Similarity Learning for Crowd-Sourced UE Positioning Accuracy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current wireless communication systems, particularly 5G NR, face challenges in accurately and efficiently determining the position of user equipment (UE) due to limitations in multiple-access technologies and positioning methods.
Innovation Solution
The implementation of a method that utilizes machine learning (ML) models to estimate the position of UE by analyzing measurements from neighboring cells and reference UEs, leveraging common cells and similarity scores to enhance positioning accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional positioning methods are used in 5G NR systems, then the positioning process is simpler, but the positioning accuracy and efficiency deteriorate
Solution Approach 1:
The patent introduces machine learning models as an intermediary between raw measurement data and position estimation. The ML model processes measurements from multiple cells and reference UEs, extracting meaningful patterns that traditional methods miss. This intermediary layer enables high positioning accuracy by learning complex relationships in the data without requiring overly complex positioning algorithms.
Solution Approach 2:
The patent replaces traditional geometric and signal-based positioning mechanics with data-driven machine learning approaches. Instead of relying on precise geometric relationships and signal propagation models, the system uses ML models to learn positioning patterns from measurement data, substituting physical modeling with statistical learning to achieve better accuracy.
2Measurement precision
If more measurement data from multiple sources is collected for positioning, then positioning accuracy improves, but processing time and latency increase
Solution Approach 1:
The patent performs preliminary actions by pre-training machine learning models offline using large datasets. The models learn positioning patterns in advance, so during actual positioning operations, they can quickly process new measurement data without requiring extensive real-time computation. This preliminary learning phase separates heavy computation from time-critical positioning operations.
Solution Approach 2:
The patent changes the parameter processing approach by using ML models to learn optimal feature representations and similarity metrics. Instead of processing all raw measurement parameters directly, the system transforms measurements into learned feature spaces where positioning decisions can be made more efficiently, reducing processing time while maintaining accuracy.
3Measurement precision
If machine learning models are used to analyze measurements from multiple sources, then positioning accuracy improves, but computational complexity increases
Solution Approach 1:
The patent uses copying by creating similarity representations of reference UEs and comparing them to the target UE. The ML model learns to generate compact similarity scores that capture essential positioning information without requiring full replication of complex processing for each comparison. This copying approach reduces computational burden while maintaining accuracy.
Solution Approach 2:
The patent segments the positioning problem into distinct processing stages: measurement collection, ML-based similarity analysis, and position estimation. The ML model handles the complex pattern recognition task separately from data collection and final position calculation. This segmentation allows each component to be optimized independently, reducing overall computational complexity.
Data Source
AI summary
Aspects presented herein may enhance the accuracy and/or latency of UE positioning based on crowd-sourcing, where a network entity may compute a position estimate of a UE based on neighbor-cell scan data from the UE and one or more reference UEs. In one aspect, a network entity receives a first set of measurements associated with at least one cell from a UE. The network entity performs a position estimation of the UE based on at least one of the first set of measurements associated with the at least one cell, a second set of measurements for each of a set of reference UEs, or a location of each of the set of reference UEs via an ML model, where the UE and the set of reference UEs include at least one common cell.