Wireless CSI Person Identification for Privacy-Preserving Presence Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing biometric identification methods, such as facial recognition and fingerprint reading, are invasive to privacy and prone to data theft, while traditional wireless signal-based methods lack the ability to detect human presence and adapt to personal preferences.
Innovation Solution
A system and method using wireless signals, particularly Wi-Fi channel state information (CSI), to identify individuals by analyzing gait and other biometric features through machine learning, enabling personalized control of machines based on detected presence and usage patterns.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional biometric methods (facial recognition, fingerprint reading) are used for person identification, then identification accuracy is improved, but privacy security deteriorates due to data theft risks
Solution Approach 1:
The patent replaces traditional biometric sensing systems (cameras, fingerprint scanners) with a wireless signal-based identification system. The machine learning model processes channel state information from wireless packets to extract gait and biometric features, eliminating the need for direct physical or visual contact between the identification system and the person being identified. This substitution maintains identification accuracy while improving privacy security by using indirect wireless signal measurements instead of direct biometric data collection.
2Object-affected harmful factors
If wireless signal-based methods are used for person identification, then privacy security is improved, but the ability to detect human presence and adapt to personal preferences deteriorates
Solution Approach 1:
The patent segments the channel state information into distinct temporal segments and processes each segment through the machine learning model independently. This segmentation allows the system to detect different aspects of human presence and behavior from the same wireless signal source, enabling both privacy-preserving identification and adaptive presence detection without requiring additional sensors or invasive methods.
3Measurement precision
If channel state information is collected and processed through machine learning models, then person identification accuracy is improved, but system complexity increases
Solution Approach 1:
The patent implements a universal machine learning model that performs multiple functions: extracting gait features, identifying biometric characteristics, detecting human presence, and enabling adaptive control of machines. By consolidating these functions into a single model that processes channel state information segments, the system achieves high identification accuracy while managing complexity through multi-functionality rather than requiring separate systems for each task.
Data Source
AI summary
Systems and methods for person identification using wireless signals is disclosed. Channel state information is collected from wireless signals received by a wireless receiver. The channel state information (CSI) is pre-processed to determine a gait and biometric features including body shape and height associated with a particular person based on selected CSI segments, which are annotated with a class indicative of the gait. The particular person is identified, using a machine learning model, based on the gait and the other biometric features, wherein the machine learning model is trained using classifier training and training data including the gait and information from the CSI segments, from which one or more parameters is determined. After training, the person can be identified based on packets from received wireless signals. The method further includes controlling a machine, in response to detecting the presence of the person, based on established usage patterns.