of the
solutions. To achieve the best vendor automatically formulates
clustering algorithm for AI energy saving effect and maintain
modeling and inference. A well- refined energy-saving policies
stable network KPIs, the system based on operators’ energy-
trained model can accurately needs to formulate detailed
identify the matching fault through saving intentions.
energy saving policies based on
fault propagation diagrams and complex inter-RAT networking In the live-network deployment,
machine learning. As a result, and network traffic modes. In based on operator’s energy-
this helps decision-making in the addition, the energy saving saving intentions, the PowerStar
downstream process based on policy needs to be dynamically automatically generates energy-
autonomous level capabilities of adjusted with the changes of the saving policies based on the
user’s network (Manual/Semi- networking structure or traffic network topology and historical
automated/Full-automated). mode. If there is no energy-saving traffic, and learns the optimal
The solution has been deployed in intent interface, the OSS needs settings of BTS energy-saving
China Mobile where we achieved to understand the energy-saving parameters based on network KPI
decrease of the number of trouble implementation details of each feedback during energy-saving.
tickets by 40% and reduction of device vendor and analyze the The AI load prediction algorithm

29 inform.tmforum.org
SPONSORED FEATURE

predicts network load changes development around the world.

and adjusts the energy saving The overarching outcome is
duration in real time. The energy- about giving operators full
saving policies supported by control over how and when
these AI algorithms achieve an to accelerate the autonomous
optimal balance between network network’s innovation process that
energy-saving effects and KPIs. eventually will deliver faster, cost
This solution has been deployed effective and freedom in user
in China Telecom which has experience. Intent operation takes
benefited 4G power saving of 8.6+ the form of software-enabled
million kWh/year and 5G power automation process to bolster
saving of 6+ million kWh/year, intent-driven intelligence analytics
a comprehensive energy saving and orchestration for ongoing
efficiency improvement of 12%+ assurance of the promised
for 4G sites and 20%+ for 5G sites. business outcomes throughout
From the above practices, the the intent lifecycle. Underpinned
intent API will accelerate the by Huawei’s Communications
Autonomous Driving Network’s Network 2030 strategy, its aim is
innovation process. These invaluable to deliver a zero-touch, zero-wait,
experiences and joint innovation and zero-trouble user experience.
projects with other operators For further information on
have enabled ADN to build better Huawei ADN intent-driven
solutions to support the high solutions and products, please
value cases of digital strategy of visit our ADN content hub at
operators and take the lead in the https://carrier.huawei.com/en/
growing trend of digital economy adn.

30 inform.tmforum.org
SPONSORED FEATURE

The foundation for 5G-era

automated operations: Oracle’s
Intent driven automation solution
Intent driven automation can bring a fundamental change to a intent of the customer in the
Communication Service Providers (CSP) operations model. The form of preferences, defined
traditional model has been mostly bottom-up and, the dominant as parameters against the
approach has been to build out the network, create service offers and associated products in the
products customers would procure those services with some level of product catalogue.
configurability, but they were largely constrained by the limitations of Service Design and

network capacity. To prepare for peak demand scenarios, CSPs often Orchestration: The business
overprovisioned the network which meant there was a massive amount intent is then translated into
of stranded CapEx. service and network intent

W
ith telco cloud and management framework. Each in the Service Design and
software defined of the solutions are a collection Orchestration platform and
networking, we can of platforms that connect the is typically captured as the
reverse this model – service customer to the network, as desired end state of the service
demand and business objectives illustrated in the chart below. and network. An internal intent
can drive the shape and form of engine computes the steps to
The four solution components are: go from the current to target
the network. This is where Intent
is critical and why it can be so Digital Experience for
 state. It is akin to using a GPS
powerful. With top-down Intent Communications: Today’s where it lets the users specify
driven automation CSPs can customers expect personalized the destination and the system
achieve a highly efficient network interactions from the time computes the best possible
and network operations, as well as they start learning about a route to get there. The user
deliver superior on-demand digital company’s offers to when they can also specify the service
experiences to customers. order service to when they and network constraints, for
need care. Digital Experience example, what the service
for Communications starts performance should be, and
Oracle’s value proposition from the perspective even which data centres
for Intent driven automation of experience for these should the resources reside in
Oracle offers four broad solution customers, both consumers etc, like the travel conditions
areas that work together to deliver and enterprises. As the primary we would typically place on
intent driven automation for customer facing platform, the GPS, such as quickest
CSPs, and maps to TMF’s intent it captures the business route or the shortest route.

