Revolutionizing the telecom industry with intelligent machines

By Isabelle Paradis
President, HOT TELECOM
July 2015

Download article in pdf format»

IPsoft is one of the global leaders in IT and business processes automation and has more recently focused its distinct expertise in statistical and cognitive solutions to help telecom operators resolve fraud, improve quality and deploy value-add applications riding on top of their Big Data platform. I met with Ravi Dube, CTO at IPsoft, and we spent some time discussing how machine learning has become a key ingredient in telecom operators’ success and how it can help them play a key role in the IoT era.


IPsoft has been a leader in business process automation for over 15 years now, but how does this expertise benefit the telecom industry?
IPsoft was created in the late 90s with the main objective of automating repetitive chores. Since then, it has grown rapidly, and that initial focus is now manifesting itself in autonomic network management and statistical machine learning solutions. When it comes to the telecom vertical, we apply these two concepts to offer leading edge solutions to help carriers deal with fraud and quality issues.

Most recently, we have also been focusing our efforts on helping operators deploy innovative applications riding on big data platforms, which are in turn powered by the machine learning paradigms. These new applications can help operators with increasing value-add capabilities to predict customer churn or improve network automation for example.


Fraud is a very topical subject in the telecom industry at the moment.What makes IPsoft a leader in the fraud management space?
Fraud management was the first application we developed for the telecom industry, partly because we believed it was the hardest test to pass for a machine. We thought that many industry solutions were simplistic using manually defined rules and thresholds and unable to accurately detect the methods that fraudsters use specifically to avoid detection. The first challenge, to be successful in managing today’s fraud, is to tackle the growing number of different types of fraud and machine intelligence needs to identify this with great accuracy. In addition, a machine has to be able to detect variations in patterns and learn from them in order to identify when a new pattern appears and hence rapidly adapt to new types of fraud. Statistical machine learning capabilities is therefore key for a fraud management system to stay ahead of the fraudsters and IPsoft is a global leader in this domain.

Our fraud management solution differentiates from others in the industry, as it excels in precision and coverage, identifying a large array of fraud types through statistical sophistication, while minimizing the number of false positive results using machine learning. As a result, our carrier customers are able to automate the process of detecting and removing fraud without human involvement. Can you imagine automating the removal of a supplier from route if you are only 50% sure that it is fraud? The blowback from sales and from the supplier would be immense.

In fact, I would recommend that carriers set some clear performance criteria and even hold a “bake-off” between fraud management systems to fully determine their accuracy. Anyone who says that occasional false positives can be fixed by tweaking the thresholds is clearly not leading edge!


End user experience optimization is another subject growing in importance with telecom operators. How does IPsoft enable wholesalers to play a role in tackling this challenge?

The key to retaining customers is to provide or even exceed the quality of experience they expect. Our Call Quality Manager provides a way for wholesalers to look at how the end customer is experiencing the quality of service on a granular basis, something that they currently do not do. At the moment, wholesalers assess quality using Average Call Duration (ACD) and average Answer Seizure Ratios (ASR) by supplier and destination. In comparison, our system enables them to monitor quality as seen by each individual subscriber, tracking call retries and call disruptions as examples.

At first glance, the value of achieving this for retailers is obvious, but not so much for wholesalers. Nevertheless, in our opinion, it is crucial for wholesalers to be able to measure quality down to each customer, as it enables them to maximize the opportunity that a call attempt brings. It is sometimes forgotten that a wholesaler cannot impact the number of attempts they receive, but they can make a big difference to the number of paid minutes each attempt generates – more paid minutes equals more revenue! In order to do that, wholesalers therefore have to be aware of customers who are being disrupted and who do not talk as long as they would like. Our system provides accurate recommendations on which supplier to use to maximize each call attempt’s potential, helping wholesalers maximize their margins and revenues.

In addition, the insight into wholesalers’ customers’ customers provided by our system enables them to proactively predict and resolve trouble tickets, before retailers even identify that there is an issue. This is a very good business practice if you want to be seen as a partner for high quality communication.


What innovative solutions are you working on, what leading edge offering can we expect to see coming from IPsoft in the next year?
We are working on new apps on the big data platform to help wholesalers launch and monetize a number of innovative services to stay a step ahead of their competitors.

One of the services we have been working on recently is IoT, particularly with the automotive industry, with the drive to intelligent and autonomous vehicles. We believe this is an ideal time for IoT to take off, as the enabling technologies are there now. One of these technologies is the 4G access bandwidth, which is required for edge devices (in many case video sensors) to be able to communicate amongst each other and with the central system/cloud.

The second technology that will enable IoT is IPX, which will support the routing of the enormous amounts of traffic generated by these edge devices back to the closest data centre hosting the central cloud, while sustaining high quality and availability. We believe IPX will be key in IoT’s success, as it will enable appropriate Quality of Service through the implementation of Class of Service (CoS) routing, resolving the congestion issues that may come with the IoT traffic tsunami.

Finally, the third enabling factor for IoT is low latency cloud access, because at the end of the day, your intelligence and analytics will reside in the cloud and delays in analysing information from a vehicle or a healthcare sensor would not be tolerated and errors could cause major inconvenience or sometimes be fatal.

So all three pre-requisites are lined up (4G access bandwidth, IPX and Cloud) like a perfect storm or an ideal convergence of factors that can enable IoT’s success. International carriers able to bring all those elements together in an all-encompassing solution will be the winners in this new opportunity.


What are the key trends in your opinion which will have greater impact on the wholesale market over the next few years?
Every so many years, an industry has to reinvent itself. This is such a time for voice wholesale. Traditional voice and SMS business is taken for granted and what premium customers are now looking for are services that build on mobile platforms to provide a richer set of value-added service.

A host of such new services are on the horizon: IoT, e-commerce, mobile payment, distribution of media content for entertainment and analytics to monetize intelligent apps for example. These services differ considerably in their network requirements - some are real-time, some are not, yet others are sensitive to loss, and so on and so forth.

Therefore, the next wave of telecom services will require app-aware infrastructures to deliver to users’ expectations. In addition, end-to-end CoS in IPX transport and virtual data centres for hosting, coupled with strong security, will be required to make these new services a reality.


What is the best advice you wish to pass on to our readers to achieve success in business?
I would like to leave your readers with two thoughts:

1. Persevere. Break-through technologies, apps and services require gestation periods which are sometimes lengthy. Therefore perseverance and the belief in your analysis are crucial to success in the longer term.

2. Think through clearly and then act boldly. An hour spent thinking upfront saves nine downstream and prevents costly rework with its usual delays.

Download article in pdf format»

About the Author:

Isabelle Paradis

Isabelle Paradis, President of HOT TELECOM is an international telecoms expert, with over 20 years working experience in the Americas, Europe, Asia and Africa.

HOT TELECOM has been serving global operators, governments, equipment vendors and telecom investors for over 13 years, providing leading edge market research and consulting services to industry leaders around the globe.

For more information on our HOT INTERVIEWS program, please contact Isabelle Paradis at: or at +1 514 270 1636