ML Geolocation Using RF and Topography
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current UE geolocation technologies in wireless communication systems, such as LTE and 5G, face inaccuracies and inefficiencies due to reliance on single measurement types, neglect of historical data, and lack of topographical information, leading to unsatisfactory accuracy, especially in suburban and rural areas.
Innovation Solution
A machine learning model is trained using historical network communication data and topographical information to estimate UE locations, incorporating known UE locations and external data for improved precision and real-time geolocation services.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional triangulation methods are used for UE geolocation, then the system can operate with existing cellular infrastructure, but the location accuracy deteriorates to hundreds of meters especially in suburban and rural areas
Solution Approach 1:
The patent combines multiple data sources including RF measurements, historical location data, topographical information, and road network data into a unified location estimation system. This integration allows the system to achieve high accuracy (within tens of meters) by leveraging complementary information from different sources, particularly improving performance in suburban and rural areas where traditional triangulation fails.
Solution Approach 2:
The system performs preliminary actions by collecting and storing historical location data, RF measurement data, and topographical information in advance. This pre-collected data is then used to train machine learning models and create location profiles that enable rapid and accurate real-time geolocation without requiring complex real-time triangulation calculations.
2Measurement precision
If single measurement types are used for geolocation, then the system remains simple to implement, but the location accuracy deteriorates significantly
Solution Approach 1:
The patent merges multiple measurement types including RF signal strength, timing advance, historical location data, and topographical features into a comprehensive location estimation framework. This multi-measurement approach enables the system to achieve accurate geolocation by cross-validating information from different sources and compensating for limitations of individual measurement types.
Solution Approach 2:
The system creates a composite location estimation model that integrates diverse data types analogous to composite materials. By combining RF measurements with historical data, topographical information, and road network data, the system achieves superior location accuracy that cannot be obtained through any single measurement type alone.
3Measurement precision
If historical information is ignored in geolocation, then the processing remains simple and fast, but the accuracy deteriorates due to lack of contextual data
Solution Approach 1:
The system performs preliminary actions by collecting, storing, and pre-processing historical location data and RF measurements in advance. This historical data is organized into location profiles and used to train machine learning models beforehand, enabling the system to leverage historical information for improved accuracy without adding significant processing time during real-time geolocation operations.
4Measurement precision
If topographical information is not utilized, then the system remains computationally simple, but the geolocation accuracy deteriorates in challenging environments
Solution Approach 1:
The patent integrates topographical information including building polygons, road networks, and terrain data with RF measurements and historical location data. This combination allows the system to improve location accuracy in challenging environments by constraining possible locations to physically plausible areas and using topographical features as additional location indicators.
Data Source
AI summary
The described technology is generally directed towards user equipment (UE) geolocation. A machine learning model can be trained to estimate UE locations based on historical network communication data associated with the UEs. In order to train the machine learning model, known previous UE locations and corresponding historical network communication data can be provided to the machine learning model. A variety of other information, such as topographical information, can also be provided to the machine learning model. The machine learning model can be trained to predict the known previous UE locations based on the corresponding historical network communication data and any other provided information. Once it is trained, the machine learning model can be deployed to estimate real-time UE locations based on historical network communication data associated with the UEs.