Wifi-based human body detection method, and smart device

A Wi-Fi based human body detection method, applied to detect a human body sign in an environment by using a change in sub-carrier channel frequency responses in a channel state information of a Wi-Fi connection of a smart device, the method comprising:

• selecting, by the smart device, in a plurality of detection environments, a sub-carrier data source device,
• acquiring the sub-carrier channel frequency responses of the sub-carrier data source device,
• checking and filtering out invalid sub-carrier channel frequency responses, and
• storing the sub-carrier channel frequency responses;
• the method further comprising:
• obtaining, by the smart device, sub-carrier amplitudes from the stored sub-carrier channel frequency responses, and obtaining a plurality of data sets by grouping the sub-carrier amplitudes;
• dividing each of the plurality of data sets into a plurality of sub-datasets based on different parts of the sub-carrier channel frequency responses;
• for each of the sub-datasets, calculating a correlation coefficient between the sub-carrier amplitudes contained in the same sub-dataset to form a sub-dataset correlation coefficient vector, and calculating a Mahalanobis distance between the sub-carrier amplitudes contained in the same sub-dataset to form a sub-dataset distance vector;
• obtaining and taking an average of variances of a plurality of sub-dataset correlation coefficient vectors as a human activity detection value, and
• obtaining and taking an average of variances of a plurality of the sub-datasets distance vectors as a human presence detection value;
• and
• in response to the human activity detection value being greater than a human activity threshold, determining a human activity state;
• in response to the human presence detection value being smaller than a human presence threshold, determining a human presence state, and
• in response to the human presence detection value being greater than or equal to a human presence threshold, determining a human absence state;
• the method further comprising a threshold calibration step for automatically calibrating the human activity threshold and the human presence threshold in different scenarios, the threshold calibration step comprising:
• after entering an automatic calibration mode, recording and continuously calculating, by the smart device, the human activity detection value and the human presence detection value, and storing a maximum value of the human activity detection value and a minimum value of the human presence detection value until a threshold calibration stop command initiated by a cloud server is received;
• and
• after receiving the threshold calibration stop command, setting, by the smart device, a new human activity threshold and a new human presence threshold;
• wherein the new human activity threshold is greater than the maximum value of the human activity detection value, and the new human presence threshold is smaller than the minimum value of the human presence detection value.