Machine Learning Positioning Database Using Wireless Feature Matching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional positioning technologies face challenges in providing accurate location information for emergency services indoors and in dense urban areas, where GPS signals are weak, and existing methods for creating positioning databases are inefficient, requiring large amounts of data and computation time, and are prone to errors due to differences between outdoor and indoor environments.
Innovation Solution
A machine-learning-based method using matching feature points from wireless communication infrastructure to create a positioning database in real-time, which extracts matching feature points from collected and positioning data to estimate optimal composite locations, independent of signal strength, and represents the output as a matching probability density function, allowing for efficient combination of data from multiple positioning resources.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional positioning database creation methods are used, then positioning accuracy can be maintained, but computation time and storage requirements increase significantly
Solution Approach 1:
The patent pre-processes and stores feature point information from wireless communication infrastructure during database creation, organizing it into matchable formats before actual positioning is needed. This preliminary organization of feature points (extracting, storing, and indexing them in advance) enables rapid matching during runtime without performing heavy computation at the moment of positioning, thus resolving the contradiction between maintaining accuracy and reducing computation time
Solution Approach 2:
The patent extracts and isolates key feature points from the collected wireless communication data, separating only the essential matching elements (feature point coordinates, identifiers, and characteristics) from the complete raw dataset. By extracting only the necessary feature information needed for matching rather than processing entire datasets, the system maintains positioning accuracy while significantly reducing the computational burden during actual positioning operations
2Measurement precision
If conventional positioning database creation methods are used, then positioning accuracy can be maintained, but storage requirements increase significantly
Solution Approach 1:
The patent extracts and stores only the essential feature point information (coordinates, identifiers, and matching characteristics) from the complete wireless communication datasets. By isolating and storing only the critical matching elements rather than entire raw datasets, the system maintains positioning accuracy while significantly reducing the storage space required for the positioning database
Solution Approach 2:
The patent segments the positioning database into structured feature point records organized by location and wireless infrastructure type, dividing the data into manageable, indexed units. This segmentation allows the system to store only relevant feature information in an organized manner, reducing overall storage requirements while enabling efficient retrieval for maintaining positioning accuracy
3Productivity
If outdoor collection methods are used for positioning database creation, then data collection can be performed, but indoor positioning accuracy deteriorates
Solution Approach 1:
The patent uses wireless communication infrastructure (base stations, Wi-Fi access points, Bluetooth beacons) as intermediary objects that exist both outdoors and indoors. By extracting feature points from these infrastructure elements during outdoor collection and storing them in a unified database, the system enables accurate indoor positioning through matching against the pre-collected feature data, thus resolving the contradiction between outdoor collection efficiency and indoor positioning accuracy
Solution Approach 2:
The patent creates a virtual copy of the wireless communication infrastructure's spatial distribution and signal characteristics in the positioning database during outdoor collection. This copied feature point information represents the indoor wireless environment without requiring physical indoor data collection, enabling accurate indoor positioning through matching while maintaining outdoor collection efficiency
4Ease of operation
If signal strength-based matching is used, then positioning can be performed, but positioning accuracy deteriorates due to environmental differences
Solution Approach 1:
The patent changes the matching parameters from signal strength (RSSI) to geometric and topological features of wireless infrastructure (coordinates, identifiers, spatial relationships). By transforming the matching basis from environment-dependent signal strength to stable structural features, the system maintains simple positioning operations while achieving accuracy independent of environmental conditions like indoor vs. outdoor settings
Data Source
AI summary
Disclosed herein are an apparatus and method for positioning of uncollected points based on machine learning using matching wireless communication infrastructure points. The apparatus includes memory in which at least one program according to an embodiment is recorded and a processor for executing the program. The program may compare collected data acquired from wireless communication infrastructure with positioning data measured by a positioning target terminal and thereby extract matching feature points; create a fingerprint database of global grid cells, including uncollected points, for the extracted feature points in real time; and estimate the optimal composite location of the positioning target terminal based on the created fingerprint database.