Device-Free Target Following with Deep Spatial and Temporal Structures of CSI

Abstract A novel device-free target tracking and following method is proposed based on both the received signal strength (RSS) channel state information (CSI) of WiFi signals. Different from typical method, we consider a scenario where under followed by device that transmits reference signals for location itself. To meet goal, deep spatial-temporal neural network model designed to learn exploit multi-resolution spatial temporal features RSSI CSI tracking. By experimental results our testbed, show average positioning accuracy can reach 0.773 meters, which has $$64\%$$ 64 % improvement over 2.164 meters same condition.