ITEM: South Korean incumbent telco KT has announced the development of a new AI platform – but unlike rival operators, KT is applying its AI internally rather than externally.
On Wednesday, KT announced it had completed development of Neuroflow, an AI platform for network operations it’s been working on since March 2017. According to the Korea Times, Neuroflow is an open-source deep-learning platform that analyzes network data and makes automatic operations decisions based on that data:
The firm noted AI technologies have been partially applied to network operations, citing a system predicting problems in the LTE network. But its latest technology marked the first AI-based platform capable of analyzing data from any kind of network – wired, wireless or the internet of things, it said.
This enables KT’s network to essentially diagnose itself for faulty network design and causes of network failure, as well anticipate when failures are likely to occur.
KT said it plans to establish an “AI-assisted network control center” by 2020.
Also, the Times reports, because Neuroflow is open-source, other companies can use it for their own projects.
KT isn’t the first Korean operator to embrace AI or develop its own AI platform, but it does seem to be the first to develop one specifically designed for network ops rather than the consumer market (although KT plays in the latter space as well – its GiGA Genie AI-powered IPTV set-top box was launched in January 2017).
SK Telecom’s Nugu AI platform, unveiled in September 2016, is strictly a consumer play, aiming to rival the likes of Amazon’s Alexa and Google Assistant. The cellco has deployed Nugu in home smart speakers as well as its mobile navigation app for vehicles, and plans to integrate it into its IPTV service next.
In December, LG Uplus threw its hat in the smart-speaker ring, but not with its own AI platform. The cellco is partnering with Naver, whose speaker product powered by its Clova AI platform adds voice commands to LG Uplus’ IPTV and home IoT services.
That said, LG Electronics does have its own AI platform called ThinQ, and the company announced at last week’s CES that it will install ThinQ into basically everything it makes from this point on. So perhaps LG Uplus will be able to leverage ThinQ at some point.
In any case, it’s interesting to see an operator with a respectable R&D budget develop an AI platform specifically for smarter network operations, let alone an open-source one. For all the hype over AI as magic software that makes your appliances and your car smarter than you are, AI and machine learning will be essential – and arguably more useful – as an enabler of reliable network automation and troubleshooting, especially as networks themselves become more complex, as well as the services they’ll be carrying in the coming years.
In fact, last November, the Telecom Infra Project (TIP) – the open-source telecoms group spearheaded by Facebook (and of which KT is not a member) – launched a project group dedicated to exploring ways to apply AI and machine learning to network planning, operations and customer behavior identification to optimize service experience and increase automation.
(PRODUCTION NOTE: KT’s Neuroflow platform is not to be confused with the startup NeuroFlow, which aims to leverage sensors and big data analytics to analyze mental health in real time.)