A Channel State Information-Based Localization using One-dimensional Convolutional Neural Networks

We propose the novel localization method based on channel state information (CSI). The proposed method estimates range without the channel propagation model estimation by using the one-dimensional convolutional neural networks (1D CNN). The 1D CNN converts CSI measurements directly into the estimated distance, and then the location estimation is performed by the least-squares solution. The simulation results show that the proposed method is accurate and robust to the variation in the channel propagation model.