Low Power Error Correction Architecture for Space

Massively redundant fault-tolerant methods, such as triple modular redundancy require enormously more weight and power than the nonfault-tolerant system and solve only one reliability problem. This paper will show that three reliability problems, random bit flips, error bursts and hard failures can be solved for less than 25 percent more power and 50-100 percent more parts using error detection codes and powered-down spares with repeated execution on detection of error.