MLLoc: Machine Learning Location System Based on RAIM in Mobile Networks

Location-based service (LBS) has been playing an essential role in various sectors of society. As the most important solution in LBS, the global navigation Satellite system (GNSS) has been limited in availability in areas such as city and canyon. With the development of wireless communication technology, widely distributed mobile networks can provide numerous quality line-of-sight (LOS) path signals in GNSS-denied environments. Therefore, this article develops a machine learning location (MLLoc) system based on receiver autonomous integrity monitoring (RAIM) to fuse mobile network signals with GNSS, in which a machine learning (ML) method is used to obtain stable time of arrival (TOA) estimation from the downlink broadcast signal of mobile networks. The field tests carried out in long term evolution (LTE) networks have verified the performance of the method. Specifically, the range accuracy of the TOA estimation based on ML and the positioning performance of the MLLoc system were evaluated through field tests. Our results verified that the developed MLLoc system is highly available in GNSS-denied environments and achieves meter-level positioning accuracy.