Comparison of Position Estimation Accuracy in Polar and Rectangular Coordinate Systems Using Wi-Fi CSI

Abstract

Prior knowledge utilizing Channel State Information (CSI) and the precise spatial coordinates is imperative for accurately determining an individual’s whereabouts through the utilization of Wi-Fi CSI. In the present study, we employ LiDAR to ascertain the precise spatial coordinates. We conduct a comparative analysis of the estimation precision across various quantities of CSI receivers, as well as different coordinate systems pertaining to the ground-truth position. The quantity of CSI receivers directly impacts the expansion of the sensing area. Conversely, there exist both polar and rectangular coordinate systems, serving the purpose of elucidating the most suitable coordinate system for position estimation. To achieve this, we employ a neural network regression model and gauge the mean absolute error (MAE) as a metric for estimation precision. The assessment outcomes indicate that estimation precision escalates with an increased number of CSI receivers. Moreover, the rectangular coordinate system yields greater accuracy as opposed to the polar coordinate system.