A Single-Version Algorithmic Approach to Fault Tolerant Computing Using Static Redundancy

Abstract

This paper describes a single-version algorithmic approach to design in fault tolerant computing in various computing systems by using static redundancy in order to mask transient bit errors in processor-memory and registers. This low-cost single-version scheme relies on a time redundancy approach. This software scheme does not intend to tolerate software design bugs. Instead of using multiple and independent versions of an application program, this single-version approach uses multiple copies of an application program. This low-cost approach is useful to tolerate various malicious code modifications and transient-faults during the run time of a computing application system without incurring any additional cost for extra hardware and extra software versions as an N-version programming scheme (NVP) or a Recovery block scheme (RBS). This proposed model is a practical and usable one that demands an affordable redundancy in time and space. The proposed scheme is capable of tolerating various operational faults that might occur during the execution time of an application.