Generalized RF Fingerprinting with Non-RF Factors
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional Wi-Fi positioning systems face challenges in scaling and accuracy due to reliance on labor-intensive calibration and limited use of non-RF related information, leading to increased errors and inaccuracies in location determination.
Innovation Solution
Incorporating non-RF related factors such as GPS quality, device type, client identification, and operating system information into the fingerprinting methodology, with a distance function derived from training datasets to optimize location determination and predict errors, allowing for the use of alternative location methods when necessary.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional fingerprinting approaches are used, then location determination can be performed using RF signal strength, but labor-intensive calibration is required and scalability is difficult
Solution Approach 1:
The system enables self-service by allowing mobile devices to automatically contribute their own location data and RF fingerprint measurements to the database without requiring manual calibration. Devices perform self-calibration by comparing their measured RF signals against predicted values generated from crowd-sourced location data, eliminating the need for labor-intensive manual calibration while maintaining location accuracy.
Solution Approach 2:
The patent transforms the calibration process by changing from manual parameter input to automated parameter extraction. Instead of manually measuring and inputting RF signal parameters during calibration, the system automatically extracts parameters from crowd-sourced mobile device data, converting a labor-intensive process into an automated scalability-friendly approach.
2Measurement precision
If conventional fingerprinting approaches are used, then location positioning can be achieved, but accuracy is difficult to evaluate without external data
Solution Approach 1:
The patent introduces an intermediary evaluation mechanism that uses predicted location data as a mediator to assess actual location accuracy. By comparing crowd-sourced location measurements against predicted locations derived from RF fingerprints and non-RF data, the system creates an indirect but effective method to evaluate accuracy without requiring complex external validation infrastructure.
Solution Approach 2:
The system achieves multi-functionality by using the same crowd-sourced data for both location determination and accuracy evaluation. The database serves multiple purposes: storing RF fingerprints for positioning, storing crowd-sourced location measurements for validation, and providing training data for predictive models, thereby simplifying the system while enabling comprehensive accuracy assessment.
3Measurement precision
If conventional fingerprinting approaches are used, then location determination can be performed, but non-RF related information is not effectively utilized
Solution Approach 1:
The patent merges RF signal data with non-RF related information including GPS coordinates, device orientation, motion data, and environmental context into a unified location determination framework. By combining multiple data sources and fusing them through predictive models, the system enhances location accuracy while demonstrating versatile adaptability to utilize diverse data types from mobile devices.
Solution Approach 2:
The system creates a composite information structure by integrating RF fingerprint data with non-RF parameters such as GPS location, device metadata, and contextual information. This composite approach mirrors composite materials in engineering, where combining different data types produces a more robust and accurate location determination system than using RF data alone.
Data Source
AI summary
An RF fingerprinting methodology is generalized to include non-RF related factors. For each fingerprinted tile, there is an associated distance function between two fingerprints (the training fingerprint and the test fingerprint) from within that tile which may be a linear or non-linear combination of the deltas between multiple factors of the two fingerprints. The distance function for each tile is derived from a training dataset corresponding to that specific tile, and optimized to minimize the total difference between real distances and predicted distances. Upon receipt of an inference request, a result is derived from a combination of the fingerprints from the training dataset having the least distance per application of the distance function. Likely error for the tile is also determined to ascertain whether to rely on other location methods.