For AI and ML, it’s Important to Turn to the Specialists.

For telecom companies, improving operational efficiency is expected to be one of the most beneficial applications of machine learning and artificial intelligence (AI), and that’s exactly what tomorrow’s networks will require. For this reason, service assurance is a key focus for many operators who are looking for a way to maximize resources and evolve their networks towards becoming more intelligent, automated, and self-healing. Their challenge now lies in finding the right technology partners.

Help Wanted: Data Translators
The hiring race for data scientists and machine learning experts has been happening for several years now, but does that really solve the problem? Data scientists do play a critical role, but they are only one part of the puzzle. They know how to code and have experience working with specialized machine learning tools, but when it comes to understanding real-world challenges and incorporating these issues into the software, where does the business knowledge come from?

While a more generalized AI platform may be fine for some applications, when it comes to highly complex telecom networks, more specialized AI tools are a must. AI and machine learning can help drive network automation, but it’s not as simple as installing a generic platform, pushing a button and getting meaningful results. Finely crafted algorithms – designed with industry-specific knowledge and carefully trained with highly relevant data sets – are necessary to build a solution capable of managing such complex networks. We must apply a rigorous, scientific method to AI/ML, which starts with clearly defined problems, moves on to testable hypotheses, and then designs experiments to test those hypotheses. This is where domain expertise makes all the difference, and it’s what defines the role of the data translator.

In the OSS world, data translators live with one foot in data science and one foot in the Network Operations Center. Since they understand the core business objectives of service assurance solutions and the real-world challenges of running a network, these specialists can identify actions based on the findings of the data that neither the data scientist nor the executive alone are able to extrapolate. For example, to create an ML algorithm that predicts network events, it only makes sense to focus on predictions that will trigger a proactive action designed to actually improve the network operation process. This kind of domain expertise would typically come from a data translator, not an algorithm developer.

For this reason, data translators are a critical part of teams designing software solutions that will be making real-time network decisions in a closed-loop environment. They help create automated, closed loop service assurance solutions by guiding how algorithms are defined and spotting biases and inconsistencies before they get baked into the software.

You can invest in data technologies and collect all the data you can possibly imagine, but it’s worthless if it’s not analyzed and acted upon in an optimal way. According to a recent McKinsey survey, only 18 percent of companies believe they can gather and use data insights effectively. But when it comes to autonomous decision making, there is little room for error. Intelligence and automation need to go hand in hand.

The Coming Age of Specialization
For years, many service providers have gone with a ‘best-of-suite’ approach to managing their network, which means they are picking a single supplier for all their OSS/BSS needs. Many operators feel this is an easier path, giving them just ‘one throat to choke,’ so to speak. But it appears this era is coming to an end. With 5G, NFV and automation, the software systems that run tomorrow’s networks need to be smarter – with intelligence baked in. Not only will closed-loop service assurance systems be responsible for spotting problems in the network, they often will also be responsible for fixing them. Because of this, along with the ease and agility of today’s open API based solutions, we’re seeing the age of system specialists coming back again.

With so much riding on the performance of software tools, operators may begin to dread the idea of putting all their eggs in one basket and relying on a single vendor. Instead, they may want to consider gathering a portfolio of specialized tools deployed using a micro-services type of architecture, all integrated through modern APIs.

At TEOCO, when it comes to machine learning and AI – our approach is different from many of our peers because we design our service assurance solutions based on actual use cases that we’ve experienced through our work with dozens of operators around the globe. We begin with a clearly defined problem, move on to a testable hypothesis, and then design experiments to test that hypothesis. And we do this again and again – thousands of times. At TEOCO, we truly believe in the future of Machine Learning and AI, and we’re investing in building intelligent algorithms by studying actual ops center behavior – not just a generic platform that claims to do everything for everyone.

It’s impossible to be an expert in everything and every part of the telecom business, which is why we focus on service assurance. This is where we have decades of experience. TEOCO’s recent launch of Helix 10.1 introduces Anomaly Detection, a machine learning algorithm capable of automatically identifying network performance issues. By efficiently analyzing massive datasets of network and service KPIs, the algorithm is capable of detecting what human engineers cannot. It uses automated troubleshooting to offer a better understanding of the patterns that may cause service issues, as well as taking proactive measures to prevent these patterns from recurring.

Just as cars are becoming more automated and eventually will drive themselves, so too will our wireless networks. But this will come in small steps over the coming years, as trust is built through proven results, and of course, trusted partnerships.

Click here for more information on TEOCO’s Helix service assurance solution, and how we are helping drive the future of communications.