AI Positioning Model Switching for Low-Overhead NR Accuracy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing NR positioning methods face challenges in resource overhead and accuracy, particularly with timing-based and angle-based approaches, and lack support for AI/ML models in signaling.
Innovation Solution
Implementing AI/ML models for positioning at user equipment (UE), network nodes, and location servers, with triggers based on line of sight (LOS) conditions to activate or switch between AI/ML and legacy positioning methods, reducing resource utilization and improving accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If timing-based positioning is used, then positioning can be implemented, but resource overhead increases
Solution Approach 1:
The patent changes the fundamental parameter of positioning methodology from traditional timing-based or angle-based approaches to AI/ML-based positioning. The location server implements AI/ML models that process positioning data differently, using machine learning algorithms to estimate position based on received signals rather than conventional timing measurements, thereby reducing resource overhead while maintaining accuracy
Solution Approach 2:
The patent substitutes the mechanical/measurement-based positioning system with an intelligent system. Instead of relying on precise timing measurements and signal processing mathematics, the system uses AI/ML models trained on positioning data to automatically determine position, replacing the traditional measurement and calculation mechanism with an intelligent decision-making mechanism
2Measurement precision
If angle-based positioning is used, then positioning resolution improves, but the number of antenna elements required increases
Solution Approach 1:
The patent fundamentally changes the positioning parameter from angle-based measurement to AI/ML-based estimation. The location server uses trained models to infer position directly from received signals, eliminating the need for complex angle measurements that would require multiple antenna elements, thus achieving high resolution with fewer antennas
Solution Approach 2:
The patent uses AI/ML models that have been trained (copied from training data with ground truth labels) to replicate the positioning function. Instead of physically implementing complex antenna arrays for angle measurement, the system copies the positioning capability through trained machine learning models that can estimate position from standard signaling
3Measurement precision
If AI/ML model is introduced for positioning, then positioning accuracy improves, but signaling support and configuration complexity increase
Solution Approach 1:
The patent implements self-service mechanisms where the location server autonomously manages the AI/ML positioning process. The server automatically selects appropriate models, configures them based on network conditions, and activates them without requiring complex manual signaling or configuration, thereby reducing the burden on the network while maintaining high accuracy
Solution Approach 2:
The patent introduces dynamic model selection and activation mechanisms. The location server can dynamically choose between different AI/ML models or traditional positioning methods based on current network conditions, UE capabilities, and service requirements, allowing flexible adaptation without fixed complex configuration
Data Source
AI summary
A method for positioning is disclosed. The method may include receiving, by a user equipment (UE), a threshold configured by a location server, determining, by the UE, a criterion associated with the threshold, activating, by the UE, a positioning model for positioning according to the criterion, monitoring, by the UE, a performance of the positioning model, and reporting, to the location server by the UE, the performance of the positioning model. The method may further be implemented at a network node (e.g., a base station) or a location server, which improves the efficiency of positioning.