GNSS Residual Grid Localization via Graph Neural Networks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional GNSS positioning methods, especially 'first fix' positioning, face challenges due to non-line of sight signals and multipath issues, leading to prolonged time in determining the location of a GNSS receiver, especially in urban environments.
Innovation Solution
The method employs residual grid representation and processing using graph neural networks to quickly and efficiently determine the location of a GNSS device by creating and aggregating residual grids based on pseudorange measurements, accounting for errors like receiver clock bias and atmospheric delays, and integrating this information with traditional weighted least squares algorithms.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of time
If traditional GNSS positioning methods are used, then the positioning process can be performed with conventional algorithms, but the time required to determine location is prolonged due to non-line of sight signals and multipath issues
Solution Approach 1:
The patent segments the positioning problem into two distinct phases: a coarse positioning phase that quickly eliminates impossible locations using residual grids, and a fine positioning phase that refines the location estimate. This segmentation allows the system to rapidly filter out NLOS and multipath affected areas before performing detailed positioning calculations, thereby reducing overall positioning time while maintaining accuracy.
Solution Approach 2:
The patent performs preliminary action by pre-computing residual grids that represent possible device locations before actual positioning occurs. These residual grids are created based on satellite geometry and pseudorange measurements, establishing a framework of probable locations that speeds up the subsequent positioning process by eliminating the need to evaluate all possible locations from scratch.
2Productivity
If residual grid representation is used, then the time required for positioning is reduced and accuracy is improved, but the device complexity increases due to graph neural network processing
Solution Approach 1:
The patent substitutes traditional iterative mathematical optimization methods with a graph neural network-based approach. Instead of using conventional algorithms that require multiple iterations to converge on a position solution, the system uses a trained neural network that can directly infer location from residual grids, significantly reducing computational complexity and processing time despite the increased model complexity.
Solution Approach 2:
The patent creates simplified residual grid representations that capture the essential geometric relationships between satellites and potential device locations. These residual grids serve as compressed copies of the full positioning problem, allowing the graph neural network to work with reduced-dimensional data that maintains positioning accuracy while reducing computational burden.
Data Source
AI summary
In some implementations, a global navigation satellite system (GNSS) device may determine its approximate location, and, for each pseudorange measurement of a plurality of pseudorange measurements performed by the GNSS device: determine a location of a respective satellite vehicle (SV) that transmits a respective GNSS signal of which the pseudorange measurement is performed, and determine a respective residual grid, where the respective residual grid is based on respective information from the pseudorange measurement and the location of the respective SV, and the respective residual grid is indicative of possible locations of the GNSS device within a geographical region including the approximate location of the GNSS device. The GNSS device may aggregate the residual grids corresponding to at least a portion of the plurality of pseudorange measurements and may determine a location estimate of the GNSS device based on the aggregation of the residual grids.