Operational drivers of autonomous networks

A recent TM Forum survey showed that one of the five main drivers of autonomous networks is reducing costs through operational transformation. This excerpt from our recent report Autonomous networks: business and operational drivers explores the impact of network automation on operations.

How to respond to the drivers has split communications service providers (CSPs). Nearly half of respondents to the survey conducted for the report Autonomous networks: business and operational drivers, chose continual improvement and automation of operations when asked about their visions for the future, while 43% said they have an ambitious automation strategy that they expect to result in a full transformation of operations. Another 10% are focusing on cutting costs.

In an interview

ahead of the FutureNet World event, Ibrahim Gedeon, CTO of Canadian CSP TELUS, said about digitally transforming to become a digital service provider that if CSPs don’t have the right network automation, they will be adding more teams because operating the network is becoming more complex. He added that with network automation, CSPs have more flexibility regarding the kinds of teams they put in place.

“Network automation should enable people to ‘fake’ multiple skill sets,” Gedeon said. “So, instead of getting this expert and data experts and optical experts, I would like to have a general engineering person with the right tools so that they can become [expert].”

He added that before cloud and network functions virtualization (NFV), network faults generally involved dealing with a single vendor. But now, he said, “I’m averaging three to eight vendors per issue.” Solving those issues needs automation.

Changing roles

The biggest change for network operations personnel is that those who remain will shift from primary troubleshooters to monitoring and handling exceptions. In our survey, 47% of respondents said they consider this scenario highly likely in the era of ANs. If the promises of AI and machine learning come true, exceptions should steadily decrease over time.

About a third of CSPs think it highly likely that there will be little change in operations roles, which says more about CSPs’ perceived ability to adapt than about the change required: They don’t think CSPs can adapt.

Nearly 90% of respondents think it is reasonably or highly likely that many roles will change from hands-on operations to designing systems or training and explaining automated, AI-driven operations tools to humans. This is essential to maintain oversight, understanding and institutional knowledge of the processes at work so humans can step in as necessary.

The systems-design part of this equation poses a real challenge for CSPs. In our 2020 TM Forum survey about automation and AI, 82% of operator respondents said they did not have enough internal expertise to develop the insights necessary for automated decision-making and closed loop operations. Being able to explain decisions by AI algorithms and a loss of control over tracking changes in the network were among the top concerns about introducing AI into the process.

Operational change becomes increasingly hard when practices and processes appear to be working well enough. It is the old, “If it ain’t broke don’t fix it” mentality. But in the face of new competition, working well is not always well enough.

In an interview published on TM Forum Inform, PCCW Global CTO Paul Gampe said that the most significant improvements his company has made through its ongoing transformation were in automation and the adoption of Agile practices. However, he added that, “the biggest challenge in this transformation was recreating the willingness and the desire to adopt new ways of working within a very successful company with a well-established operating model.

When something works well there’s often no urgency for change, so part of the challenge was to create that desire to want to embrace new ways of working.”

Time-consuming tasks

Such change begins with a hard look at where effort is being applied. More than half of respondents to

our survey said they are spending the most time on automating repetitive tasks, such as re-keying common data into multiple forms and initiating system backups, to more complex but still repetitive standard device configuration, incident report distribution, testing, inventory updates and low-level parameter changes. About 25% said lifecycle related tasks are most time consuming, and 20% said closed loop tasks are.

Getting rid of repetitive tasks does not need a centralized or coordinated effort but is part of the overall AN strategy. However, this consumes human resources that could be better applied to an enterprise-wide drive toward AN. Eliminating repetitive tasks is necessary and desirable, but does not automatically address the most urgent automation priorities for operations. There are bigger payoffs, for example, in automating service management and assurance than managing partners, at least for now, because their own house must be in order regarding automation before they can implement any type of zero-touch partnering arrangement. Hence CSPs see service management and assurance as among the most urgent automation needs.

Network planning and design also sits high on the list. As the result of a 2019 Catalyst project, BT was able to reduce the time it took to perform planning tasks by 70% through automation of its network planning and fulfillment processes, plus having a centralized plan that could leverage automation from adjacent functions. Note that these benefits and possible advantages were before 5G, cloud adoption, a virtualized core and broad AI adoption. The third most urgent system in need of automation is the fulfillment process: Fulfillment and assurance are increasingly intertwined and will likely be automated as part of a unified effort.

Read the full report to find out more.