Positioning Measurement Referencing for AI/ML Accuracy in NLOS Conditions
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing wireless positioning systems in 5G NR face challenges in environments with non-line-of-sight paths and low signal-to-interference plus noise ratios, leading to inaccurate positioning.
Innovation Solution
Implementing positioning models trained with AI/ML that dynamically reference measurements based on a common reference point, using measurement configurations to correct for drifts and changes in wireless positioning devices, and applying common referencing methods to ensure accurate positioning outputs.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional wireless positioning methods are used in non-line-of-sight environments, then the system structure remains simple, but positioning accuracy deteriorates due to signal drifts and environmental changes
Solution Approach 1:
The positioning model is trained offline in advance using labeled measurement data to learn the relationship between reference signal measurements and positioning outputs. This preliminary training phase enables the model to compensate for drifts and environmental variations without requiring real-time system adjustments, thereby improving positioning accuracy while maintaining operational simplicity
Solution Approach 2:
An AI/ML positioning model is introduced as an intermediary between the reference signal measurements and the final positioning output. This model processes the measurements and applies learned corrections, enabling accurate positioning in challenging environments without directly modifying the underlying wireless communication infrastructure
2Measurement precision
If reference signal measurements are used without common referencing, then the measurement process is simpler, but positioning accuracy deteriorates due to drifts and changes in wireless positioning devices
Solution Approach 1:
The system implements a feedback mechanism where the positioning model receives reference signal measurements, compares them against learned patterns from training data, and adjusts its output accordingly. This feedback loop enables the system to compensate for drifts and changes in wireless positioning devices, maintaining measurement accuracy over time
Solution Approach 2:
The system dynamically adjusts measurement parameters by applying different referencing methods based on the specific measurement configuration and environmental conditions. The positioning model selects and applies appropriate correction parameters from its training, enabling accurate positioning despite variations in device characteristics and environmental factors
Data Source
AI summary
A wireless positioning device, such as a user equipment (UE) or a transmission reception point (TRP), may receive a measurement configuration. The wireless positioning device may receive a set of positioning signals. The wireless positioning device may measure the set of positioning signals. The wireless positioning device may reference the measured set of positioning signals based on the measurement configuration. The wireless positioning device may output the referenced measured set of positioning signals to a positioning model. The wireless positioning device may output the referenced measured set of positioning signals to a positioning model by transmitting the referenced measured set of positioning signals to a wireless device including the positioning model, by training the positioning model based on the referenced measured set of positioning signals and a set of labels, or by calculating a positioning output using the positioning model based on the referenced measured set of positioning signals.