Telecom in the Era of AI

Telecom in the Era of AI

Beyond “Why is my broadband so slow?” AI in telecom tackles outages, predicts surges, and delivers optimal service even as 6G looms. Discover data-driven solutions by Cloudera.

Like many other industries, telcos have woken up to the power of artificial intelligence (AI) that not only helps with improving services but also with more human-centric roles to make day-to-day tasks easier. 

With 6G coming down the line and consumer demand for seamless services greater than ever, providing high-performing connectivity is key. If AI is to be successful at helping achieve this mission, telcos need to ensure they have access to large volumes of high-quality data. 

“Why is my broadband so slow?”

Telecoms networks are highly complex. They consist of multiple generations of technologies and many interconnected systems, making predicting and preventing network failures and downtime extremely challenging. Networks also have physical elements to contend with – such as weather and dynamic volumes of traffic on the network. AI can show enormous value in tackling issues in both of these areas. An Accenture analysis estimates that AI can potentially reduce network downtime by up to 50% for telecom companies.

This is because AI systems can do things that humans simply can’t, and in the blink of an eye. Take outages and network faults. An AI can identify patterns of weather that can be layered with machine learning (ML) algorithms trained on past incidents. By analysing previous instances of adverse weather correlated with network mix, loading and other contextual data, these technologies can recommend preventative measures engineers can take to mitigate the impact of an outage or even avoid it altogether. 

An example might be predicting the change in load on different parts of the network when a storm blows through, with AI able to call on whether network consumption might surge in the suburbs and slump in the cities, creating pressure points that could lead to service degradation.

AI and ML take a lot of the guesswork and grunt work out of engineers’ hands, enabling them to address problems before they become major issues. When a major issue occurs, AI-powered decisions also drastically reduce the mean time to repair.

AI can also aid proactivity in predicting surges of traffic and advising customers. Systems can be trained to autonomously manage and optimise network workloads, enabling telcos to make informed decisions about what technologies should be used at times of high demand. Various technologies are available to manage demand – from 3G-5G in wireless networks and copper and fibre in wired networks. 

Most telcos will have these capabilities in use – all of which are useful for different solutions and enable telcos to be flexible. But only if they’re smart about how to use them to provide network stability. When one network comes under pressure, the other can relieve that pressure.

For example, the pandemic put a huge strain on networks. Parents worked at home while their children were streaming TV shows and playing games online. This stretched fibre networks and slowed speeds dramatically. But with AI powering decisions and identifying bottlenecks, telcos were able to come up with solutions to solve this problem and advise customers on how to get the best service. 

These remediations were often counter-intuitive, such as asking parents to stream TV on wireless networks instead of fibre. Still, they made a huge difference in keeping the network stable and customers happy. Making accommodations for such strategies in advance will also recognise other planning requirements, such as optimising power in the wireless network.

AI is only as strong as its data

As AI continues to mature, other use cases will emerge. However, organisations must understand that AI is only as good as the data it learns from before letting it loose on the network. Models trained on data from only a subset of an organisation’s data may miss crucial insights or provide incomplete recommendations. 

Modelling customer experience based solely on billing data, for example – which remains the standard in many telcos today – ignores the impact of service quality and network performance on customer experience. Growing billing amounts may indicate a growing dependency on the network and could be interpreted as a satisfied customer. But if intermittent quality issues accompany that growing dependency, higher spending may indicate a higher churn risk.

With the global AI market for telcos projected to grow from $1.2 billion in 2021 to almost $40 billion by 2030, AI solutions are the industry’s future. So, the technology must be unbiased, fair, secure, and well-rounded, relying on clean and accurate data.

So, organisations must build AI use cases from robust groundwork, giving it access to a complete, clean, compliant and trusted data set. This will require a modern data architecture built around a unified data platform that enables AI to draw insights from data across the enterprise – from cloud environments to on-premise data centres. Strict governance must also be enforced always and everywhere, ensuring consistent levels of data quality across the business.

Honing connectivity with AI

Having refocused on core offerings, AI will play a crucial role as telcos look to deliver a better service. The landscape will only become more complex with 6G networks on the horizon. Telcos must prepare now and unify their data so services are unhindered.

Without quality data powering AI models, there will be a risk of the AI failing and misunderstanding context – or even providing inaccurate recommendations that will tarnish the brand reputation. To fully unleash AI’s potential, the industry must prioritise curating diverse, unbiased data coupled with thoughtful data practices to protect that data. AI has the potential to revolutionise the industry, but only if it’s built on solid foundations.