31 inform.tmforum.org
SPONSORED FEATURE

Oracle’s Intent driven automaon soluon

Business Intent Digital Experience for Communicaons

Service orchestraon & Assurance

Service Design & Orchestraon Service Assurance
Intent engine Visualizaon
Service Intent Service
Network
& Service Universal
Design & Policy & Rules Inventory Fault + Perf. Mgmt
Topology
Assign
E2E Service Orch. ML/Event analycs

Network Intent Network orchestraon & Automaon

Service acvaon Configuraon management NFV Orchestraon SDN Orchestraon

SDN 5G
Communicaons & IT Tradional
Cloud
Network Infrastructure Physical NFV Virtual OTN

3rd party components

The below chart illustrates from the original intent, and For use cases such as Network
a top-down intent-driven action must be taken to as a Service or 5G slice-based
orchestration flow. course correct. You could services, closed loop automation
Network and Service
 loosely relate this to the traffic can be highly valuable for CSPs
Inventory: the ability of service congestion or incidents in the to achieve top-down intent driven
orchestration to translate GPS analogy – an intelligent operations delivering superior digital
intent into accurate service navigation system offers experiences with lower opex. The
and network configuration alternate choices to the combination of service orchestration,
actions relies on an accurate destination to overcome the assurance, and inventory, together
view of the current state of evolving traffic conditions. with domain level network
the network. In other words, Oracle’s recent acquisition, orchestration platforms forms
and in the GPS analogy, you Federos, provides unified the basis to achieve autonomous
first need to know where you assurance solutions, that operations in the 5G era.
are currently before starting performs exactly this function.
the journey towards the Federos ingests diverse data Modular and cloud native
destination. The service and sources such as network
telemetry, events, and logs, Oracle’s intent driven automation
network inventory provides solution has been built using
an ‘as-is’ end-to-end visibility and normalizes and enriches
it with other data sources a highly modular architecture.
of the current state of the It allows the deployment of
network resources and such as inventory data and
intent parameters. It then the modules in a best of breed
services, which is used by environment, integrating into
service orchestration to build a applies machine learning
based event analytics and the CSPs’ choice of systems
journey plan to reach the end using Open APIs proposed by
network and service state as real time topology to perform
root cause analysis. Based on various industry bodies such as
defined by the intent. TMF, ETSI, MEF, 3GPP, etc. The
the outcome of this analysis,
Service assurance (Federos):
 it can drive policy-driven solution has been rearchitected
Once the service is configured actions back into the service to be cloud native so CSPs can
and activated as per the and network orchestrator to achieve all the business benefits
intent, the service and reconfigure resources, enabling around IT agility and minimize
network performance must both guided and closed loop impact due to planned and
be monitored for deviation automation. unplanned outages.

32 inform.tmforum.org
SPONSORED FEATURE

Intent driven Service Orchestraon

CX Product
Order Business Intent
Orchestraon

• Processes Product Orders created in CX

Product

• Dynamically orchestrates orders across commercial systems

• Maps Product Order lines to Service Order lines

Service
Order

Service Intent
Service Design
Orchestraon

• Processes Service Orders as (service) acons

• Designs new Service Configuraon represenng intended service state
• Analyses Service Configuraon changes and calculates (technical) acons for down-
stream systems to align network with service intent
Technical
Order
Service Delivery
Orchestraon

• Dynamically orchestrates technical acons (packaged as Technical order lines) across

an arbitrary set of delivery systems
• Maps operaons/parameters from order lines to appropriate API payloads Network Intent
APIs
Orchestraon

