An Intelligent Indoor Localization System in the NLOS Environment Based on Deep Learning and CSI Images

Abstract

With the rapid development of the smart service industry, indoor high-precision localization technology in the non-line-of-sight (NLOS) environment has attracted great interest. Fingerprint-based indoor localization has been widely employed due to its low hardware cost and high localization accuracy. In this paper, we propose an indoor intelligent localization system based on a 2-dimensional deep convolutional neural network (2D DCNN) and image fingerprints extracted from the channel state information (CSI). Simulation results show that the proposed localization system can achieve a high localization accuracy in the NLOS environment.