• Telcos globally are targeting network automation for increased efficiencies
• Some major operators have begun assessing their network automation levels with the help of a methodology from the TM Forum
• The TM Forum measures autonomy in the network on a scale between 0 and 5 (where 5 is ‘full automation’)
• The industry, based on assessment pilots, is currently halfway to achieving complete automation, with an average level of 2.5
• Several big obstacles need to be overcome, however, so that telcos can fully automate their networks, according to the TM Forum’s Andy Tiller

Following years of talk, hyperbole and planning, much of it fuelled in recent times by a surge in AI developments, telcos globally are moving at a steady and still often cautious pace towards the implementation of network automation processes, but it’s a tough task for most network operators, with industry association TM Forum highlighting a number of major challenges associated with trust, skill gaps and cost justification on the road to fully AI-driven processes in the network.

Operators have been striving to automate many processes in order to improve all manner of efficiencies for decades and most particularly since the advent of network virtualisation and software-defined networking (SDN). But as with many important developments in an increasingly software-enabled sector, pinpointing the most effective strategy and measuring the effectiveness and impact of automation is hard. As a result, an increasing number of telcos have been embracing ways to identify the steps required to achieve their automation goals and are in need of a methodology to track their progress.

One such network automation assessment approach has been developed by industry organisation the TM Forum, which has designed a methodology that evaluates the autonomous level of a given telco’s network on a scale of 0 to 5: A score of 0 equates to ‘manual operations and maintenance’ with no automation; a score of 1 is given for those with ‘assisted operations and maintenance’; 2 is awarded for a ‘partial autonomous network (AN)’ (some closed-loop operations and maintenance for specific operations); a score of 3 is given to telcos with a ‘conditional autonomous network’; 4 is for a ‘highly autonomous network’ that includes ‘decision-making based on predictive analysis or active closed-loop management of service-driven and customer experience-driven networks via AI modelling and continuous learning’; and 5 is for a ‘fully autonomous network’ with ‘closed-loop automation capabilities across multiple services, multiple domains (including partners’ domains) and the entire lifecycle via cognitive self-adaptation’. For more on those automated network levels, see this TM Forum post.

So far, 55 companies, including telcos and vendors, have signed the TM Forum’s Autonomous Networks Manifesto, which contains an agreement by the signatories to use the association’s categorisation of autonomy based on autonomous networks levels and to adopt a building-block approach to achieve their autonomous network target architecture.

Are we there yet?

So how far down the autonomous networks path are the telcos? 

Earlier this year, Laurent Leboucher, group CTO and SVP of Orange Innovation Networks, told TelecomTV that the operator group is currently at level two on average, meaning it has “already started to implement the right processes”, such as using AI for root cause analysis and for predictive network management, although the company has not yet integrated the technology at scale for various use cases.

Leboucher noted that the goal for Orange is to implement its automation capabilities at scale but pointed out that there are challenges in terms of making the data accessible (the so-called “data democracy”), collecting it in a data lake and making it available to cloud-based processing and storage.

Telecom Argentina, another signatory to the Forum’s manifesto, told us that it is “diligently enhancing autonomy levels for select use cases, proven to yield significant benefits for the company.” Some cases, it added, currently operate at level 2, and the telco expects to advance to level 3 across key domains by 2025.

The Argentinian telco’s goal is to attain level 4 for specific processes by 2026 or 2027, and it intends to do so by creating “essential use cases that streamline the delivery of new on-demand network services.” In addition, “we must implement new technologies to abstract the network domains and bolster their programmability and service domains accordingly,” the operator noted.

Another major telco group, Vodafone, has also signed up to the manifesto but declined to provide an update on the level of network automation it has reached. In response to TelecomTV, the company stated that it “uses significant levels of automation within its IT infrastructure and networks across Europe to improve the customer experience, reduce energy consumption and free up our engineers to be able to work on developing new products and services, among other benefits.” The telco added that the TM Forum’s Autonomous Networks initiative “is of interest to us, and we are currently assessing how best to get involved.”

Andy Tiller, EVP of member products and services at the TM Forum, told TelecomTV that it estimated the industry on a whole has an average autonomous network level of around 2.5 – based on self-assessments provided by a total of 22 operators across Asia, Europe, Latin America, the Middle East and Africa. These include Thailand’s AIS (Advanced Info Services), Antel Uruguay, Finland’s Elisa, Globe Telecom in the Philippines, Indonesia’s Indosat Ooredoo Hutchison and Telkomsel, Africa’s MTN Group, France’s Orange Group, Spain’s Telefónica, Saudi Arabia’s STC, Telecom Argentina and several anonymous operators.

One assessment pilot by the TM Forum discovered that in a scenario to determine the autonomy level used for fault management on the radio access network (RAN), the average score among the 22 operators was 2.54 out of 5. In terms of fault management on the core network, their average automation level was slightly higher – at 2.64 out of 5.

Tiller noted that while the methodology for scoring the automation levels across various scenarios is “very standardised”, there is more work to be done regarding the actual interpretation of the questionnaire. Therefore, the TM Forum is planning to create an unambiguous scoring system that would mark the start of “a proper industry benchmark” for the particular scenarios already examined (fault management on the RAN and on the core network). It then plans to get the operators to repeat their analysis based on the updated scoring system, and to publish its new results by November.

Tiller further noted that an AN level of 2.5 is “a reasonable score” given that it is based on an average of more than 22 networks globally. He also highlighted fault management and network optimisation as “the clear winners” in terms of areas of high priority for telcos when it comes to automation. “These are things that operators have already done a lot of automation on because they are processes that lend themselves to being automated and, in some cases, are already using machine learning to support optimisation, for instance,” explained Tiller.

He added that operators are prioritising, in particular, the ability to see data from the network and tune various parameters in real time – either to heal a fault automatically or to optimise performance and customer experience.

So some progress is being made, but what are the major challenges associated with achieving network automation? We’ll delve into those in the second part of this feature article. 

- Yanitsa Boyadzhieva, Deputy Editor, TelecomTV