Abstract
As a result of the increased importance of engineered electrical systems to modern civilization, it is necessary to design systems that sustain ideal levels of performance despite the potential for internal faults and external attacks. Designing systems that exhibit resilience, also known as fault tolerance, is the primary method by which optimal performance is preserved despite adverse conditions. This paper is a review of a variety of computational and electromechanical fault tolerance techniques from the literature in order to evaluate the state of the art and identify potential areas for improvement. Our findings suggest that the existing literature has only focused on a limited number of resilience challenges, and that no single resilience-enhancing solution, either hardware- or software-based, is capable of addressing all of the major types of possible faults. Further, we classify the papers using the resilience matrix, which combines four resilience phases put forth by the National Academy of Sciences and four Network Centric Warfare domains. We identify the matrix components insufficiently addressed: particularly, we have found no relevant literature on the cognitive and social domains. Even within the parts of the resilience matrix that have received attention in the literature to date, we observe that there is relatively less emphasis placed on the adaptation of the computational and electromechanical systems so that a repeated fault will not incur significant disruption in subsequent occurrences. Therefore, based on this review, we find that while significant and sustained attention has been dedicated to enhance the resilience of engineering electrical systems, substantial work remains to fully address resilience challenges that instill confidence in our ability to engineer resilient systems.