• Performs lifecycle operaons on managed enes using the logical, physical or virtual
Network

abstracons of network configuraon exposed by the API of each delivery system

• Converts vendor agnosc API payloads to vendor specific configuraons

Network Configuraon Protocols

Oracle and TM Forum: As the TM Forum has evolved Conclusion

A History and Future of to the Open Digital Framework,
In summary, the Oracle suite of
Collaboration Oracle has continued to align with
solutions comprising of Digital
these standards and contribute
Oracle has a rich history with Experience for Communications,
to their definition. A signatory
the TM Forum dating back to Service Design and Orchestration,
to both the Open API Manifesto
its early days in the 1990s when Service and Network and
and the Open Digital Architecture
Oracle laid the groundwork for Service inventory, and Service
Manifesto, Oracle is committed to
the development of TM Forum’s Assurance, provides an end-
help CSPs operationalize intent
widely adopted information to-end continuum of intent
driven automation by adopting
framework (SID) and solution across business, service, and
these principles throughout
certification (eTOM) and network enabling CSPs to deliver
architectural design. Recent Open
achieved the first-ever TM Forum superior digital experiences,
API certifications include Service
solution certifications for two reduce opex and increase
Inventory Management (first to
solutions spanning business and operational efficiencies. With
certify), and Resource Inventory
operations support systems. the Oracle solutions discussed
Management for Oracle’s 5G-ready
Oracle sits on the TM Forum above, CSPs can truly embrace
unified inventory management.
board and many subject matter and operationalize intent
experts participate in working driven automation and build
groups, SpecJams, and thought the foundation for 5G-era
leadership activities. automated operations that takes
full advantage of the inherent
capabilities of telco cloud and
software defined networks.

33 inform.tmforum.org
SPONSORED FEATURE

Don’t care how.

I want it now.
Shifting the conversation of intent beyond
network orchestration to service orchestration

In 1971, the world fell in love with Transformation: The software-defined networks, are
Willy Wonka and his mystical triggering a massive shift in how
chocolate factory. Veruca Salt
shift from IT to network digital services are valued from
was the golden ticket winner who For the last several decades, the customer’s perspective.
sang the infamous lyrics “I don’t CSPs have focused their
care how; I want it now.” transformations efforts primarily in
the IT domain. When the European Redefining service value
Fast forward fifty years, and
Telecommunications Standards Not long ago, CSPs were
the tune isn’t far off from how
Institute (ETSI) first advocated for differentiating their products
the industry is defining intent.
network functions virtualizations and services primarily on pricing
Meaning, customers today are
(NFV) back in 2012, it was the and packaging. The commercial
less concerned ‘how’ their service
first time the industry shifted the attraction for a customer was
works and more focused on what
transformation narrative from the ability to buy a product or
they want, and how soon they can
being an IT-centric conversation, service in different packages, at
get it. What they want could be a
to a network conversation. CSPs different rates. With transformation
better video streaming to watch
across the globe are investing moving towards the network, what
a sporting event, or a better
billions of dollars in infrastructure once was a pricing and bundling
connection for Zoom calls from a
that will power next generation conversation is evolving to a service
home office.
digital economies. For example, experience conversation.
While Communications Service U.S. operator Cox Communications
Consider a consumer customer
Providers (CSPs) are keenly announced in February 2022
experiencing a slow internet
aware of the shifting expectations its plans for a multibillion-dollar
connection at their home.
of today’s digital customers, investment in network and product
the solution is not so simple. It upgrades. This comes in addition The customer would likely engage
involves a different mindset and a to their previous $19 billion invested their local CSP via a self-service
new strategy for transformation, over the past ten years, according to portal (e.g. virtual agent) and the
one less concerned about IT and the company. conversation might go like this, “I
more focused on the network. am having choppy Internet service
These investments, along with
at my house.” As the interaction
new technologies like 5G and
continues, the customer might

