On channel modeling based on semi-deterministic and stochastic approaches for 5G and beyond

• Ano de defesa: 2021
• Autor(a) principal: Pessoa, Alexandre Matos
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: eng
• Palavras-chave em Português: 5G; Modelagem de canal; SCRV; Mobilidade dual; Soma-de-senóides
• Link de acesso: http://www.repositorio.ufc.br/handle/riufc/58916

The evolution of mobile communication systems associated with the network heterogeneity and the explosive growth in the number of connected devices have transformed channel modeling in a challenging task for the current and future generation of wireless systems. This challenge consists in developing new approaches that allow the channel models to provide realistic simulations in a broad range of scenarios and still remain scalable with the number of nodes and coverage area in the network. A realistic channel simulation is associated with the accuracy of the model in describing the propagation effects that affect the electromagnetic waves as well as with the capability of the model on supporting fifth-generation (5G) features for channel modeling, such as double directional arrays, massive multiple-input multiple-output (MIMO), millimetre-wave (mmWave), blockage, and spatial consistency. Additionally, to be accurate, the practical implementation of a channel model needs to deal with limitations in execution time (or computational complexity) and storage (memory requirements). In this context, the focus of this thesis is on the development and calibration of two channel models for 5G networks, including terrestrial and aerial links considering both single mobility (SM) and dual mobility (DM). Different approaches for generating the large scale parameters (LSPs) and small scale parameters (SSPs) are investigated, which yield different trade-offs among accuracy, complexity, and storage. The first model, namely 5G-Remote, is devoted to remote rural areas and takes advantage of the semi-static characteristics of the scattering by considering fixed power delay profile (PDP) and power angular profile (PAP). It is relatively simple to implement and has low complexity. The second model, namely 5G Stochastic Radio channel for dual Mobility (5G-StoRM), is a stochastic channel model (SCM) for 5G networks in general and can be used to perform accurate system and link-level simulations with support for various 5G use case scenarios. A key concept in 5G-StoRM is the use of a low complexity (in terms of computations and storage) sum-of-sinusoids (SoS) method that allows generating spatially consistent random variables (SCRVs) with DM in terrestrial links from a predefined autocorrelation function (ACF). The SoS method is further generalized to the n-dimensional space and proved to be capable of generating any wide-sense stationary (WSS) Gaussian process (GP) in Rn characterized by a positive semi-definite (PSDe) ACF. The generalized SoS method is used to extend 5G-StoRM to also support air-to-ground (A2G) and air-to-air (A2A) links. Due to its simplicity, 5G-Remote is validated from a closed-form expression to its ACF while 5G-StoRM is extensively validated by numerical simulations at 6, 30, 60, and 70 GHz using the results reported by different sources in terrestrial and A2G links, considering indoor, urban, and rural environments. Finally, the memory consumption of 5G-StoRM is proved to be invariant with the number of base stations (BSs) deployed in the scenario, been especially useful to perform simulations in 5G massive wireless networks.