Stochastic Geometry-Based Modeling and Performance Analysis for Terahertz Heterogeneous Networks

Abstract

With the rise of numerous bandwidth-intensive applications and a significant proliferation of wireless smart devices, the Terahertz (THz) band (0.1–10 THz) is increasingly recognized as a vital enabler for the next generation wireless networks. Although THz communication holds significant potential advantages, its adoption has remained stagnant until recently, primarily due to the challenges posed by limited coverage and poor penetrability of THz links. In order to overcome the aforementioned obstacles and fully leverage the potential of the THz band, it is of paramount importance to undertake a comprehensive investigation of the THz heterogeneous network (HetNet) comprised of both THz base stations (TBS) as well as the already extensively deployed millimeter-wave (mmWave) base stations (MBS).

For the sake of tractability, it is a common practice to assume spatial independence among all tiers in a HetNet. However, considering the strong directivity, limited coverage, and low penetrability of TBS, it is more practical to densely deploy TBS in a clustered manner for indoor communication scenarios. Therefore, in the thesis, we focus on a HetNet consisting of THz and mmWave nodes with spatial dependence deployed in a finite area. Specifically, the mmWave nodes are spatially distributed following a Poisson Point Process (PPP), while the THz nodes are clustered around the mmWave nodes, forming a Poisson Cluster Process (PCP) with the mmWave tier serving as the parent process. This approach allows us to introduce inter-tier spatial dependence between the THz and mmWave tier, providing a more realistic representation of the application scenarios for THz communication.

In this thesis, our primary focus is on elaborating a stochastic geometry framework for the spatially dependent THz-mmWave HetNet. By considering various features in THz communication such as network topology, THz channel model characteristics, antenna pattern, and blockage model, we derive the closed-form expressions of the contact distance distribution, association probability, and the Laplace transform of HetNet interference. The proposed framework offers valuable insights into how the spatial dependence between the THz and mmWave tiers, as well as the clustering settings, impact the overall network performance. By quantitatively analyzing these aspects, we are able to reveal the impact of inter-tier spatial dependence on network performance. It is revealed that this spatial dependence introduces flexibility in node deployment by adjusting the scattering variance and the number of nodes per cluster. Furthermore, through numerical simulations, we demonstrate the significant impact of bias ratio, density, and blockage on the coverage probability, indicating the importance of selecting the optimal combination of these parameters.

In order to delve deeper into the sensing performance of THz HetNets, we present a stochastic reflection model to evaluate its performance at the link level. Specifically, we adopt the Boolean Line Model (BLM) to represent ambient obstacles. Taking into account the first-order specular reflection, we statistically analyze the distribution of reflection link distances, which is further extended to derive the distribution of received signal strength (RSS). The reflection model is validated by comparing simulations and the experimental data collected from real-world scenarios. By leveraging the proposed reflection model, a fine-grained analysis of sensing and localization performance of THz HetNet is conducted by evaluating the successful sensing probability which is determined by the number of multipaths exhibiting sufficient signal-to-noise ratios for environmental sensing. The stochastic geometry model proposed in this thesis provides a comprehensive and in-depth analysis of THz HetNets, encompassing both communication and sensing aspects.