RT-Fall WiFi Fall Detection Using CSI Phase Difference
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing fall detection systems using WiFi devices are limited by assumptions that users can only perform a few predefined activities and that activities cannot be performed continuously, making them unsuitable for real-world settings where various daily activities are naturally and continuously performed.
Innovation Solution
The system uses the phase difference of Channel State Information (CSI) between two receiver antennas to segment fall and fall-like activities in continuously captured WiFi wireless signal streams, leveraging variance in phase difference and time-frequency domain features for accurate detection, and employs a real-time fall detector called RT-Fall with one transmitter antenna and two receiver antennas.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If wearable sensor-based approaches are used for fall detection, then detection accuracy is improved, but compliance difficulty increases due to always-on-body requirement
Solution Approach 1:
The patent replaces mechanical sensor-based detection with electromagnetic wave-based detection. WiFi signals are transmitted through the environment and reflected off the human body, with the reflected signals captured by receiving antennas. The system processes Channel State Information (CSI) to detect falls, eliminating the need for physical contact between sensors and the user while maintaining detection accuracy.
2Ease of operation
If ambient device-based approaches are used for fall detection, then compliance ease is improved, but false alarm rate increases due to other pressure or sound sources
Solution Approach 1:
The patent extracts specific features from the ambient WiFi signal that are unique to human body interactions. By analyzing the phase and amplitude information of CSI reflected from the human body, the system isolates fall-related signal characteristics from other environmental sources. The use of multiple receiving antennas and signal processing techniques further separates fall events from false alarm sources.
3Area of stationary object
If computer vision-based approaches are used for fall detection, then detection coverage is improved, but privacy intrusion increases and computation complexity increases
Solution Approach 1:
The patent replaces vision-based detection with electromagnetic wave-based detection. Instead of using cameras to capture images or video sequences for scene recognition, the system uses WiFi signal reflections and processes CSI to detect falls. This substitution eliminates the need for intensive real-time video processing while maintaining detection coverage and avoiding privacy concerns associated with visual monitoring.
4Device complexity
If prior art WiFi fall detection is used with predefined activities assumption, then device complexity is reduced, but adaptability decreases for real-world settings
Solution Approach 1:
The patent implements a dynamic detection framework that adapts to various human activities without requiring predefined activity categories. The system continuously monitors WiFi signal characteristics and automatically adjusts its detection parameters based on the observed signal patterns. This dynamic approach allows the system to handle diverse real-world scenarios including walking, standing, sitting, and falling, while maintaining relatively simple device architecture.
Data Source
AI summary
The present invention relates to a fall detection method and system. The fall detection method comprises: receiving, by a first receiving antenna, a first WiFi signal stream propagating through an environment; receiving, by a second receiving antenna, a second WiFi signal stream propagating through the environment; determining a physical layer Channel State Information (CSI) stream, namely, a first CSI stream, of the first WiFi signal stream; determining a physical layer CSI stream, namely, a second CSI stream, of the second WiFi signal; determining a phase difference, namely, a CSI phase difference, between respective phase of the physical layer CSI stream of the first WiFi signal stream and the physical layer CSI stream of the second WiFi signal stream at the same time point, to form a CSI phase difference stream; and determining, according to the CSI streams and the CSI phase difference stream, a fall event.