WiMPP: An Indoor Multi-person Positioning Method Based on Wi-Fi Signal

In the era of Internet of things, convenient and high-precision location service is of great importance for the connection among things. In recent years, the indoor positioning technology based on Wi-Fi devices has developed rapidly, but there is still space for the improvement of accuracy in multi-target positioning. In this paper, a multi person positioning method named WiMPP based on Wi-Fi signal is proposed for the high-precision positioning in indoor scenes. WiMPP first collects the Wi-Fi sensing signals in environment with only one pair of transmit and receive antennas, and then estimates AOA, TOF and other parameters using two-dimensional MUSIC algorithm; Then, the estimated parameters are constructed as a heat map which is then inputted into a two-dimensional convolution neural network for training and classification such that the positioning of targets can be obtained. The experimental results show that WiMPP can achieve high precision positioning accuracy (average error distance is 6 cm, median error distance is 8 cm) under the condition that two persons are in the indoor scene. Compared with other location methods based on Wi-Fi signal, WiMPP not only can position multiple persons, but also improves the location accuracy to a certain extent.