Artificial intelligence in telecoms

There's huge potential for AI and data analytics in telecommunications, with service providers investing heavily in targeted and personalised marketing, chatbots for customer self-service, robotic process automation, and new machine learning approaches to network optimisation and predictive maintenance. But look with a critical eye and you'll notice that these early applications all primarily benefit the telco. So what's in it for you, the customer? Will you ever see AI used directly for your benefit? The answer is yes, through wifi optimization.

Wifi quality becomes a telco responsibility

We all need good quality wifi. With the rise of streaming, gaming and videoconferencing, fast, reliable wifi has become more important than ever. It's hardly an exaggeration to say that for most of us, our wifi experience IS our experience of the internet at home. But maintaining great wifi quality is increasingly hard.  More and more competing wifi networks, unseen radio interference and physical obstacles within the home complicate the already tricky physics of cramming so much high-speed data into a small part of the spectrum.

Until now, you've been on your own to get it to work well - your telco effectively drew the line on their responsibility at the connection to your home. And although much could be done to track and optimize wifi quality, little of this could be done by the customer.  Which left a process of trial and error : moving your devices, moving your router, maybe upgrading or spending extra on a range extender that might or might not help. Without even really knowing whether your wifi was working as it should or not. The end result has typically been failure, ongoing frustration and a substantial waste of time and money.

Fortunately telcos are recognising that, from the customer perspective, wifi is as just as much a part of their internet service as the fixed network connection. The end-to-end experience is all that matters to you, the customer. So now service providers are moving to do the complex work of wifi optimization for you, automatically. Why now? Because finally they can - thanks to the emergence of AI and sophisticated algorithms with the power to exploit the gold-mine of data already available on your router.

The amazing data on your router : TR-069 & TR-181

Every application of AI is built on great data, and luckily there's a wifi treasure-trove just waiting to be used. Your humble router doesn't just provide your wifi connection to the internet, it also constantly monitors its own performance.  It can track service quality indicators like the upload and download speeds it achieves for you, as well as the error and retransmission rates that indicate how hard it had to work to get you this performance. Better still, it can do this separately for each device you have connected, so can track performance issues with your phone separately from those with your laptop, TV or gaming console. It can also sense details of the wifi environment it was working in, such as signal strength. In principle all this information can be tracked every minute to give an amazingly rich picture of your wifi experience. And if this isn't enough, the router also has the ability to conduct a highly detailed diagnostic scan of the neighbouring wifi networks which it is competing with. Not bad for a largely forgotten little plastic box that often sits gathering dust, in a corner, out of sight!

Building on telco AI investment and data analytics capabilities

Through the foresight of international standards committees, this data has been defined consistently (TR-069 & TR-181) and available for a decade, but until recently it's been difficult to make use of it for anything other than reacting to individual faults. It wasn't possible to manage this volume of complex data without ubiquitous cloud computing and the power of AI and sophisticated algorithmic approaches. But, having already invested heavily to develop these capabilities for customer and marketing analytics, telcos are now able to deploy them directly on service quality management for each individual customer like you.

Most telcos have built large-scale data lakes that take advantage of the incredibly cheap storage and processing available in the cloud, so can handle volumes of data which would previously have been uneconomic. And the wide range of AI techniques required for the analysis offer new challenges for their in-house data science teams that go well beyond the complexity of marketing analytics: feature detection and feature engineering in unintelligible datasets, clustering to spot subtle patterns, anomaly detection, sophisticated time series analysis using e.g. recurrent neural networks, recommender systems and automated test-and-learn approaches etc.

Wifi optimization

With all of this data and analytical power your telco will be able to constantly monitor the quality of wifi service you're receiving and make sure it's up to the highest standard across the network. It could benchmark your performance over time, against your neighbours', and even among your own connected devices to identify high probability opportunities for improvement. It could determine which combination of equipment and settings deliver the best performance in which environments. It could do proactive maintenance, anticipating emergent problems before they arise and preventing them. And it could automatically diagnose the cause of problems which do arise so they can be resolved faster.

Most importantly, it could take proactive measures to improve your performance and the network as a whole. It could automatically optimise the settings of your router, for example selecting the right channel not just for you but for you and your neighbours as a group, selecting the best overall solution rather than triggering futile, destructive competition for bandwidth between individual customers.

It could also identify customers most likely to benefit from an equipment upgrade, for example where performance variations between connected devices indicate likely range issues or physical impediments. It could even recommend the best solution for each customer based on similar situations and positive or negative experiences with attempted remedies. So instead of just standard solutions, or even an individual engineer's judgement on good options, an AI system could learn from everyone's experience to make the best suggestion.

AI will revolutionize telco service quality management

Filipe Lopes, a wifi optimization and Lead Developer of Axiros's AXWIFI system, expects customers to be major beneficiaries as AI revolutionizes the approach to service quality management. "This data finally shows telcos the customer's own perspective of their service experience - the only perspective that matters" he said. "You can't help but be excited by the possibilities AI offers, we're on the cusp of a powerful new way to truly manage service quality and that's great news for customers". And, with immediate cost savings from fewer inbound calls to stimulate investment, he expects telcos to drive progress rapidly.

So in the near future you can expect your telco to be working invisibly to make your wifi experience better: higher speeds, less buffering, lower latency, greater reliability. And with the focus of telco leadership on customer satisfaction metrics like Net Promoter Score (NPS), what makes customers happy makes them happy too. So this will likely be just one in a series of improvements that telcos will make as they leverage their data and AI capabilities beyond marketing into operations and customer service. These are all important steps on the road towards the ultimate goal of zero-touch customer service, when frustrating and ineffective call centre experiences will be a thing of the past. And that's something that customers and telcos can be equally happy about.