Research into a least multipath based wireless local positioning technique for massive MIMO systems in extreme multipath conditions

点击翻译按钮获取中文摘要

In future wireless communication networks (5G, 6G, etc.), massive MIMO systems increase the achievable data rate. Since the achievable data rate depends on the independence of the spatially recorded information (spatial diversity), large antenna arrays are favorable. In this project, the spatial diversity of Massive MIMO systems is used to enable exact indoor localization even under extreme multipath conditions. While UWB systems attempt to isolate the direct path from the multipath propagation by increasing the bandwidth, the impulse responses of the line of sight (LOS) and the multipath propagation overlap in localization systems with limited bandwidth. Therefore, a LOS identification, which is solely based on the evaluation of a single path, is not possible. Hence, in this project the spatial diversity of Massive MIMO systems is used to separate the common LOS of all receive paths from the superimposed multipath propagation, which decorrelates with increasing antenna distance. For this purpose, a least multipath metric is developed to replace the commonly established least squares metric. Here, the aim is to estimate the multipath propagation, remove it from the signal, and thereby, drastically improve the localization accuracy. For this purpose, the receive channel of each antenna is superimposed by an individual hypothetical multipath channel during the measurement evaluation. While the LOS path at each receiver directly depends on the transmitter’s position, the multipath propagation towards all receivers differ and are sparse, because they are caused by few reflections at the environment. Hence, at the correct transmitter position, the remaining receive signal can be described with only little further multipath propagation, whereas at an incorrectly assumed transmitter position, the LOS path has to be described using additional multipath propagation. According to the compressed sensing theory, the multipath propagation can then be estimated with a small L1 norm at the correct transmitter position. Hence, the least multipath position estimation is performed by a simultaneous minimum search of both the transmitter position and the corresponding multipath propagation. Since the evaluation of the receive channels’ relative phase information is increasing the resolution of the transmitter position, the LOS paths can now be separated from the multipath propagation. Vice versa, the effect of the multipath propagation is simultaneously subtracted from the evaluated phases and hence, the localization accuracy increases. Equally important to the novel localization concept, an equivalent calibration concept will be investigated. For this purpose, again the proposed least multipath metric will be used. The novel calibration and localization concept will be validated by measurements using a unique massive MIMO array, which is already available at the LHFT.