LIVE FROM TIP FYUZ 2025, DUBLIN: Vendors believe operators remain crucial in the process of RAN automation, with the data they hold considered vital in efforts to develop systems which will cope in real-world situations.

The consensus across the board during a panel session is the better the quality of the information fed into automation systems, the better the output.

Sameh Yamany, CTO at Viavi Solutions (pictured, far right), explained data which has been synthesised for testing purposes often fails in the field, making collaborations with operators and vendors important.

He noted these relationships must be backed up by a “commercial-grade testing” environment, one which enables more data to be trialled in real-world scenarios.

Aira Technologies founder and CEO Anand Chandrasekher (pictured, second from right) was clear vendors “absolutely” still need operator data to train network automation systems.

He explained his company can get most of the way on its own, with the base models it uses to train its products covering at least 70 per cent of the information it needs.

Detail from operators is required to get the training closer to completion, which is “not intense work” and can be done in around a month.

Chandrasekher argued this would become a self-fulfilling prophecy, with collaboration helping to reduce the amount of operator data required to train automation systems, in turn speeding the process.

Evolution
Rodrigo Correia, head of Solution Line SMO at Ericsson (pictured, far left), called on operators to find a way to provide greater quantities of and more realistic data, explaining it is important the information on hand is evolved “often” to keep pace with emerging developments around agentic AI and other technologies.

Rakuten Symphony benefits from being part of its namesake Group and its mobile operator in Japan.

Anshul Bhatt, chief product officer in the OSS Business Unit and head of Intelligent Operations (pictured, second from left), said its experience is that people must remain involved in any RAN automation system.

People with knowledge of their field can “qualitatively analyse and improve and keep” data under watch.

Bhatt said the technology and data are largely in place, meaning attention is now turning to how to access the information required.