Device-Free Crowd Counting Based on the Phase Difference of Channel State Information
Jing Zong, Binke Huang, Liang He, Bo Yeun Yang, Xiaoyan Cheng · 2020
Estimating the number of people accurately in an area of interest plays a critical role in various applications. In this paper, we perform crowd counting from 0 to 7 persons with a closed meeting room scenario. Triple processes (linear transformation, the phase difference extraction and Savitzy-Golay filtering) are performed on the CSI raw phase so that we can successfully extract the features related to the human moving, eliminate environment interferences and phase errors effectively. We propose the support vector machine (SVM) model based on the phase difference expanded matrix of channel state information (CSI) between adjacent receiving antennas. The mean and standard deviation of the phase difference series calculated in a time window at each time spot are added into the initial phase difference matrix. The proposed phase difference expansion matrix can improve the accuracy and robustness in the subsequent SVM classification. This model achieves 98.82% counting accuracy with only a maximum of 1-count-difference.