Geometry-based Radio Channel Characterization and Modeling: Parameterization, Implementation and Validation

Sammanfattning: The propagation channel determines the fundamental basis of wireless communications, as well as the actual performance of practical systems. Therefore, having good channel models is a prerequisite for developing the next generation wireless systems. This thesis first investigates one of the main channel model building blocks, namely clusters. To understand the concept of clusters and channel characterization precisely, a measurement based ray launching tool has been implemented (Paper I). Clusters and their physical interpretation are studied by using the implemented ray launching tool (Paper II). Also, this thesis studies the COST 2100 channel model, which is a geometry-based channel model using the concept of clusters. A complete parameter set for the outdoor sub-urban scenario is extracted and validated for the COST 2100 channel model (Paper III). This thesis offers valuable insights on multi-link channel modeling, where it will be widely used in the next generation wireless systems (Paper IV and Paper V). In addition, positioning and localization by using the phase information of multi-path components, which are estimated and tracked from the radio channels, are investigated in this thesis (Paper VI). Clusters are extensively used in geometry-based stochastic channel models, such as the COST 2100 and WINNER II channel models. In order to gain a better understanding of the properties of clusters, thus the characteristics of wireless channels, a measurement based ray launching tool has been implemented for outdoor scenarios in Paper I. With this ray launching tool, we visualize the most likely propagation paths together with the measured channel and a detail floor plan of the measured environment. The measurement based ray launching tool offers valuable insights of the interacting physical scatterers of the propagation paths and provides a good interpretation of propagation paths. It shows significant advantages for further channel analysis and modeling, e.g., multi-link channel modeling. \par The properties of clusters depend on how clusters are identified. Generally speaking, there are two kinds of clusters: parameter based clusters are characterized with the parameters of the associated multi-path components; physical clusters are determined based on the interacting physical scatterers of the multi-path components. It is still an open issue on how the physical clusters behave compared to the parameter based clusters and therefore we analyze this in more detail in Paper II. In addition, based on the concept of physical clusters, we extract modeling parameters for the COST 2100 channel model with sub-urban and urban micro-cell measurements. Further, we validate these parameters with the current COST 2100 channel model MATLAB implementation. The COST 2100 channel model is one of the best candidates for the next generation wireless systems. Researchers have made efforts to extract the parameters in an indoor scenario, but the parameterization of outdoor scenarios is missing. Paper III fills this blank, where, first, cluster parameters and cluster time-variant properties are obtained from the 300~MHz measurements by using a joint clustering and tracking algorithm. Parameterization of the COST 2100 channel model for single-link outdoor MIMO communication at 300~MHz is conducted in Paper III. In addition, validation of the channel model is performed for the considered scenario by comparing simulated and measured delay spreads, spatial correlations, singular value distributions and antenna correlations. Channel modeling for multi-link MIMO systems plays an important role for the developing of the next generation wireless systems. In general, it is essential to capture the correlations between multi-link as well as their correlation statistics. In Paper IV, correlation between large-scale parameters for a macro cell scenario at 2.6 GHz has been analyzed. It has been found that the parameters of different links can be correlated even if the base stations are far away from each other. When both base stations were in the same direction compared to the movement, the large-scale parameters of the different links had a tendency to be positively correlated, but slightly negatively correlated when the base stations were located in different directions compared to the movement of the mobile terminal. Paper IV focuses more on multi-site investigations, and paper V gives valuable insights for multi-user scenarios. In the COST 2100 channel model, common clusters are proposed for multi-link channel modeling. Therefore, shared scatterers among the different links are investigated in paper V, which reflects the physical existence of common clusters. We observe that, as the MS separation distance is increasing, the number of common clusters is decreasing and the cross-correlation between multiple links is decreasing as well. Multi-link MIMO simulations are also performed using the COST 2100 channel model and the parameters of the extracted common clusters are detailed in paper V. It has been demonstrated that the common clusters can represent multi-link properties well with respect to inter-link correlation and sum rate capacity. Positioning has attracted a lot of attention both in the industry and academia during the past decades. In Paper VI, positioning with accuracy down to centimeters has been demonstrated, where the phase information of multi-path components from the measured channels is used. First of all, an extended Kalman filter is implemented to process the channel data, and the phases of a number of MPCs are tracked. The tracked phases are converted into relative distance measures. Position estimates are obtained with a method based on so called structure-of-motion. In Paper VI, circular movements have been successfully tracked with a root-mean-square error around 4 cm when using a bandwidth of 40 MHz. It has been demonstrated that phase based positioning is a promising technique for positioning with accuracy down to centimeters when using a standard cellular bandwidth. In summary, this thesis has made efforts for the implementation of the COST 2100 channel model, including providing model parameters and validating such parameters, investigating multi-link channel properties, and suggesting implementations of the channel model. The thesis also has made contributions to the tools and algorithms that can be used for general channel characterizations, i.e., clustering algorithm, ray launching tool, EKF algorithm. In addition, this thesis work is the first to propose a practical positioning method by utilizing the distance estimated from the phases of the tracked multi-path components and showed a preliminary and promising result.