Trustworthy 6G Positioning in Complex Environments: A Data-Driven Approach

Purpose and goal

This project aims to deliver a novel, trustworthy cellular positioning solution that performs robustly in mixed indoor-outdoor scenarios and at scale. Building on existing AI methods, it integrates with traditional time-of-arrival positioning and augments accuracy through graph neural networks for line-of-sight classification, addressing GNSS limitations indoors and in challenging environments. The goal is a dependable, scalable framework suitable for widespread deployment.

Expected effects and result

The work is expected to yield improved positioning accuracy and reliability across diverse settings, thanks to AI-enhanced fusion, robust line-of-sight assessment, and automatic calibration of 5G/6G antenna locations. By utilizing robotic reference units, the system can validate performance and provide a trustworthy baseline for future networks and applications, benefitting public safety, logistics, and mobility.

Planned approach and implementation

The project is structured as a close Ericsson-NTU-KTH collaboration combining industry deployment experience with academic research strengths. Teams will coordinate on milestones, data collection, and validation campaigns, with Ericsson providing network expertise and field resources, NTU and KTH leading algorithm development and experimental design, and joint workstreams for integration, testing and dissemination.