Indoor 5G Positioning using Multipath Measurements

Abstract: Positioning with high precision and reliability is considered as an important feature of new wireless radio networks such as 5G. In areas where satellite positioning is not available or is not reliable enough, 5G can work as an alternative. An example is inside factories where autonomous vehicles might need to be positioned in complex environments. This work aims to investigate if multipath propagation of radio signals can be exploited to improve indoor positioning. A 5G simulator that simulates the propagation of a reference signal in a factory environment is used. Distances corresponding to several paths between the user equipment (UE) and the transmission/reception point (TRP) can be estimated given the received reference signal. These distance estimates are used together with a partially known map of the environment to develop and evaluate the algorithms in this thesis. The developed multipath-assisted algorithms are based on two different target tracking methods, an extended Kalman filter (EKF) and a particle filter (PF). Both alternatives use a data association algorithm to determine how measurements should be paired with propagation paths. Both filters that exploit multipath propagation are shown to greatly improve positioning accuracy compared to a line-of-sight (LOS) based alternative. The multipath-assisted algorithms can achieve an accuracy below 0.9 m in 90 % of all cases in a complex environment, which is more than tenfold better than the LOS based alternative considered here. The PF also shows an ability to track a UE in a complex environment using very few TRPs, while the EKF and LOS based methods do not succeed in this case.