Masters Theses
A variable node optimization model for Byzantine Fault Tolerant systems
Abstract
"Byzantine Fault Tolerance (BFT) has been a major subject of study over the last two decades with increasing societal dependance on secure, correct, and reliable computer systems and online services. This research presents a model for high-level optimization of emerging systems that rely on these BFT algorithms and use a variable numbers of decision nodes. The model highlights the relationship between the security of a system and its efficiency. Two experiments were performed to determine system performance by varying the number of compromised nodes, decision nodes, and total nodes. They examine the probability that a transaction will be compromised based on these variables using hypergeometric distribution, a subset of combinatorics. It was found that the compromise probability follows predictable patterns, with certain combinations of decision nodes performing better than others. The results show a trichotomous relationship where one in every three decision nodes results in lower security risk than its neighbors. The purpose of this model is to assist system developers in deciding how to best construct their systems to improve security while minimizing resource usage"--Abstract, page iii.
