Ericsson is using machine learning and artificial intelligence (AI) to create 5G networks that learn and improve.

The NWDAF network data analytics function helps service providers improve the user experience by using the data generated by the network to flag and fix problems, thus improving service quality.

For 5G, automation is a continuous endeavour ranging from automating specific simple and complex tasks to supporting zero-touch autonomous networking, which requires the application of technologies such as AI, machine learning (ML) and reasoning. NDWAF achieves this by getting the right data at the right cost to feed into the AI-ML models.

NWDAF enhances network operability by providing insight that helps the service provider identify and solve network problems on different levels as well as input for taking business decisions. One example is automated closed loops – data retrieval, analytics processing, insights generation and action enforcement – on different network and application levels.

Ericsson says its NWDAF can drive up to 45 per cent reduction in opex and capex compared with typical third-party NWDAF implementations with event exposure. Smart data acquisition allows more efficient integration and testing, simpler orchestration, and optimal footprint.

“At BT we connect for good and our customers increasing reliance on the network requires us to think differently,” said Neil McRae, British Telecom group chief architect. “With Ericsson we are working on cloud native solutions for 5G core architecture to support multiple generations of technologies and we see the need for greater automation and analytics to manage complex networks and ensuring the best quality experience for our customers. I’m pleased to see that Ericsson is launching NWDAF building upon Ericsson’s built-in software probes and expert analytics, a natural next step development to enhance closed loop automation and service assurance that could benefit BT’s customer network experience platform.”

NWDAF is a reusable set of microservices integrated into Ericsson’s cloud-native dual-mode 5G core and in its expert analytics. Flexible deployment and model portability enable the NWDAF to be located both centrally and as a built-in NWDAF in the 5G core products. It is based on 3GPP release 17, defining a distributed network analytics architecture where inference and training, and data repository are separated.

“Everything 5G, especially 5G core networks, must be automated,” said Monica Zethzon, head of area packet core at Swedish firm Ericsson. “Automation will improve operational efficiency, reduce time to market for new services and is required to handle the magnitude of tailor-made services that will be introduced with 5G. With the launch of NWDAF we build the automated closed loop and assurance, enabled by AI and machine learning, to leverage data-driven operation based on network data.”

James Crawshaw, principal analyst at Omdia, added: “Ericsson’s NWDAF adds intelligence to the 5G core, bringing new insights that enable automation and optimisation. With this, Ericsson shows its technology leadership in 5G core.”