Network Reliability Analysis under the Multivariate Normal Model

Abstract

High Frequency (HF) radios have been used since the early 20th century for long-distance communication. HF communication systems primarily utilize skywave propagation in which the ionosphere is used to reflect radiowaves back to Earth. The performance of HF communication links is directly tied to the ionospheric propagation medium. The ionosphere is a highly variable and irregular environment that creates many challenges in the design of robust HF communication networks. Ionospheric characteristics vary temporally and will be spatially correlated. As a result, link failures within an HF network may also be correlated. In this thesis we develop a novel model for probabilistic link failures that captures the correlation expected in HF communication networks. We focus on two problems related to the design of robust HF networks- 1.) The Network Reliability Problem, which seeks to compute the probability a network is operational in the presence of random link failures, and 2.) the Most Reliable Path Problem, which seeks to identify the most reliable path between two nodes in a network.