How business requirements are shaping Orange's AN strategy

When it comes to autonomous network development and deployment “the reality check is having a business representative in the room,” according to Alexis Koalla, Director Operations Strategy & Autonomous Network Transformation, Orange Group.

“At the end of the day, efficiency and autonomy mean little if they do not reduce costs, support revenue growth, and improve how the business meets customer needs,” explains Koalla. And it is this focus on business outcomes that has led him to call for a rethink on how the industry approaches high-value autonomous network scenarios.

In 2023 Orange Group committed to achieving Level 4 of network autonomy by 2025 where it was justified from a business perspective and with the aim to “build momentum behind a shared vision,” he explains.

The company began by undertaking an autonomy maturity assessment across its operating companies (OpCos) in Europe and Africa.

“It was a huge job,” states Koalla. “At the end of this first assessment most of them were standing below level three. None were at three.” Today "the good news is that most OpCos are between three and four in terms of autonomy level for many high value-scenarios that count."

The next steps were a gap analysis; the identification of automation enablers; the definition of a related OSS 'urbanism', which refers to the architecture of OSS tools; and the creation of a road map to help guide operating companies towards achieving level four of network autonomy.

Then in the last 18 months, Orange Group started broadening the scope of its AN strategy to include the requirements of business leaders, who join regular workshops.

Business-driven use cases

In addition to hearing about business realities, operating companies (OpCos) within the Orange Group can also draw on the centralized expertise and skills of Alexis’s teams and the Orange Automation Practice, which was set up so that OpCos can learn from the experiences of their peers in other countries.

However, each OpCo decides where and how to invest in AN, based on their pain points and business needs, according to Koalla, with the high-value use cases of energy efficiency and fault management consistently among the top priorities.

“It's up to them to put the money where they can [achieve] significant business outcomes, whether that is the RAN, transport, or core,” he says.

OpCos make use of TM Forum’s ANLET, which is a comprehensive, self-service scenario-based assessment tool. It provides questionnaires that CSPs can use to gauge their own progress toward Level 4 autonomous networks. It also helps operators identify gaps and areas for improvement in specific high-value domains, including fault management or energy efficiency.

However, as Orange Group’s deployment of AN matures and is more closely coupled with business outcomes the company has identified a need for cross-network domain high-value scenarios. And Alexis would like TM Forum’s ANLET to reflect this requirement.

“We found that trying to bridge the high-value scenarios ... is not the right way,” says Koalla. “What we are seeing is what we want to work on with TM Forum. Instead of having ... one single high-value scenario, start thinking about cross-domain high-value scenarios,” he continues. “So, we need to rethink the way we are building autonomous network level evaluation tool, the ANLET, so that we can identify a high-value scenario that is covering or providing significant business [benefits] for two, three or four domains.”

It's a direction that closely aligns with TM Forum’s current strategic focus, according to Andy Tiller, EVP products & services at TM Forum: “This is exactly the problem we are setting out to solve with AI-Native ODA,” he says.

Tiller explains that going beyond Level 4 of AN requires building cross-layer and cross-domain autonomous flows, which in turn requires common semantic grounding and common governance of AI agents working in different parts of the business. TM Forum’s Open Digital Architecture (ODA) therefore aims to provide the industry with a common foundation for intent-driven automation, working across layers, domains, cloud platforms and software stacks.

“We are having some good successes today with AI point solutions providing autonomy for individual high-value scenarios, but maximizing the potential of AI requires us to move beyond point solutions,” notes Tiller. “ANLET will also need to evolve to support evaluation of Level 4+ autonomous flows”.

Koalla for instance believes that cross-domain high-value scenarios will help deliver “savings of business outcome at scale, " adding that "we can correlate and also improve many things on the business side, such as customer complaint management.”

Tooling up AN

The development of a cross-domain approach to high-value scenarios isn’t the only change Alexis wants to see.

“The main thing [OpCps] are facing, and this is the challenge ahead, is that to implement many of those high-value scenario they need tools, enablers, and these tools today, some of them exist, but need to be customized. Many of them don't exist.”

When it comes to RAN performance management, for example, Orange Group has developed an in-house tool, described within a document about RAN performance monitoring enablement published on TM Forum's site. In-house solutions, however, do not necessarily save time and money, points out Koalla.

"To deliver the best tooling for our affiliates, we perform 'make or buy' assessments. However, even if we buy an off-the-shelf solution we request for high openness and tools offering the capability to build on the top of it. This way, we go faster and capitalize on our internal expertise."

And regardless of whether Orange develops tools in-house or sources them externally, it “needs to make all of them consistent and converge.”

A composable OSS

Today, operating companies must modify the tools that help them deploy autonomous networks, which take time and money, states Koalla.

Orange is therefore evolving its OSS from a structure that Koalla likens to firmly cemented bricks towards a Lego-like, modular, composable architecture, called OSS 3.x.

“If the OSS doesn't evolve, then your autonomy is still a kind of dream,” points out Koalla.

With “OSS 3.0 we provide a catalog of tools in each domain of networks,” he explains. “It could be cross-domain, but we'll focus first on [individual] domains. We'll not impose a list of tools, we'll provide a catalog that they can choose from with flexibility, and that they can also customize,” he explains.

“The direction is towards platforming where we are building a common foundation. Each affiliate can customize the use cases to their own needs, but this customization can be put in the common pot of use cases catalog and can be easily replicated for others. This is the direction that we are building progressively with the OSS 3.0 journey.”