R-DEHM: CSI-Based Robust Duration Estimation of Human Motion with WiFi

Abstract

As wireless sensing has developed, wireless behavior recognition has become a promising research area, in which human motion duration is one of the basic and significant parameters to measure human behavior. At present, however, there is no consideration of the duration estimation of human motion leveraging wireless signals. In this paper, we propose a novel system for robust duration estimation of human motion (R-DEHM) with WiFi in the area of interest. To achieve this, we first collect channel statement information (CSI) measurements on commodity WiFi devices and extract robust features from the CSI amplitude. Then, the back propagation neural network (BPNN) algorithm is introduced for detection by seeking a cutting line of the features for different states, i.e., moving human presence and absence. Instead of directly estimating the duration of human motion, we transform the complex and continuous duration estimation problem into a simple and discrete human motion detection by segmenting the CSI sequences. Furthermore, R-DEHM is implemented and evaluated in detail. The results of our experiments show that R-DEHM achieves the human motion detection and duration estimation with the average detection rate for human motion more than 94% and the average error rate for duration estimation less than 8%, respectively.

A Review of the State of the Art in Non-Contact Sensing for COVID-19

“…This technique can be used for monitoring the vital signs of patients independent of their activities [ 83 ]. The RF signals detect the movement by observing the Channel State Information (CSI), which can show amplitudes of the RF signals while movement occurs between a RF transmitter and receiver [ 84 , 85 ]. The Emerald system has been developed to monitor COVID-19 patients using RF signals.…”

Higher Order Feature Extraction and Selection for Robust Human Gesture Recognition using CSI of COTS Wi-Fi Devices

“…WFID [40] also applies PCA to uniquely identify humans from groups of nine and six people, and achieves an accuracy of 91.9% and 93.1% respectively with SVM. R-DEHM [41] achieves 94% accuracy in detecting human movements with an average error rate of 8% for duration estimation by implementing a back propagation neural network (BPNN) algorithm on principal components of CSI. R-TTWD [42] applies the majority vote algorithm on PCA based features to identify the presence and absence of human movement using one class SVM with true positive >99%.…”

Sign language gesture recognition with bispectrum features using SVM

“…State of the art device free sensing leveraging CSI applies PCA, a feature extraction method on the pre-processed CSI traces. For example, PCA was adopted in detection application such as localizing human, and able to achieve 97% accuracy using SVM with less localization error in across different environment [14].The other works such as WFID [15], R-TTWD [16], FallDeFi [17], R-DEHM [18] and WiFind [19] uses PCA and achieves remarkable recognition accuracy. BodyScan [20] computes power spectral density (PSD) from PCA features for recognizing activities with 72.5% and detecting breathing rate with 97.4% accuracy.…”