34 inform.tmforum.org
SPONSORED FEATURE

share all the ways his family uses

Internet services, “All of my kids Recommend
Knowledge Assurance
stream Netflix on their devices, and
Decisioning
I work from home so I need Zoom
Workflows
to work without any interruption.” AI/ML Discovery
Service Catalog
In this example, the interaction
NLP/NLU
started with an issue (e.g. case). But Decomposition

as the conversation progressed, the Notification

Fulfillment

customer’s intent was deciphered

to wanting faster internet service
User Domain Orchestration Domain Implementation Domain
to meet his and his kids’ needs.
Now instead of a service issue, it Figure 1: ServiceNow model of key domains required to enable intent-based engagement
becomes an upsell opportunity for
the CSP. Understanding intent as
part of the customer interaction To do this requires looking workflows that span across the
helps reduce risk of churn while at intent across three key entire organization, helping
delivering value to the customer. domains: user, orchestration, and ensure services get delivered
Now imagine an enterprise implementation. quickly and efficiently.
customer wanting SD-WAN
services for multiple office User Domain Implementation Domain
locations.
Engagement starts the moment Once customer intent is
Underneath this complex service a customer interacts via the deciphered, the orchestration
order lies a combination of different channel of their choice. The domain needs to interact with
network services that need to data from that initial interaction the implementation domain
be identified and explored, and can be leveraged to assess to deliver the intent. This
it is likely partner services are the context of the intent, the interaction could be based on
required to deliver the service services the customer already either fulfillment and service
to all the locations. As service has, any past account history, delivery, where a customer’s
complexity grows, it is becoming etc. Technologies like natural intent results in an order and
more and more critical for CSPs language processing and instantiation of the right service.
to understand the experience understanding (NLP/NLU) as The interaction could also
customers are expecting from their well as knowledge management happen with the assurance
service and break that down into are used to deliver the right domain, where AIOps could be
a series of tasks that need to be insights at the right time, to the leveraged to ensure the service
completed, both from the CSP and right teams. (e.g. working with downstream
its partners. monitoring systems).
Orchestration Domain
I ngredients for intent-driven Embracing intent
With the customer’s intent
engagement
known, the necessary Intent based engagements and
In today’s digital world, CSPs can no workflows and tasks now networks are here to stay and
longer wait for a customer to make need to be dynamically cannot be ignored. At ServiceNow,
a request. They need to anticipate orchestrated. Reference we see the massive potential for
what the customer is going to ask to these workflows are CSPs to embrace the power of
for and, in turn, ask more intuitive driven by the service and intent to take customer engagement
questions. The goal is to make product catalog, and to the next level, which will require
customer interaction as efficient as include decomposition and a strategy focused on both network
possible, focusing on addressing their orchestration, validation and and service experience. Doing so will
needs more quickly to accelerate notification to the right teams, truly help CSPs transform, both their
service delivery. CSPs will need both internal and external. networks and the experiences they
to discern the customer’s intent, Where possible, automation intend to deliver.
inherently knowing the business and needs to be leveraged. To learn more about ServiceNow’s
service outcome the customer is Artificial intelligence (AI) and Telecommunications solutions,
looking for and translating that into machine learning (ML) can be visit servicenow.com
an actionable service request. applied to the orchestration

35 inform.tmforum.org
tm forum open
digital framework

A blueprint for intelligent operations fit for the 5G era

The TM Forum Open Digital Framework (ODF) provides a migration path from legacy IT systems and processes
to modular, cloud native software orchestrated using AI.
The framework comprises tools, code, knowledge and standards (machine-readable assets, not just documents).
It is delivering business value for TM Forum members today, accelerating concept-to-cash, eliminating IT &
network costs, and enhancing digital customer experience.
Developed by TM Forum member organizations through our Collaboration Community and Catalyst proofs of
concept, building on TM Forum’s established standards, the Open Digital Framework is being used by leading
service providers and software companies worldwide.

Goals of the Open

Digital Framework
The aim is to transform
business agility (accelerating
concept-to-cash from 18
months to 18 days), enable
simpler IT solutions that
are easier and cheaper
to deploy, integrate and
upgrade, and to establish a
standardized software model
and market which benefits
The framework comprises TM Forum’s Open Digital Architecture all parties (service providers,
(ODA), together with tools, models and data that guide the their suppliers and systems
transformation to ODA from legacy IT systems and operations. integrators).

