Single Base Station Massive MIMO Fingerprint Location Based on GAN

Abstract

Massive multiple input multiple output (MIMO) base stations are equipped with antenna arrays with a large number of antennas. Therefore, users can be located by making rational use of the signal received by the antenna arrays. In the indoor nonline of sight (NLOS) scenarios, the traditional location method based on time of arrival, angle of arrival, and received signal strength, etc has poor location accuracy. Therefore, the channel state information (CSI) in massive MIMO systems can be used as location fingerprints to improve the location accuracy. However, in order to improve the accuracy of fingerprint location, a large number of fingerprint data need to be collected, and the sampling cost increases proportionally. Therefore, we propose to use the Generative Adversarial Networks (GAN) to expand the dataset of massive MIMO fingerprint. The experimental results show that the expanded fingerprint dataset can achieve better location accuracy than the real dataset in the scenarios with low SNR, and can achieve similar location accuracy with the real dataset in the scenarios with high SNR.