DNN-based Positioning with Optimum Input Parameter in Indoor VLC LOS/NLOS Environment

Accurate indoor positioning has been a major research topic in recent years. Indoor positioning solutions using radio frequency(RF)-based communication technologies have been used such as WiFi and. However, RF-based technology is difficult to provide the accurate positioning due to the rapid change in the received signal from the movement of obstacles and people in an indoor environment. Therefore, in this paper a visible light-based communication is adopted for user positioning. In addition, in view of two aspects of positioning accuracy and positioning processing time, deep neural network(DNN) was applied to perform the precise positioning. For the DNN model, hyperparameter optimization was considered to achieve high accuracy and fast processing time. The trained DNN model is designed to output the user's three-dimensional actual position, and it can be seen from the simulation results that the proposed method achieves more precise positioning than the existing method.