Position Error Outlier Classifier for Satellite Navigation Integrity
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Position errors in satellite navigation systems do not follow a normal distribution, leading to challenges in error correction and integrity assessment, as outliers are either underrepresented or overrepresented, affecting navigation system availability and integrity.
Innovation Solution
A classifier is developed to identify outliers in position error distributions using pattern recognition based on boundary conditions, trained with machine learning to distinguish between position information with high errors and valid data, discarding outliers before navigation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If statistical upper error limits are calculated based on normal distribution, then position information with lower position error is well represented, but outliers are underrepresented
Solution Approach 1:
The patent changes the distribution assumption from normal distribution to heavy-tailed distribution to accurately model position errors. This parameter change allows the system to properly represent both typical position errors and outliers, resolving the contradiction between representing low-error data well and detecting outliers accurately.
Solution Approach 2:
The patent replaces traditional statistical methods based on normal distribution with machine learning-based classification. This substitution enables the system to learn complex patterns in position error data and automatically distinguish between typical errors and outliers, improving both representation accuracy and detection reliability.
2Reliability
If statistical upper error limits are calculated based on normal distribution, then outliers are well covered, but position information with lower position error is overestimated
Solution Approach 1:
The patent changes from using normal distribution parameters to heavy-tailed distribution parameters, which allows the system to cover outliers adequately while avoiding the overestimation of typical position errors. This parameter adjustment ensures that both outlier coverage and error estimation precision are improved simultaneously.
Solution Approach 2:
The patent replaces normal distribution-based statistical methods with machine learning classification that can adapt to the actual heavy-tailed error distribution. This substitution enables the system to accurately estimate position errors for both typical and outlier cases, resolving the contradiction between outlier coverage and error estimation precision.
3Reliability
If position information is sorted by position error, then integrity and system availability are increased, but device complexity increases
Solution Approach 1:
The patent performs preliminary classification of position information into outlier and non-outlier categories before the actual navigation processing. This preliminary action separates the complex task of handling outliers from routine navigation processing, improving system availability while keeping the overall device complexity manageable through structured processing stages.
Solution Approach 2:
The patent replaces complex manual sorting and analysis methods with automated machine learning classification. This substitution reduces the operational complexity of handling position errors while improving system availability through automated outlier detection and exclusion from navigation calculations.
Data Source
AI summary
A classifier for identifying outliers of a position error distribution of position information is disclosed. The classifier is configured to identify position information with an increased position error as an outlier using the position information and temporally correlating boundary conditions.