Applying Hypervisor-Based Fault Tolerance Techniques to Safety-Critical Embedded Systems

Thumbnail Image
Publication date
Defense date
Journal Title
Journal ISSN
Volume Title
Google Scholar
Research Projects
Organizational Units
Journal Issue
This document details the work conducted through the development of this thesis, and it is structured as follows: • Chapter 1, Introduction, has briefly presented the motivation, objectives, and contributions of this thesis. • Chapter 2, Fundamentals, exposes a series of concepts that are necessary to correctly understand the information presented in the rest of the thesis, such as the concepts of virtualization, hypervisors, or software-based fault tolerance. In addition, this chapter includes an exhaustive review and comparison between the different hypervisors used in scientific studies dealing with safety-critical systems, and a brief review of some works that try to improve fault tolerance in the hypervisor itself, an area of research that is outside the scope of this work, but that complements the mechanism presented and could be established as a line of future work. • Chapter 3, Problem Statement and Related Work, explains the main reasons why the concept of Hypervisor-Based Fault Tolerance was born and reviews the main articles and research papers on the subject. This review includes both papers related to safety-critical embedded systems (such as the research carried out in this thesis) and papers related to cloud servers and cluster computing that, although not directly applicable to embedded systems, may raise useful concepts that make our solution more complete or allow us to establish future lines of work. • Chapter 4, Proposed Solution, begins with a brief comparison of the work presented in Chapter 3 to establish the requirements that our solution must meet in order to be as complete and innovative as possible. It then sets out the architecture of the proposed solution and explains in detail the two main elements of the solution: the Voter and the Health Monitoring partition. • Chapter 5, Prototype, explains in detail the prototyping of the proposed solution, including the choice of the hypervisor, the processing board, and the critical functionality to be redundant. With respect to the voter, it includes prototypes for both the software version (the voter is implemented in a virtual machine) and the hardware version (the voter is implemented as IP cores on the FPGA). • Chapter 6, Evaluation, includes the evaluation of the prototype developed in Chapter 5. As a preliminary step and given that there is no evidence in this regard, an exercise is carried out to measure the overhead involved in using the XtratuM hypervisor versus not using it. Subsequently, qualitative tests are carried out to check that Health Monitoring is working as expected and a fault injection campaign is carried out to check the error detection and correction rate of our solution. Finally, a comparison is made between the performance of the hardware and software versions of Voter. • Chapter 7, Conclusions and Future Work, is dedicated to collect the conclusions obtained and the contributions made during the research (in the form of articles in journals, conferences and contributions to projects and proposals in the industry). In addition, it establishes some lines of future work that could complete and extend the research carried out during this doctoral thesis.
Embedded systems, Safety-critical software, Aerospace electronics, Virtual machines, Hypervisors
Bibliographic citation