A State Recognition Approach Based on Distribution Difference for Passive People Counting

People counting in corridor scenarios is attractive for various applications. However, existing WiFi-based passive people counting systems need to use all signal measurements during the entire activity time, which contain a large amount of irrelevant information. These systems prevent the people counting technologies from being applied to many real-time smart applications. To this end, we propose a novel state recognition approach based on distribution difference for the passive people counting system. According to the relative position between the person and the transceiver, we first define the propagation channel into three states, i.e., non-fading, reflection, and shadowing states, and export the signal distributions of the three states as template profiles. Then, we collect the real-time CSI measurements and match them with the templates to achieve state recognition. If the state is identified as shadowing, the corresponding CSI measurements will be used for people counting. Experimental results demonstrate the effectiveness of the system, which can estimate the human count with high accuracy while reducing the scale of the data.