Futuristic fault-finding on Vodafone’s mobile network

Artificial intelligence will help engineers detect and fix network issues before they affect customers, as part of Vodafone’s continual, ever-evolving investment in its network.

Vodafone will use artificial intelligence (AI) on its networks across Europe to help detect and correct issues before they affect customers.

The Anomaly Detection Service, designed in partnership with Nokia, comes as mobile network infrastructure becomes increasingly complex. The Anomaly Detection Service uses a technique called machine learning to repeatedly analyse the millions of performance indicators for a network, a process that isn’t feasible for people to undertake given the overwhelming amount of data.

In doing so, it understands what the network should look like when it’s working at its best – as well as what it looks like when it isn’t. As soon as any performance indicator isn’t reading as it should, the Anomaly Detection Service will notify Vodafone engineers who can then investigate further.

By automating the mundane and time-intensive task of data analysis, the Anomaly Detection Service frees up time for Vodafone staff to improve the network itself and analyse other maintenance data, rather than continually monitor performance indicators.

The Anomaly Detection Service is just one of the many ways Vodafone is continually improving its network. For example, as ever-increasing numbers of sensors and other unmanned devices use mobile signal, Vodafone has to adapt its coverage for this so-called Internet of Things (IoT).

Vodafone’s Narrowband IoT (NB-IoT) coverage now reaches 98% of Vodafone’s UK network. In areas where it does not reach, Vodafone will work with its business customers to find a viable solution.

NB-IoT signals travel further and through thicker obstacles than 4G or 5G, reaching places that those technologies currently can’t – such as the middle of England’s forests. While data speeds are slower, this is less important for IoT sensors that only need to send small amounts of data occasionally.