AI/ML Direct RF Positioning

Technical Explanation

Classical cellular positioning relies on geometric trilateration of Time-of-Arrival (ToA) and Angle-of-Arrival (AoA), which degrades severely under Non-Line-of-Sight (NLOS) reflections in complex urban canyons and indoor factories. Rel-18 standardizes AI-driven positioning where deep neural networks process raw Channel Impulse Responses (CIR) as spatial fingerprints, inferring sub-meter positioning even when the direct line of sight is obstructed.

Key Functions

• Extracts deep spatial features from multi-path channel impulse responses
• Identifies and corrects severe Non-Line-of-Sight (NLOS) propagation delay bias
• Achieves sub-50 cm horizontal accuracy in dense indoor factory (InF) environments
• Operates without requiring line-of-sight geometric line intersections