Computationally Efficient mm-wave Scattering Models
The use of millimeter-wave (mm-wave) frequency bands for fifth-generation (5G) and beyond cellular mobile communications has led to intense interest from academia and industry over these spectrum resources. This underutilized frequency band opens a considerable number of challenges in terms of propagation effects, antenna systems, circuit design, etc. In terms of propagation, the provision of reliable and accurate mm-wave channel models is key for supporting the deployment of ultra-high capacity 5G networks. With this vision, extensive mm-wave radio-channel measurement campaigns have been carried out worldwide in various propagation environments and at multiple frequencies during the last decade. These studies emphasize the importance of scattering at mm-waves.
Blockage and backscattering by objects are two relevant scattering phenomena at mm-wave. Hence, modeling of these phenomena needs to be considered for obtaining meaningful system- and link-level predictions. In terms of blockage modeling, several computationally efficient alternatives for calculating blockage from rectangular screens are available in the literature. However, as regards to backscattering models, existing solutions following a similar approach are only accurate in the far-field. That is, radar cross section (RCS) based approaches.
First, two three-dimensional (3D) computationally efficient methods for calculating backscattering from smooth rectangular surfaces based on the Fresnel integrals and the error function are proposed in this thesis. In addition, applying the same methodology, existing blockage models are modified to capture backscattered fields. Second, after identifying the similarity between the blockage and backscattering phenomena, a combined model for calculating both blocked and backscattered fields from rectangular smooth surfaces is proposed. These two first contributions of the thesis are restricted to single-surface contributions. Third, a multiple-scattering model for calculating backscattered fields due to consecutive interactions with several rectangular surfaces is proposed. In addition, the multiple-scattering model is later extended to capture single-blockage between backscattering events.
Fourth, and last, the models are extended to support key aspects in 5G-and-beyond mobile communications such us multiple-input multiple-output (MIMO) antenna systems, polarization effects, and non-rectangular shapes.
All the above mentioned models are exhaustively validated both against simulations and measurements through a significant number of test-cases and numerical convergence to physical optics (PO) models is shown.
The proposed models can be used as a complement to more detailed electromagnetic (EM) models, offering a first-order approximation with minimal conceptual and computational complexity, which makes them easy to implement and allows for their use in exploratory studies, for example in conjunction with optimization tools. These models would be well suited for simulations of indoor or outdoor scenarios where fast, efficient calculations are of the essence, for example, in system- and link-level simulations of wireless systems, particularly when Monte Carlo simulations are applied. Such problems typically involve very many, hundreds or more, large objects (in terms of wavelengths), which makes rigorous EM modeling impossible or prohibitively time-consuming.
Also, for ray-tracing modeling, where the number of potential paths can grow exponentially for higher-order diffuse and diffraction-like ray interactions, means for early pruning is of the essence. Hence, integrated approximate, but computationally efficient, path loss models could then be used to prune non-contributing candidate paths on-the-fly. The models can be applied to surfaces which either represent an actual scenario or are generated based on distributions for position, size, orientation, etc.