Research On Identity Authentication Based On Channel State Information

With the rapid development and prosperity of technology construction,WiFi signals have been widely used and applied in public life.As an emerging research hotspot,WiFi signal authentication is gaining widespread attention in various fields such as ubiquitous computing,human-computer interaction,and intrusion detection.WiFi signal research has the advantages of simple deployment mode,concealed deployment location,wide application scenarios,and low cost.At the same time,channel state information in the WiFi signal can be used to analyze human behavioral gestures and identity features,thereby providing a powerful data foundation for identity authentication technology.First of all,each person's gestures have some subtle and unique behavioral characteristics,including but not limited to factors such as exercise cycle and distance of motion.Experiments have shown that under the same environmental conditions,the amplitude changes of channel state information caused by different people's gestures are different.Therefore,this paper proposes a method for completing identity authentication using channel state information in a gesture scenario.After the denoising and dimension reduction process of the data,the method segments the action interval and the action interval in the data through a window-based action interval segmentation algorithm,and analyzes the movement of the gesture according to the segmentation result..In order to achieve identity authentication for different gestures,this paper applies the SVM classifier to recognize the gestures of the action interval data.In the process of motion analysis,the Hilbert transform is used to calculate the motion distance and motion velocity in the motion interval,and form the feature set of the motion sequence.Finally,BP neural network is used to complete the identification process of various gestures.In the experimental analysis stage,the authentication accuracy of this method in 8 gestures reached 95.8%.Secondly,this paper proposes a method for completing identity authentication through channel state information in the human behavior scenario,which can realize the identity of the actor through continuous human behavior.In the data preprocessing stage,the method performs ambient noise cancellation on the channel state information data through a Butterworth low pass filter.In order to reduce the amount of data calculation,a time interval identification algorithm based on principal component analysis combined with principal component threshold is proposed.By setting the component ratio parameter,the algorithm can perform corresponding proportion component extraction and action segmentation process on the channel state information,and perform the process of simplifying invalid data on the subcarriers of the channel state information according to the time set of the action interval.In order to speed up the certification and improve the accuracy of the authentication results,this paper uses the DenseNet-BC type dense convolutional neural network model to complete the identity authentication process.In the phase of experimental verification and comparative analysis,the overall authentication time of this method is significantly less than that of similar methods,and the average accuracy in the process of multi-person authentication is above 96%.