Transient fault-tolerance through algorithms

Abstract

This article describes that single-version enhanced processing logic or algorithms can be very effective in gaining dependable computing through hardware transient fault tolerance (FT) in an application system. Transients often cause soft errors in a processing system resulting in mission failure. Errors in program flow, instruction codes, and application data are often caused by electrical fast transients. However, firmware and software fixes can have an important role in designing an ESD, or EMP-resistant system and are more cost effective than hardware. This technique is useful for detecting and recovering transient hardware faults or random bit errors in memory while an application is in execution. The proposed single-version software fix is a practical, useful, and economic tool for both offline and online memory scrubbing of an application system without using conventional N versions of software (NVS) and hardware redundancy in an application like a frequency measurement system