OPTOSS AI featured in De Telegraaf!

Original article (NL): https://www.telegraaf.nl/financieel/1900427/storingen-op-netwerk-voorblijven

More and more businesses are dependent on their network connections. An outage is the last thing you want to experience. Above all, the underlying problems that cause these outages are only discovered after the fact. It shouldn’t have to be this way, according to Taras Matselyukh.

Taras (left) believes network issues can be discovered before outages occur

Taras is a telecom expert who has developed a self-learning system that detects and mitigates these underlying problems before they can cause damage to your network. Born in Ukraine, he movef to the Netherlands 19 years ago. After spending years working as a telecom consultant with Cisco and Juniper Networks (amongst others), he eventually ventured out and started his own consultancy business.

“As a consultant, I helped companies locate the problematic parts of their system(s). If something is not working properly, some part of the system will reflect this. Often, however, issues are overlooked due to the massive amount of data such systems output. This is especially true in the Telecom industry. We’re talking about millions of lines of data per hour. Humans simply cannot process such vast amounts.”

Developments in Artificial Intelligence inspired Taras to automate the process of detecting network issues. He built a system that collects data sources, analyses them, and detects anomalies in real time. A self-learning set of algorithms recognise patterns, and are able to structure them into clusters of anomalies. “When something goes wrong, it can be detected in real time, and allows the network operator to focus on the specifically affected area.”

With support from the European Space Agency (ESA) in Noordwijk, and the University of Amsterdam (UVA), Taras built the Optoss platform into a fully functional product that is capable of detecting, and remembering anomalies. This means that if an anomaly has been detected once, it will recognise it in the future and immediately pursue the correct course of action.

“Optoss still needs human input. An expert has to teach it whether an anomaly is good or bad. In the future Optoss will be able to determine this autonomously.” Taras underscores that Optoss does not render the human operator useless. “Network infrastructures are so complex. They consist of so many devices and software configurations which make it very difficult to pinpoint the source of outages. The Optoss system effectively turns human operators into super-humans.”