Fault tolerant design of digital systems

摘要

Fault-tolerant design concepts, such as self purging, sift-out modular redundancy, overlapping parity, and cyclic duplication coding, are presented. In most of these methods, fault location can be identified and even possibly be isolated or corrected. Practical case studies involving these methodologies are presented. In addition, methodologies to test the voter circuit for possible malfunction or fault condition are given. Such methods can fault isolate the critical points such as the voters and data routers, and thereby improve the system reliability. With the increasing use of computing systems in such crucial areas as medicine and space, there has come a great need for computers that remain operational in spite of hardware failures. This paper provides a brief overview of several approaches to fault-tolerant computing. Five hardware redundancy techniques are reviewed: static, dynamic, hybrid, self-purging and the reconfiguration scheme.