A Novel Approach to Device-Based Indoor Positioning Using Channel State Information

The development of information technologies and control systems places increasingly high demands on the speed and accuracy of positioning of observation objects. Based on this, the paper proposes a new approach to indoor positioning based on wearable devices. The wearable devices are radio devices based on WiFi technology. Channel matrices, namely the signal amplitude, are used as the analyzed data. The work of the algorithm is divided into two stages: online and offline. At the Offline stage, information about the controlled premises and possible locations of the object are collected. After this, the proposed neural network is trained to solve the positioning problem. At the Online stage, information from routers is translated into coordinates of an object of interest. The use of MIMO and OFDM technologies makes it possible to measure amplitude over 56 subcarriers and 9 pairs of antennas. The total amount of analyzed data per unit of time is 504. It is this fact that made it possible to apply artificial intelligence methods and achieve positioning accuracy of more than 99.8%.