Minimizing time overhead to detect soft errors through preceding variable analysis

Soft error is a significant reliability concern for nanometer technologies. Shrinking feature sizes, lower voltage levels, reduced noise margins, and increased clock frequency improves the performance and lowers the power consumption of integrated circuit. But it causes the integrated circuit more susceptible to soft error that can corrupt data and make systems vulnerable. In computer systems, where the reliability is a great concern, the impact of soft errors may be very catastrophic. This paper proposes a new approach to detect soft errors through variable dependency analysis. The proposed method has lesser time overhead in comparison to existing dominant approach.