A better-than-worst-case robustness measure

In presence of increasing soft error rates due to shrinking feature sizes, design tools are required to analyze fault tolerance and robustness of circuits. Here, we propose a new measure that identifies hot-spots in the design. On the one hand the measure is more accurate than a "worst-case analysis" that ignores excitation probabilities. On the other hand the computation of the new measure is more efficient than a "probabilistic analysis" that considers excitation probabilities at the cost of a higher computational complexity. Both of these extremes can be embedded in the new measure. Experimental results on circuits with protection against soft errors show that the new measure can be calculated effectively.