Wireless communication indoor positioning method in 5G sub-station using deep neural network and location fingerprint algorithm

Abstract

The use of outside positioning systems based on Global Navigation Satellite System (GNSS) for inside location is impractical due to a number of drawbacks which makes the use of machine learning for its solution. Location fingerprinting has historically relied on shallow learning methods. This research proposes novel technique in indoor positioning technique for 5 G substation by deep learning techniques to analyse the indoor positioning using deep neural networks using 5 G NR signal based weighted cross correlation kernel support vector machine. The experimental results shows the comparative analysis between proposed as well as existing technique in terms of SNR, positioning accuracy, positioning error, average relative error, spectrum efficiency which attained SINR of 88 %, positioning accuracy of 96 %, Positioning error of 65 %, average relative error of 56 %, spectral efficiency of 98 % and computational complexity of 51 %. It requires one base station and works with the majority of wireless networks. Its application potential is greater and its computational complexity is less than other indoor localization techniques now in use.