Identification of Hearing Impaired People in Crowded Environments Using Wi-Fi Signals

In this article, Wi-Fi signals are used to find hearing-impaired people in a crowded environment with the help of different gestures namely “empty, one-hand right, one-hand left, two-hand right, and two-hand left”. The existing system for recognizing hearing-impaired people is based on cameras. This has some drawbacks, such as poor photo quality at night, privacy concerns, and the high cost of putting these systems into everyday life. The data collected from the Wi-Fi is represented in the form of channel state information (CSI) values. Five activities were performed by the subject namely empty, one-hand right, one-hand left, two-hand right, and two-hand left. Support vector machine (SVM) (Linear SVM), Ensemble (Subspace discriminant), and neural network pattern recognition were performed on collected CSI values. The simulation results showed that 94.7% of classification accuracy was achieved by neural network pattern recognition while classifying gesture data.