Research And Implementation Of Motion Recognition System Based On CSI

Motion recognition has a very broad application prospect in the fields of human-computer interaction,security monitoring and motion analysis.The technology has always been a focus of research in both academic and commercial product areas.Motion recognition system based on Wi Fi overcomes the limitations of the system based on vision and wearable sensor in the past,it's the most ideal design for the implementation of this technology.Channel State Information(CSI),as a fine-grained feature of the Wi Fi signal,can reflect the multi-path characteristics of the channel and has a very high perceived accuracy.Therefore,it is of great significance to design and implement CSI-based motion recognition systems.It is currently the main research object of researchers in related fields in the world.Firstly,based on the deep understanding of the application background and implementation principle of the motion recognition system,this paper designs the system framework based on the current research status at home and abroad.With the help of Intel 5300 wireless network card and wireless router,the algorithm used in the four links of signal acquisition,signal preprocessing,feature extraction and motion recognition were designed respectively under a strong multipath interference experimental environment and a weak multipath interference experimental environment.A relatively complete motion recognition system based on CSI amplitude was initially implemented.Secondly,a joint base signal optimization algorithm based on amplitude and phase difference is proposed to solve the single problem for selection of CSI base signal.The algorithm can overcome the problem that phase can't be used because of the error of the phase distribution by performing linear transformation and extension compensation on the phase of CSI,and on this basis,the amplitude of the CSI value of the received signal is combined with the phase difference of the multi antenna at the receiving end,which is used as a joint base signal.In the subsequent feature extraction part,the feature of the joint base signal is extracted as the feature set.This algorithm effectively improves the recog nition resolution of different actions.In addition,an optimization algorithm based on amplitude distribution variance is proposed to solve the problem of high complexity of the system.The algorithm defines the stability of the CSI amplitude distribution for different action by comparing the variance of the amplitude distribution of adjacent packets with the variation of the subcarrier at different time,thus realizing the rapid distinction between the motion and the static behavior,reducing the system operation complexity and effectively shortening the system's recognition time for different movements.The design of the two optimization algorithms improves the recognition accuracy of different actions while ensuring the complexity of this system.Finally,this paper validates the performance of the system before and after optimization.The effectiveness and robustness of the system are verified by performing multiple comparison tests in two experimental scenarios: strong multipath interference and weak multipath interference.The experimental results show that the system designed in this paper can recognize three different movements with higher recognition precision,and the recognition accuracy of the system has been improved about 4.7% after optimization.At the same time,the rate of discrimination between static and motion behavior is improved.In addition,in order to improve the degree of display and practicality of the system,this paper also designed a GUI experiment interface.