Machine Learning User Device Positioning from CSI in NLOS Networks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing indoor positioning technologies face challenges in achieving sub-meter accuracy due to harsh radio environments and multi-path scattering, with camera-based systems being costly and radio systems providing only meter-level accuracy, while 3GPP developments for 5G and 6G networks struggle to improve precision in non-line-of-sight conditions.
Innovation Solution
A machine learning-based approach using neural networks to process channel measurement data from locator devices, embedding time and angle of arrival information in latent space vectors, and applying a third neural network to estimate user device position, potentially enhanced by auxiliary position estimators like maximum likelihood estimators.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If camera-based systems are used for indoor positioning, then positioning accuracy is improved, but cost and device complexity increase significantly
Solution Approach 1:
The patent replaces camera-based optical/mechanical positioning systems with a radio-frequency based neural network system. Instead of using cameras to capture visual data for positioning, the system uses radio channel measurements (CSI) processed through neural networks to estimate position, thereby achieving high accuracy without the high cost and complexity of camera systems.
Solution Approach 2:
The patent transforms the approach by changing from direct visual measurement to indirect radio channel parameter analysis. It extracts positioning information from channel state information (CSI) parameters and uses neural networks to map these radio parameters to position estimates, achieving sub-meter accuracy through parameter transformation rather than direct optical measurement.
2Device complexity
If radio systems relying on RSS are used for positioning, then cost is reduced, but positioning accuracy deteriorates to several meters
Solution Approach 1:
The patent fundamentally changes the radio parameters used for positioning. Instead of relying on received signal strength (RSS), it uses channel state information (CSI) which contains phase and amplitude information across multiple frequencies. This parameter change enables the neural network to extract much more precise positioning information, achieving sub-meter accuracy with cost-effective radio systems.
Solution Approach 2:
The patent introduces neural networks as an intermediary between radio channel measurements and position estimation. The neural network acts as a mediator that processes complex CSI data and transforms it into accurate position estimates, bridging the gap between inexpensive radio measurements and high-precision positioning requirements.
3Device complexity
If traditional positioning methods are used in harsh radio environments, then implementation is simplified, but positioning accuracy deteriorates due to multi-path scattering and non-line-of-sight conditions
Solution Approach 1:
The patent converts the harmful effect of multi-path scattering into a beneficial signal characteristic. Instead of treating multipath components as noise to be eliminated, the neural network learns to recognize and utilize the distinctive patterns in multipath channel responses to accurately determine position, even in non-line-of-sight conditions. This transforms the previously harmful scattering into a useful positioning indicator.
Solution Approach 2:
The neural network serves as an intermediary that handles the complexity of harsh radio environments. It processes the distorted and scattered radio signals, extracting positioning information that would be impossible to obtain with traditional methods. The neural network absorbs the environmental complexity, providing simplified and accurate position estimates despite challenging conditions.
Data Source
Figure 1
Figure 2
Figure 3A
AI summary
Devices, methods and computer programs for machine learning (ML) -based user device positioning in wireless networks are disclosed. At least some example embodiments may allow processing measurements done in positioning to cope with challenging environments, non-line-of-sight (NLOS) conditions, and/or dense multipath scattering.