Nokia sells telecommunication network systems, that are full of sensor-equipped chipsets and regulate operating temperature using fans. Failing to detect broken chipsets and failure of the systems can result in costly system maintenance and replacements that include repair and travel costs, with later detection leading to even higher costs. As a result, it is crucial to detect any small degradations or deviations from standard operating behavior (so-called anomaly detection) before it is too late.

By creating the first version of an anomaly detection system and developing a modular framework for its implementation, we enabled Nokia to take the first steps towards predictive maintenance of their telecommunication network systems. The developed framework provided Nokia’s data scientists with the flexibility to update existing pipelines and develop new ones, paving the way for continuous improvement in their anomaly detection capabilities. Our Bright Cape solution has the potential to save Nokia significant costs and increase the overall reliability and efficiency of its telecommunication network systems, making it a valuable asset in their ongoing efforts to deliver high-quality services to their customers.