How AI will be used to manage 5G networks

Telco infrastructure provider Nokia launched a new service aimed at helping operators use artificial intelligence and automation to better manage their 5G networks and avoid network and service-level failures.

“Traditional network and service management approaches are no longer sustainable,” Nokia said in a press release announcing its AVA 5G Cognitive Operations. The new framework, pitched at communications service providers (formerly called mobile network operators), relies on virtualization and network slicing to improve wireless network performance.

Network slicing is a way to use different services on the same network infrastructure—voice and IoT, for example. It’s a kind of flexible zone which lets different applications use varying bandwidth, and importantly, maintain varying levels of reliability depending on need.

The AI-as-a-service offering becomes particularly helpful, Nokia points out, with respect to service level agreements (SLA). A communications service provider can commit to delivering certain performance levels, and AVA 5G Cognitive Operations will help the telco provision network resources based on the terms of the contract. In addition, the platform will help telcos slice their networks properly, creating cost savings because the slices will be appropriately sized.

AVA 5G Cognitive Operations is also designed to help operators predict network and service failures up to seven days in advance, Nokia says. In addition, the platform will be able to resolve the anticipated outage issue faster than traditional network management methods, the company claims.

“Operators face a perfect storm of rising traffic and consumer expectations, so it is crucial to be able to predict and prevent service degradations at an earlier stage, while solving issues that arise significantly faster,” said Dennis Lorenzin, head of the network cognitive service unit at Nokia, in a statement.

Nokia’s product is run in the cloud, with Microsoft Azure. Other public and cloud options are available, the company says.

Nokia isn’t the first to suggest that AI will become prevalent in 5G network operations. Ericsson, another infrastructure company and Nokia competitor, surveyed communications providers in 2018 and found they overwhelmingly expected to see AI running 5G. More than half (53%) said they expected to see AI in their networks to some extent by the end of 2020. The respondents said potential areas of improvement due to AI included network planning and performance as well as customer service.

On the flip side, Ericsson’s survey found that network managers were concerned about AI-related data challenges in their networks. Worries included AI data quality, excess AI data from too many sources, and storage of AI data in too many systems.