End-to-end statistical process/device/circuit/system design

Abstract

In this paper we present a methodology for the end-to-end statistical process/device/circuit/system analysis and optimization. We use standard software at the every design stage when ordinary design procedure is performed. But approximated dependencies, which were obtained through the use of response surface methodology, are used to conduct statistical analysis in Monte-Carlo loop for investigation of influence of process parameters deviation on output process/device/circuit/system performances. A rudimentary simple example of the cell inverter design, formed on the basis of the MOS-transistor, illustrates the efficiency of the methodology.