Creanord, Lounea partner on AI, ML for network congestion prediction

Creanord and Lounea announced that they are co-operating on developing artificial intelligence (AI) and machine learning (ML) capabilities to predict congestion in data transport networks. The first results of the research are expected to be ready by June this year.

“Using AI and ML to look into the future enables the operator to react before an anomaly affects the end users’ experience. Providing an excellent user experience is a priority for every operator.” — Claus Still, CEO at Creanord

Users of communication services today have very high expectations on the quality of the connectivity and the user experience of using the data communication services. Research shows that users are likely to abandon a video stream if it takes more than two seconds to start playing the video. Moreover, video is the major part of the traffic on today’s networks and congestion drastically affects the quality of real-time services like voice communication and video conferencing. Being able to identify and predict in advance which parts of the network are likely to become congested allows the communication service provider to fix capacity or other issues before they affect the users. Moreover, it enables the operator to focus investments to parts of the networks where it matters the most.

“Using AI and ML to look into the future enables the operator to move from a reactive mode to a proactive mode in order to react before an anomaly starts to affect the end user’s experience of the network. The network experience is the single most important factor why users select a specific operator or choose to leave their current provider. Around 40% of users use the quality of the network as their selection criteria. Providing an excellent user experience should be on top of the mind of every operator,” says Claus Still, CEO at Creanord.

The research will utilize data collected by the Creanord PULSure solution from Lounea’s production network. PULSure collects end-to-end quality KPIs such as latency, jitter, and packet loss and the AI/ML algorithms will be trained on the data to predict congestions automatically in the network. Both unsupervised and supervised methods will be evaluated during the research.

“Lounea is known for its high-quality network and has been repeatedly recognized for Finland’s fastest fixed network speeds and low latencies by performance test sites. We continuously strive to improve our network and we are strongly looking into how to utilize AI and ML in our daily operations to always offer an excellent unparalleled user experience. I am looking forward to seeing the first results of this research,“ says Riku Päärni, CTO at Lounea.

Predictive analytics is the first goal of the research, but both Creanord and Lounea see huge potential to further extend the research to other types of use cases such as network optimization and predictive maintenance.

CT Bureau