We started a cooperation with our customer Lunet and the Swedish research institute RISE in 2020 to explore how to make use of AI/machine learning to analyse operational data.
The target was to make use of the newly introduced streaming telemetry data from our access switches to increase the availability and simplify the operations of fiber access networks.

Streaming telemetry data has the advantage over traditional data collection, such as SNMP, of being near real time making it possible to identify a broader range of network issues.
Datapoints such as bitrate has a complex structure though, and varies over time of day, day of week, season and specific events among others, which means that fixed thresholds can not be used to identify anomalies.
This is where AI and more specifically machine learning provides the required tools to find the fingerprints of various error situations.

The project was expanded in 2020 where it was made part of the Celtic-Next supported pan-European flag ship project AI-NET, including 98 organizations from 7 countries.
Waystream’s cooperation was extended with the AI experts of the Swedish SME Savantic and the German Fraunhofer Heinrich Hertz Institute.

The project was conducted as a live network field trial where more than 1’400 Waystream access switches in the Lunet open access network were providing near real time data to an AI-based data analytic pipeline constructed by Savantic. Studies included machine learning algorithm comparisons, pipeline capacity, advantages of near real time data among other.

The result has been disseminated at multiple conferences including Denog15 in Berlin and SNCNW in Linköping.
Together with the overall AI-NET program it won the CELTIC-NEXT Innovation Award for 2024.

“The project has among other shown the advantages of near real-time data compared to traditional data collection, and experiences from the project will be important both in terms of functionality in products and collaboration with technology and product partners”, says Johan Sandell, CTO at Waystream.

Want to learn more about our AI Project?

See paper “SNCNW2024 On the use of streaming telemetry data for network health monitoring and anomaly detection” for further reading.