Open Digital Architecture Transformation Tools Learn more about

Architecture framework, Guides to navigate digital member collaboration
common language and transformation
If you would like to learn
design principles Tools to support the
more about the Open Digital
Open APIs exposing migration from legacy
Framework, or how to get
business services architecture to ODA
involved in the TM Forum
Standardized software Maturity Tools & Data Collaboration Community,
components Maturity models and please contact George Glass.
Reference implementation readiness checks to baseline
and test environment digital capabilities
Data for benchmarking
progress and training AI

36 inform.tmforum.org
tm forum research reports

knowledge knowledge knowledge

REPORT REPORT
REPORT

improving

verticals
enterprise
the growing
importance of

learned
lessons data
governance
on the journey
Authors:
Mark Newman,
to cloud native Author:
Dean Ramsay,
placing the
right bets
Chief Analyst Principal Analyst
Dean Ramsay, Editor:
Principal Analyst Author:
Ian Kemp,
Dawn Bushaus, Mark Newman, Managing Editor Author:
Author: SPONSORED BY
Contributing Analyst Chief Analyst
Ed Finegold, Charlotte Patrick,
Ed Finegold, Contributing Analyst Editors: December 2021 Contributing Analyst
Contributing Analyst inform.tmforum.org
Dawn Bushaus, Editor:
Editor:
Editor: Contributing Editor
Ian Kemp, Dawn Bushaus,
Ian Kemp, Managing Editor Ian Kemp, Managing Editor
Managing Editor Managing Editor SPONSORED BY:
SPONSORED BY:
January 2022 October 2021 December 2021
January 2022 inform.tmforum.org inform.tmforum.org inform.tmforum.org
inform.tmforum.org

February 2022 | www.tmforum.org February 2022 | www.tmforum.org November 2021 | www.tmforum.org

next generation
next
generation
CEM
now!
the Telefónica way

Author: Mark Newman, Chief Analyst, TM Forum

sponsored by:
Editor: Annie Turner, Contributing Analyst

Author: Dr. Mark H. Mortensen, Contributing Analyst Author: Dean Ramsay, Principal Analyst sponsored by: sponsored by: with the support of:
Editor: Dawn Bushaus, Contributing Editor Editor: Ian Kemp, Managing Editor

37 inform.tmforum.org
meet the Research & Media team

Report Author:
Dana A. Cooperson
Contributing Analyst

Report Editor:
Chief Analyst:
Ian Kemp
Mark Newman Managing Editor
[email protected] [email protected]

Principal Analyst: Editor in Chief, Inform:

Dean Ramsay Joanne Taaffe
[email protected] [email protected]

Customer Success & Commercial Manager,

Operations Manager: Research & Media:
Ali Groves Tim Edwards
[email protected] [email protected]

Global Account Director: Digital Marketing Manager:

Carine Vandevelde Anna Kurmanbaeva
[email protected] [email protected]

Sponsor Success Manager: Digital Media & Events Coordinator:

Maryssa Ramsey Ellie Hsu
[email protected] [email protected]

Report Design:
Intuitive Design UK Ltd
© 2022. The entire contents of this publication are protected by copyright.
[email protected] All rights reserved. The Forum would like to thank the sponsors and
advertisers who have enabled the publication of this fully independently
researched report. The views and opinions expressed by individual authors
Published By: and contributors in this publication are provided in the writers’ personal
TM Forum, 181 New Road, Suite 304, capacities and are their sole responsibility. Their publication does not imply
Parsippany, NJ 07054, USA that they represent the views or opinions of TM Forum and must neither
tmforum.org be regarded as constituting advice on any matter whatsoever, nor be
Phone: +1 973-944-5100 interpreted as such. The reproduction of advertisements and sponsored
Fax: +1 973-944-5110 features in this publication does not in any way imply endorsement by TM
ISBN: 978-1-955998-18-5 Forum of products or services referred to therein.

38 inform.tmforum.org
for more information about tm
forum’s work on autonomous
networks and intent please contact
Aaron Boasman-Patel