Intelligent Indoor Positioning Based on Artificial Neural Networks

Abstract

LBS has become an integral part of people's life nowadays. However, the GPS restricted by the shielding effect is unavailable for indoor environments. Therefore, accurately locating an electronic device indoors has become a challenging issue in recent years. This work employs the CSi combined with neural networks to achieve an accurate indoor positioning. The CSi refers to known channel properties of a communication link in wireless communications. This information describes how a signal propagates from the transmitter to the receiver and represents the combined effects of, for example, scattering, fading, and power decay with distance. This work will evaluate several neural networks for the positioning, including the FCNN, CNN, and GCNN. in multi-carrier communication systems, the CSi of adjacent subcarriers has a high correlation, and hence, the CNN is promising to learn and extract the features of this input information corresponding to the location of radio devices. Beyond that, we also investigate an improved CNN, that is, the GCNN, which has more talent to locate in indoor environments than traditional CNNs. Experimental results show that the proposed GCNN can achieve a root-mean-square error (RMSE) of less than 0.08m and 0.3m for 16 and two antennas, respectively. in addition, the computational complexities and required numbers of parameters of compared deep neural networks have been analyzed as well.