Unsupervised Learning Signal Map for Wireless Location Tracking
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional methods for locating mobile units in wireless communication systems, especially indoors, are labor-intensive and require costly hardware, as they rely on manual signal mapping or extensive sensor deployments, which are inefficient and prone to environmental changes.
Innovation Solution
The use of unsupervised learning techniques to form and dynamically maintain signal maps using unlabeled signal data, relating signal values to locations through mean signal functions, eliminating the need for manual data collection and extensive sensor deployments.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual signal mapping is used to locate mobile units, then location tracking can be achieved, but the process becomes time-consuming and laborious
Solution Approach 1:
The system uses existing mobile unit transmissions to automatically populate and update the signal map. Mobile units serve themselves by providing the signal data needed for location tracking, eliminating the need for manual data collection. The unsupervised learning algorithm automatically processes this self-provided data to maintain accurate location information.
Solution Approach 2:
The patent replaces manual mechanical data collection processes with automated computational methods. Instead of physically moving test users through locations to collect signal data, the system uses unsupervised learning algorithms to automatically analyze signal transmissions and construct signal maps computationally.
2Reliability
If manual signal mapping is used, then location data can be collected, but constant updates require regular repetition of the time-consuming data gathering process
Solution Approach 1:
The system continuously processes signal transmissions from mobile units to maintain the signal map. Instead of periodic manual updates, the unsupervised learning algorithm continuously analyzes incoming signal data, ensuring the signal map remains current with environmental changes without interrupting normal operations.
Solution Approach 2:
The signal map automatically updates itself using ongoing signal transmissions from mobile units. The system serves itself by continuously collecting and processing its own operational data, eliminating the need for external manual intervention to maintain accuracy.
3Measurement precision
If a network of sensors is deployed throughout the interior space, then automatic location tracking is achieved, but costly hardware and extensive deployment are required
Solution Approach 1:
The system creates a computational model (signal map) that represents the physical signal environment without requiring physical sensors at every location. The signal map is a virtual copy of the signal characteristics across the space, constructed from limited measurements and propagated through unsupervised learning to cover the entire area.
Solution Approach 2:
Existing mobile units and base stations perform multiple functions: they provide communication services and simultaneously serve as signal sources for location tracking. The system eliminates the need for dedicated sensor hardware by making the communication infrastructure itself serve the dual purpose of data collection and location determination.
4Measurement precision
If sensors are deployed at known locations, then labeled data can be obtained, but a sufficiently large number of sensors are needed to maintain sufficient location accuracy
Solution Approach 1:
The system uses a small number of actual signal measurements and computationally generates the complete signal map through unsupervised learning. Instead of placing sensors everywhere, the algorithm copies and propagates signal characteristics from limited measurement points to reconstruct the entire spatial signal environment.
Solution Approach 2:
The system transitions from a discrete sensor deployment approach to a continuous computational model. By moving to the mathematical dimension of signal space and using unsupervised learning, the system achieves complete spatial coverage from minimal physical measurements, effectively adding a computational dimension to solve the physical constraint of sensor quantity.
Data Source
AI summary
The present invention provides a method of unsupervised learning and location for tracking users in a wireless communication system. One embodiment of the method includes forming a signal map of a geographic area using unlabeled values of one or more signals so that the signal map relates locations in the geographic area to values of the signal(s).