Indoor Localization Using Dynamic Path Loss Exponent Estimation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing indoor localization methods relying on Wi-Fi signals for device positioning are limited by the need for offline training, which becomes unreliable due to environmental changes, and assume constant path loss exponents, requiring repeated training and failing to adapt to dynamic conditions.
Innovation Solution
A system and method using a log-distance path loss model with Gaussian Processes to iteratively update path loss exponents and device location based on correlations between neighboring locations, allowing unsupervised localization without the need for continuous training, by determining initial estimates and revising path loss exponents using received signal strength measurements from access points.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If offline training is used to measure RSS levels in the indoor environment, then localization accuracy is improved, but the system becomes unreliable when environmental changes occur and requires repeated training
Solution Approach 1:
The patent transforms the static offline training approach into a dynamic online learning system. The path loss exponents are continuously updated based on current RSS measurements and device location estimates, allowing the system to adapt to environmental changes in real-time without requiring repeated offline training campaigns.
Solution Approach 2:
The system performs self-calibration by automatically updating its own path loss exponent parameters using online RSS measurements and location estimates. This eliminates the need for external intervention or repeated offline training, making the system self-adapting to environmental changes.
2Ease of operation
If predetermined path loss exponents are used in the path loss model, then the localization process is simplified, but the values become inaccurate when environment changes occur
Solution Approach 1:
The patent makes the path loss exponents dynamic rather than static. Instead of using predetermined fixed values, the system continuously updates the path loss exponents based on online RSS measurements and device location estimates, maintaining both simplicity and accuracy.
Solution Approach 2:
The system implements feedback by using the device's estimated location and current RSS measurements to continuously refine and update the path loss exponent values. This closed-loop approach ensures that the path loss model remains accurate despite environmental changes.
3Measurement precision
If offline training is repeated whenever the environment changes, then localization accuracy is maintained, but the time and resources required increase significantly
Solution Approach 1:
The patent implements continuous online learning where the path loss exponents are continuously updated during normal operation using RSS measurements and location estimates. This eliminates the need for periodic offline training interruptions, maintaining accuracy without time loss.
Solution Approach 2:
The system performs automatic self-updating of path loss parameters during normal operation, eliminating the need for external training interventions. This continuous self-calibration maintains accuracy while minimizing time and resource consumption.
Data Source
AI summary
A method for tracking a device determines correlations among locations of the device including a set of previous locations of the device and an initial estimate of a current location of the device, and determines, for each access point (AP), a current path loss exponent for the current location of the device using previous path loss exponents determined for the previous locations of the device and the correlations among the locations of the device. The method determines the current location of the device according to a path loss model using received signal strengths (RSS) of signals received from each AP at the current location and the current path loss exponent determined for each AP. The current path loss exponent for each AP are updated using the current location of the device and the RSS of signals received from the corresponding AP.