RTT Positioning Correction via AI/ML for NLOS Accuracy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing multiple point round trip time positioning algorithms in new radio (NR) are hindered by reliance on line-of-sight conditions, and AI/ML approaches like CIR-based methods are impractical due to sensitivity to clock instability and synchronization requirements.
Innovation Solution
A system that uses a transmit-receive point-based correction module, such as a trained AI/ML model, to correct non-line of sight round trip time data into virtual line of sight data by combining round trip time and angle of arrival data, allowing for accurate location estimation without tight network synchronization.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If multiple point round trip time positioning algorithm is used, then location estimation can be implemented, but positioning accuracy degrades significantly under non-line-of-sight conditions
Solution Approach 1:
The patent introduces an AI/ML-based correction module as an intermediary between the raw RTT measurements and the positioning algorithm. This correction module processes the RTT data to compensate for NLOS effects, thereby maintaining positioning accuracy without requiring strict line-of-sight conditions. The correction module acts as a mediator that transforms imperfect NLOS measurements into corrected values suitable for accurate positioning.
Solution Approach 2:
The patent transforms the RTT measurement parameter by applying AI/ML-based corrections that adjust the measured values to account for NLOS propagation effects. The correction module changes the parameter representation from raw NLOS RTT to corrected RTT values that reflect what the RTT would be under line-of-sight conditions, thereby improving positioning accuracy in NLOS environments.
2Reliability
If CIR-based direct AI/ML approach is used, then line-of-sight dependency is avoided, but the method becomes highly sensitive to clock instability and synchronization variations
Solution Approach 1:
The patent extracts and isolates the clock synchronization sensitivity issue from the positioning process by using RTT-based measurements instead of CIR-based measurements. RTT measurements are inherently more robust to clock drift because they measure round-trip time directly rather than relying on precise timing of signal arrivals that are highly sensitive to clock variations. This extraction of the timing measurement approach reduces sensitivity to synchronization issues.
Solution Approach 2:
The patent employs a practical correction module that can be trained and deployed without requiring extremely tight synchronization infrastructure. The approach uses readily available RTT measurements and applies corrections through a trained model, avoiding the need for expensive, highly synchronized infrastructure that would be required for CIR-based methods to work reliably in NLOS conditions.
3Reliability
If training dataset includes all possible clock behavior variations, then clock-related issues are addressed, but the solution becomes impractical for real system deployments
Solution Approach 1:
The patent performs preliminary training of the AI/ML correction module using a manageable training dataset that captures typical NLOS propagation characteristics. The correction module is pre-trained offline with representative data, and then deployed for real-time positioning without requiring continuous retraining or extremely large datasets. This preliminary action allows the system to handle clock variations and NLOS effects practically.
Solution Approach 2:
The patent uses a practical training dataset that covers the most common and significant NLOS scenarios and clock behavior variations, rather than attempting to include every possible variation. The correction module is trained on representative samples that capture the essential characteristics of NLOS propagation, providing sufficient accuracy for real-world deployments without the impractical complexity of exhaustive training data coverage.
Data Source
AI summary
The technology described herein is directed towards obtaining an estimated location of user equipment, based on measured round trip time data and angle of arrival data to correct the measured round trip time data into virtual round trip time data, including for non-line of sight communication links from an unknown location of a user equipment and a transmit-receive point in an environment. The virtual round trip time data obtained from transmit-receive points is combined into a vector dataset input to a line of sight-based position determination/calculation function to obtain the estimated user equipment location. Correction can be per transmit-receive point, e.g., via a trained AI/ML (artificial intelligence/machine learning) model for the transmit-receive point, analytical function or lookup table-based correction module.