AI Is in Diverse Parts of a Mobile Operator's Business: Here’s Why
You cannot escape the noise surrounding artificial intelligence (AI) right now. Over the last year, AI has become the latest in a long list of technologies to command column inches and webinar hours in the telecom industry, replacing network functions virtualization (NFV) as the industry's second most talked-about topic (after 5G, of course).
One reason for this is the rash of announcements over the last few months of virtual agents that mobile operators have developed, using AI to underpin smarter responses to text- and voice-based interactions with customers. Vodafone was an early adopter with its Hani virtual agent that uses an AI-based platform from specialist 7.ai. More recently, in April 2017, Orange placed an AI-powered virtual assistant at the center of its Djingo smart home solution; in November 2017, Deutsche Telekom announced that it was planning to enhance its Tinka digital assistant (launched by T-Mobile Austria) with voice-first capability, using the Omega solution from i.am+, the tech company owned by entertainer will.i.am. A host of tech startups are active in the AI chatbot and voice-based natural-language processing space.
But this active area is a tiny part of the potential for AI in mobile telecom. For 20 years, neural networks have provided the basis for fraud detection and threat intelligence services; today, as networks become increasingly complex, AI solutions are being used to optimize antenna beam patterns and to switch carriers to optimize voice-over-LTE (VoLTE) call quality at the cell edge.
Marketing departments are using AI to ensure that proposed new products are sold at the right price -- based on an intelligent evaluation of competitor products and impact on existing offers -- and offered at the best time to appeal most to potential buyers, on a per-customer basis.
Network engineering teams are offered a choice of recommended actions to undertake preventative maintenance work, with an AI system having evaluated network performance parameters and learned which actions have the best impact. And dealing with trouble tickets is getting a whole lot easier now that the learning from millions of tickets per year can be taken into account -- far too many for the best human engineering teams to match.
The success of these applications has spurred vendors of many kinds to develop and try to "productize" AI solutions and platforms for telecom applications; though, in practice, many of the telecom-specific applications involve a great deal of bespoke development. But it is the case that AI is popping up in more and more aspects of operators' systems. The next five years will see a very rapid rise in deployment and use of such systems as the value becomes clearer.
The Heavy Reading report Artificial Intelligence in Mobile Networks examines what AI means in the context of mobile telecom, identifying the techniques that are being used, where AI processing is best done and what it is being used for. It maps the supply-side of the market for AI solutions used by mobile operators, identifying some representative players of various kinds and profiling leading and interesting vendors.
— Danny Dicks, Contributing Analyst, Heavy Reading