Home > Publications > Doctoral dissertation & Master thesis
Domestic Conference
Title Modeling and Analysis of 5G Channels: Geometry-Based Stochastic Channel Model Point of View
Degree Ph.D.
Author Sangjo Yoo
Advisor Kiseon Kim
Graduation Date 2019.02.15 File Link icon
    Date 2018-12-21 21:15


The fifth generation (5G) wireless technology is becoming more available and expected to play a key role in the fourth industrial revolution. 5G is not just a simple advancement of 4G in capacity, but aims to support new services with new business models and use cases, such as Internet-of-things (IoT), public safety, critical mission control, and cellular V2X (C-V2X), to name of few. Such new use cases will be supported based on both sub-6GHz and millimeter wave band with new radio (NR) technologies as envisioned in 3GPP Releases 15 and beyond. In order to support such heterogeneous traffic, users, and services, the new wireless technologies needs to be designed and tested based on the fading channel models, which realistically represent diverse propagation scenarios. Also, comprehensive understanding and analysis of the channel fading characteristic are necessary to improve physical layer design and for efficient fading simulator development.


Among several channel modeling methodologies, the geometry-based stochastic channel modeling approach have been widely used for 5G channel model developments due to its simplicity and adaptability to diverse propagation scenarios. Depending on the scatterer modeling method, geometry-based stochastic channel models (GBSCMs) are classified into two categories: regular-shaped GBSCM (RS-GBSCM) and irregular-shaped GBSCM (IS-GBSCM). The former assumes that the scatterers are distributed on regular geometrical shapes, such as rings, ellipses, and cylinders. The key advantage of the RS-GBSCM is its mathematical tractability, yielding closed-form channel correlation functions (CFs) that are useful for theoretical analysis. Such closed-form solutions are obtained by approximating the total propagation distances (TPDs) traveled by scatterered waves under the local scattering assumption. Hence, the RS-GBSCM has to be used with care, as the TPD approximations and local scattering assumptions become erroneous for 5G channels, where spherical wavefront, small cell-size, and wideband propagation characteristics are typical. On the other hand, the IS-GBSCM does not limit the distribution of scatterers to any regular shapes, but rather distribute the scatterers based on the realistic geometries, similar to the propagation environment of interest (e.g. street canyon, straight road, and intersection). The IS-GBSCM does not rely on the local scattering assumption, but aim to realistically describe the channel characteristics. Hence, the IS-GBSCM is more realistic than the RS-GBSCM, but it is significantly difficult to analyze the channel statistical properties due to the model complexity. Since there is a trade-off between analytical simplicity and accuracy for the two model types, both model types should be properly applied based on the specifics of the problem.


The first major part of this dissertation is devoted to the thorough analysis on the accuracies and validity ranges of conventional RS-GBSCMs, relying on the TPD approximations and the local scattering assumption. In the literature, there are a number of publications on the RS-GBSCMs and their statistical analysis results. Since it has been unknown that how is the accuracies and validity ranges of such approximation-based models, the practicality of the RS-GBSCMs have been questionable. To fill this gap, we develop a reference RS-GBSCM for space-time-frequency (STF) correlated channels and analyze its accuracy and validity range. We show that the conventional RS-GBSCMs become inaccurate for urban pico / micro / macrocells, vehicular-to-vehicular (V2V), and wideband channels, where large beamwidths and relative propagation delays are typical. Also, the TPD approximations lead to errornous spatial CFs for the massive MIMO arrays. In order to improve the model accuracy, we propose new TPDs using Taylor-series approximations. Then, we derive a closed-form solution for the STF-CF using Jacobi-Anger expansion and show its superior accuracy and validity range. The applicability of the new solutions to conventional RS-GBSCMs are also tested and verified based on measured data.


In the second part of this dissertation, we present theoretical investigations on the impact of roadside scatterers (RSSs) on the DPSD characteristics of V2V channels for a straight road, which is an important 5G C-V2X use case. So far, several measurement results have been shown that RSSs, such as houses, buildings, trees, and sound blockers distributed along with roadsides, produce the most significant Doppler spread and constitute a background spectra of measured DPSDs. Such a significant Doppler spread by RSSs leads to high frame error rate at the receiver by degrading channel estimation accuracy. Despite of its importance, only handful theoretical investigations have been conducted due to the complexities of the channel models and the corresponding DPSD solutions. In this thesis, we propose an indirect method for the DPSD derivation of a generic two-dimensional (2D) RSS model, which is an IS-GBSCM. Compared to the conventional methods, leading to impractical multiple integral solutions, our method yields a single integral-form, more useful for analytic studies, model validation/parameter estimation, and fading simulator design. Using the new DPSD solution, we investigate the impact of different RSS layouts on the DPSD characteristics and also provide several new insights. In addition, we present comparisons of the proposed DPSD with the DPSDs measured in highway and urban canyon V2V environments. Our research results demonstrate not only the validity of the generic 2D RSS model, but also the significant contribution of RSSs to V2V channels.


In the last part of our dissertation, we propose much simpler DPSD solutions of the V2V channels in the presence of RSSs. Although the proposed DPSD of the generic 2D RSS model is simpler than conventional solutions, it is in an integral-form, which is too complicated to extract any insight on the impact of RSSs on the DPSD shapes. In order to derive a simpler DPSD solution, we use a geometrical one-dimensional RSS model, which considers two line segments on both roadsides to represent RSSs. In order to obtain the exact DPSD, we use the indirect method mentioned above. Then, angular approximations are additionally used to obtain two closed-form DPSD solutions. We show that the new closed-form DPSDs are simple rational-forms and are scaled, shifted, and truncated sums of the cubic Jakes’ DPSD. The new DPSDs are also close to the DPSDs measured in highway V2V channels. Hence, our new DPSD solutions can be good alternatives to the conventional closed-form solutions, such as classical Jakes and double Jakes DPSDs, which have been widely used for V2V system analysis and fading simulator design. 

광주과학기술원 한·러 MT-IT 융합기술연구센터 광주과학기술원정보통신공학부