Project Palantir

Important results from the project

The objective for project “Palantir” was to build knowledge in AI/ML and to assess how this toolset could be utilized in fiber access networks to increase availability and stability, as well as to simplify operations and maintenance. The project was conducted by Waystream (FTTx access switches and routers), the research institute RISE and the city operator Lunet. The result included assessment of various algorithms as well as positive demonstration of how AI/ML could address a specific failure scenario use case.

Expected long term effects

Project “Palantir”, conducted by Waystream (FTTx access switches and routers), the research institute RISE and the city operator Lunet, has successfully demonstrated how AI/ML could address failure scenarios for fiber access networks. The project, that also has resulted in knowledge sharing among the project parties, has generated a foundation for the development of new services further improving the availability and stability of fiber access networks, as well as also simplifying operations and maintenance.

Approach and implementation

Based on experiences from RISE we adopted an “ML Pipeline” approach. It consists of an initial phase including problem formulation, data collection and data pre-processing followed by an iterative phase where ML models are selected, trained and evaluated. This process has worked very well and can be recommended. Due to Covid-19 we have not been able to have live workshops. Despite this situation we have managed to reach our objectives, but can conclude that face to face meetings would have made the work more efficient.