Electrical and Computer Engineering ETDs

Author

John Guthrie

Publication Date

8-25-2016

Abstract

The exponential growth of malware designed to attack soft real-time embedded systems has necessitated solutions to secure these systems. Hypervisors are a solution, but the overhead imposed by them needs to be quantitatively understood. Experiments were conducted to quantify the overhead hypervisors impose on soft real-time embedded systems. A soft real-time computer vision algorithm was executed, with average and worst-case execution times measured as well as the average power consumption. These experiments were conducted with two hypervisors and a control configuration. The experiments showed that each hypervisor imposed differing amounts of overhead, with one achieving near native performance and the other noticeably impacting the performance of the system.

Keywords

hypervisor, virtualization, soft real-time, Docker, KVM, embedded, performance, power

Sponsors

Air Force Research Laboratory

Document Type

Thesis

Language

English

Degree Name

Computer Engineering

Level of Degree

Masters

Department Name

Electrical and Computer Engineering

First Advisor

Heileman, Gregory

First Committee Member (Chair)

Lamb, Chris

Second Committee Member

Shu, Wennie

Share

COinS