Design And Research Of Digital Key Recognition Based On WiFi Signal

In recent years,the development and utilization of modern communication systems has become a hot topic of research.The realization of new human-computer interaction has become the focus of researchers.Digital keypad is a common way of human-computer interaction,which is commonly used to unlock mobile phone keys,pay with digital password and so on.On the one hand,as the relationship between people in the social environment becomes more and more close,the possibility of dangers such as privacy leakage and security breaches caused by the identification of digital keystrokes by WiFi is becoming higher and higher in the public social environment.On the other hand,with the gradual improvement of the working bandwidth of hardware devices and WiFi protocols,it is possible to implement a virtual digital keyboard for WiFi signals,which can bring about major changes for the development of human-computer interaction and people's lifestyle.In this context,this paper designs a WiDig system for identifying digital keystrokes under the WiFi signal.Through data acquisition and experiment,our results show that WiDig can average reach 90.2% of the user recognition rate,85% of the digital keystroke recognition rate in indoor environment.This paper is divided into two parts,detection and extraction of digital keystroke waveforms and matching and identification of digital keystroke waveforms.1.Detection and extraction phase of digital keystroke waveforms.First,the original CSI(Channel State Information)value is extracted from the wireless network card for smoothing,then the Butterworth low-pass filter is used to filter out highfrequency noise,principal component analysis is used to simplify dimensionality reduction,and the ReliefF+ algorithm is put forward to select subcarriers,which can improve classification accuracy.2.Matching and identification phase of digital keystroke waveforms.Firstly,based on wavelet analysis and discrete wavelet transform,the sample data volume is reduced,and the short-time energy segmentation window algorithm is used to obtain the specific mode of the digital keystroke and the duration of detecting the digital keystroke.Then,a keystroke detection algorithm that combines the time window is proposed to estimate the start and end time of the digital keystroke.The distance between the feature waveforms is then calculated using dynamic time warping.Finally,the K-nearest neighbor algorithm is used to classify the keystrokes and users to obtain the final recognition result.