Space and aerial architectures to expand global connectivity

Abstract

Currently, 46.4% of the world's population does not have access to the Internet. Bringing the more than 3.5 billion individuals still unconnected online is the primary goal for multiple international organizations, including the ITU and the UN Broadband Commission. In the last ten years, there has been steady growth in the number of Internet users (around 200 - 300 million per year), but this has been considered insufficient to meet the target of having 60% of the world's population be connected by the end of 2020 (as set in Resolution 200 - Connect 2020 Agenda for Global Telecommunication/ ICT Development). Besides, even more ambitious targets (75% of the world's population connected) have been proposed for 2025. Two important barriers that restrict connectivity are the lack of infrastructure and affordability.

To address these barriers, several novel concepts that involve space-borne and airborne platforms have been proposed to provide connectivity at a lower cost (improve affordability) to a wider reach of people (extend infrastructure). This thesis explores the tradespace of architectures for space and aerial communication network concepts to extend global connectivity. In particular, constellations of geostationary satellites, large constellations of MEO and LEO satellites, and high- and low-altitude aerial platforms are studied. For each of these concepts, I develop end-to-end system models that include the RF propagation, atmospheric channel, power- and mass-sizing, system dynamics, and costs. Different frequency bands are considered, including current state-of-the-art Ku- and Ka-bands and future scenarios with extremely high-frequency bands (V/Q, E, and optical). The potential of each of these concepts is then analyzed from a techno-economic perspective.

Given the large scale of the problem (global connectivity), the different spatial and temporal scales on which each of the concepts operate, plus the large tradespace of potential architectures, evaluating the potential impact requires the development of large simulation models to compute realistic estimates for performance and cost, as well as to identify trade-offs among concepts. However, due to limited computing resources, an exhaustive evaluation of all design configuration is impractical and often not possible; consequently, the resources devoted to concept exploration need to be carefully allotted, balancing exploration and exploitation within the tradespace. To that end, this thesis presents a Bayesian optimization approach tailored for System Architecture problems, to explore tradespaces efficiently when there is a tightly-constrained budget for objective function evaluations.