Orange highlighted the importance of data and AI in its innovation model, marking their use as crucial for developing smarter networks and enhancing operational efficiency.
At a webinar organised by the company today (18 May), SVP of Orange Labs Networks Emmanuel Lugagne-Delpon said AI could “bring value to almost every phase of the network lifecycle”, including planning and design, investment optimisation, smart maintenance and security.
He pointed to a use case from Spain where AI and machine learning (ML) were used to determine which sites were most profitable and how to reduce churn levels. This delivered a 10 per cent to 20 per cent improvement in its capex efficiency, with the operator now planning to replicate the model in other markets.
In another example, Orange used AI to detect international call fraud, reducing related costs by €37 million, an approach Lugagne-Delpon said was “highly replicable” for other transactions.
Orange also employed AI to predict the evolution of mobile traffic to prevent congestion, and to reduce network energy consumption.
“These are the principles to apply to all the segments of the networks, on fixed and mobile access, on transport network, on core networks, data centres”, he said, adding Orange will increase automation to improve monitoring and speed detection, troubleshooting and recovery.
Future networks
Steve Jarrett, SVP of data and AI, explained these fields will be particularly useful in future “software centric” networks. While noting there are challenges, he said such infrastructure was “a really wonderful opportunity, because data will be much more available and easier to get than we have today in many cases”.
Orange aims to test potential uses and learn from this, but also scale successful programmes across its footprint, he said.
Jarrett added Covid-19 (coronavirus) highlighted the “need for fundamentally data-driven decision-making”, in turn driving “further investment in data and AI” while also delivering a “real cultural transformation of the understanding of the need to be data-driven”.
Comments