Abstract: In response to the increasing demand for intelligent human-computer interaction and the problem of poor convenience and high cost of human detection methods based on sensing devices of home appliance, a human detection technology based on Wi-Fi channel state information is proposed, which fully utilizes home appliance Wi-Fi modules to achieve contactless detection. This method utilizes subcarrier amplitude information to filter outliers and extract static environment related features in the offline stage of the algorithm. It trains a fingerprint library for parameter storage, and extracts real-time signal amplitude in the online stage. The results are compared with the training stage threshold to output results with or without people. This method extracts multiple types of feature information in the time domain and subcarrier domain, and can achieve differential detection of human motion, unmanned and human micro motion. It has achieved accuracy in real-world experiments and has certain value for the application of intelligent detection in household appliances.

Key words: Sichuan Hongmei Intelligent Technology Co., Ltd. Mianyang 621011