Using Machine Learning Models to Generate Connectivity Graphs for Conditional Handover Procedures

Ericsson has been granted a European patent publication (EP4476936A1) for a system using machine learning models, specifically graph neural networks and transformers, to generate connectivity graphs between user devices and network nodes for conditional handover procedures in cellular networks. The technology classifies connections based on signal measurements and predicts handover success probabilities. Inventors are Selim Ickin and Jalil Taghia, with the application covering all designated EPC contracting states.

This is a patent publication rather than a regulatory instrument, imposing no compliance obligations on third parties. Technology companies developing telecommunications equipment or network optimization solutions should review the claims for potential licensing considerations or to assess freedom-to-operate in the European market. Patent attorneys may wish to monitor related proceedings or family applications in other jurisdictions.