Date of Award

Spring 2020

Project Type

Thesis

Program or Major

Computer Science

Degree Name

Master of Science

First Advisor

Radim Bartos

Second Advisor

Robert Noseworthy

Third Advisor

Marek Petrik

Abstract

Monitoring mechanisms are a critical component of the security and maintenance of high precision timing networks. Any and all guarantees of determinism and correctness are invalidated if a synchronous network malfunctions or is compromised by an attacker. Existing mechanisms allow for a comprehensive view of the distribution of time throughout a network, but they do not scale to large networks. I propose a new method called aggregated reverse time transfer (ARTT), which redefines the existing mechanisms to include a new aggregation scheme that serves the dual purpose of distributed data summarization and anomaly detection for networks of any size. With this thesis I provide a full specification and implementation of the ARTT mechanism, test both the outlier detection and model accuracy on a real timing network, and detail the steps necessary to perform stable-state outlier detection and aggregation on large-scale networks.

Download

Share

COinS