Statistical methodologies for integrated circuits design

Abstract

The continuous scaling of physical dimensions has strongly increased circuit performance variability and the traditional corner-case methodology is becoming unreliable. As a consequence, there is an urgent need for new and more accurate statistical models. In this scenario, the purpose of this paper is twofold: 1) to give the reader the basic concepts of statistical modeling, and 2) to discuss a viable statistical approach that could be adopted into a traditional IC design flow for the next technology generations.