Using big data to combat fraud in the telco network

Telecom service providers (telcos) across Australia and New Zealand have been at the forefront of ground-breaking innovations and progress in connecting people to each other and to the Internet, with this only becoming more apparent over time.

Just recently, Telstra Wholesale has confirmed the availability of 400Gbps wavelength services on inter-capital routes in Australia, effectively providing high-capacity connectivity. Meanwhile, Optus has confirmed EBITDA growth is up 4.5% with 425,000 more mobile subscribers for the year-end March 31.

Within this ever-changing landscape, 5G brings with it the promise and hype of new revenue streams, but the fact of the matter is that a telco's core business is focused on consumers. However, there is also a growing urgency to cut down on fraudulent operations within the network operations. While some activities are more serious than others, all these, if unchecked, make a serious dent in the bottom line. This is where observability becomes a priority.

Detecting telco fraud as big data and 5G become the new 'normal'

The strategies that telcos have deployed to combat fraud are constantly evolving. Not to mention, major investments have been made in data and analytics in recent years with a view to curb or mitigate the instances of fraud.

For example, telcos have always collected vast troves of data about their consumers in terms of call data records (CDRs), data consumption and billing information, as well as data related to the network, from telemetry to other parameters. However, it is the customer data that, when analysed correctly and at scale, can be used to effectively mitigate fraud.

The advent of 5G has further pushed telcos into investing heavily in real-time monitoring of network traffic to detect and prevent fraud by scanning and analysing call patterns and Internet and SMS usage to identify unusual activities or anomalies.

If a device registered on the network with one SIM card suddenly gets registered with multiple SIMs, then the data analysis can raise a flag. Or if an unusually high volume of call traffic occurs during off-peak hours or if a device suddenly connects to a base station in a foreign country, event triggers can set off alarms that customer service teams can quickly intervene to resolve.

Gaining visibility on data to get ahead of fraud prevention

Real-time monitoring would not be possible without telcos utilising their vast customer data to check for anomalous activity. However, this can be taken a step further to predict potentially fraudulent activity in the future. For example, if an analysis of CDR data shows large numbers of calls being made from an international number for unauthorised long-distance or international calling, then alarms can be triggered for further investigation. Similarly, if calls are being forwarded to premium rate numbers, this would trigger additional charges for unsuspecting consumers.

And this is only one piece of the observability conversation. As Australian telcos embrace hybrid cloud for their digital transformation, data-driven approaches to their business will become more prevalent, leading to data observability's growing importance. While monitoring data alone will tell you when something's not as it should be, that doesn't answer the question 'why'. Telcos will find that observability tools appeal to the fundamental pillars of data trust by helping to identify and troubleshoot anomalies that they can't see across the entire data lifecycle. With data-driven decision-making done the right way, telcos can not only quickly detect and disarm threats as they happen but also optimise workloads, increase efficiency and ultimately save costs on infrastructure and operations, all the while enhancing customer experience.

Trusted data is the foundation of good AI

Predictive models based on generative artificial intelligence (AI) or machine learning are a relatively recent strategy being deployed by telcos to combat fraud. With so much data being processed through their networks, telcos can deploy machine learning algorithms to analyse the large data sets that are generated from customer data.

However, generative AI is only as good as the data they have been trained on, with the right enterprise context. For telcos to benefit from trusted, secure and responsible AI, they need to be able to trust their data. Telcos that invest in a data infrastructure that is enterprise AI-ready with security and governance to detect fraudulent activities will go further in unlocking the value from their data, empowering their business, and truly democratising AI.

Modern data architectures are necessary for fraud prevention

Telcos have been investing in modern data architectures to help them cope with a number of business pain points. With fraud prevention, one of the major challenges is the number of data sources to be tracked, as well as the sheer volume of data to be tracked. These include data coming from the network, from their customers' data, as well from marketing and promotional data. While all of these together can help them create a 360-degree view of the customer, these datasets typically reside in silos. A lot of progress has been made to stitch these silos together to generate a single view of the data universe in a telco. That's, in essence, what a modern data architecture promises – but more needs to be done.

One way to address this is the Cloudera Data Platform (CDP), which has been deployed by many of the world's leading telcos. CDP helps telcos with the ingestion of vast datasets in real-time while also creating the right tools to enable analysis for several fraud prevention scenarios.

Developments with CDP also pave the road towards greater observability, with recent solutions bringing a wide range of use cases under one product, including active system monitoring, expedited issue resolution, workload optimisation, financial governance and self-service analytics.