Research On Multi-behavior Perception Method Of Indoor Personnel Based On Channel State Information

With the advancement of information technology and the deployment of basic network communication equipment,the research of indoor personnel behavior perception technology based on Wi-Fi signals has attracted more and more attention.Compared with traditional sensing technology,wireless sensing technology is not affected by line-of-sight and light factors,and the deployment cost is low.This paper studies the multi-behavior sensing technology of indoor people based on Channel State Information(CSI).According to the different characteristics of the specific behavior types and the multiple choices of basic signals,different people behavior recognition algorithms are designed in a targeted manner.And use commercial router equipment to build an experimental platform.Through repeated experiments in different experimental environments,the signal characteristics and algorithm performance are analyzed and evaluated,and compared with other existing research.The main work of this paper is as follows:(1)In order to effectively use the static signal model and improve the feature utilization,a static behavior detection method based on the CSI fingerprint model is proposed,which uses various daily static behaviors of personnel as the entry point for research.Gaussian filtering is used to denoise the original CSI signal,and then Principal Components Analysis(PCA)is used to reduce the data redundancy and extract the feature set,and then use the tree segmentation Least Square Support Vector Machine(LSSVM)conduct offline training and online classification,and verify the effectiveness and accuracy of the algorithm in different experimental environments.(2)In order to improve the accuracy and robustness of dynamic behavior perception,a dynamic behavior detection method based on CSI feature matching is proposed.By establishing the CSI time domain model of sports behavior,the template matching principle is used to achieve behavior classification.First use Discrete Wavelet Transform(DWT)to preprocess CSI time domain data,and perform feature segmentation on the data,and then use Dynamic Time Warping(DTW)combined with spatial distance classification methods to match specific behaviors the work.The characteristics of the time-domain signal are studied,and the parameters that affect the performance of the algorithm are evaluated in the experiment to verify the performance of the algorithm.(3)In order to improve signal utilization and increase the diversity of perceived behavior categories,a behavior detection method based on the combination of CSI amplitude and phase is proposed.Extend the research behavior from single-person behavior to specific two-person interaction behavior,use the processed amplitude information and phase difference information to combine to establish a behavioral perception algorithm with stronger comprehensive performance.The algorithm uses the Random Forest(RF)algorithm to identify different types of behaviors through feature extraction and fusion.Through repeated experiments,the influence of different signal characteristics and classification methods on the system is studied,which effectively verifies the advanced nature of the algorithm in signal selection and behavior perception types.