Skip to main content

Domain knowledge: the secret to AI/ML success in broadband networks

Domain knowledge: the secret to AI/ML success in broadband networks

Speak with any broadband provider today and they will tell you one of their key challenges is dealing with increasing complexity in their networks brought about by explosive growth in end-points, connected devices, nodes, bandwidth demand, traffic types, technologies…you name it. But they’re rising to the challenge by increasing the level of automation. Advanced automation tools allow operators to respond to network changes with clarity, speed and accuracy, thus reducing waste and increasing productivity. Artificial intelligence (AI) and machine learning (ML) play an extremely important role in automation and can help further augment operators’ capabilities in managing networks.

What is AI/ML? Is this some kind of magic wand we can use to make all our problems disappear? Unfortunately, this paints a picture that is far too rosy. Effective AI/ML tools are fickle and notoriously hard to get right, even in light of today’s massive computing power and the availability of large amounts of data. The timeless adage “garbage in, garbage out” applies here: any algorithm of AI and ML can only be as good as the quality of the data that feeds them. This makes data the most valuable component of the whole AI system.

Preparing top-notch data input requires domain knowledge, a core skill that is often overlooked when discussing AI/ML. Domain knowledge is the detailed understanding of the real-world environment in which the AI/ML algorithms operate. Once upon a time, AI hardliners considered that domain knowledge was an inferior skill. After all, AlphaGo Zero was able to achieve superhuman performance, and even determining the best in-game Pokémon characters can be done without a human expert. But in most real-world scenarios, rules are ambiguous and data comes with a lot of noise. So the view has now changed dramatically: whether it’s in improved breast cancer diagnosis or saving the elephants, or learning from Amazon’s gender bias tribulations, domain knowledge is what differentiates an under-performing, even potentially damaging AI solution, from a best-in-class, value creator.

Luckily, at Nokia, domain experts in broadband communications are an abundant resource. To illustrate how this enables us to mainstream AI in our products, let’s look at how we leveraged deep learning for network troubleshooting with POST Luxembourg. Our domain expertise proved key in analyzing their service, device and network data to predict customer issues and prescribe the best actions. This resulted in significant improvements to their troubleshooting operations, replacing time-consuming investigations, and recognizing impairments more quickly than technicians who have been working in the field for many years, and with a tremendous increase in accuracy and very low false-positive rates.

Recent AI/ML developments in telecom also show promising results in areas such as network care and operations. We expect AI/ML to be used at all levels in fixed networks to make faster, better and more predictable decisions on how broadband networks should be organized. That will lead to productivity increases for more and more tasks, especially because AI/ML relies on learning from data and, these days, there is more and more data available to train algorithms.

The reliable capture, transmission, storage and processing of all these data sets is therefore essential, which is why we’ve embedded these features into the hardware requirements and key algorithms of the Altiplano cloud platform, the heart of Nokia’s software-defined access solution. Massive telemetry streaming and data collection is a key enabler for all AI/ML use cases. The Altiplano platform supports a highly scalable infrastructure for streaming and collecting data, and makes this data available to OSS and other applications via open APIs.

If you’re ready, we can help you go a little deeper into automation, and AI/ML, starting with this great white paper. The benefits of AI/ML are unquestionable; all it needs is the right approach and the right partner to unlock them.

Filip de Greve

About Filip de Greve

Filip De Greve is Product Marketing Director for the Fixed Networks division at Nokia. In that role, he is focused on the go-to-market for innovative fixed access broadband solutions. Filip has over 20 years experience in the telecommunications industry and held various roles on provider and supplier side providing leadership in executing projects, technical consultancy and customer delivery. He holds a Ph.D. in the Telecommunications and Information Technology from the University of Ghent, Belgium.

Connect with Filip on LinkedIn or follow him on Twitter

Article tags