Title :
A method for statistical design centering
Author :
Wojciechowski, Jacek ; Vlach, Jiri
Author_Institution :
Inst. of Electron. Fundamentals, Warsaw Univ. of Technol., Poland
Abstract :
A method for geometrical centering is extended by including statistical noncorrelated properties of the design parameters. A metric is introduced that replaces evaluation of a multidimensional integral by evaluation of a number of simple integrals. Yield maximization is formulated and solved as a geometrical problem in n-dimensional metric space; yield derivatives are not required. Optimization techniques are used in solving the problem, simultaneously also providing means for the constraint region approximation and yield estimation. The optimization procedure requires that parameter probability density functions be sufficiently regular. An approximation meeting this condition is proposed
Keywords :
constraint handling; network synthesis; probability; statistical analysis; constraint region approximation; geometrical centering; n-dimensional metric space; parameter probability density functions; statistical design centering; statistical noncorrelated properties; yield derivatives; Circuit synthesis; Constraint optimization; Design methodology; Extraterrestrial measurements; Gravity; Multidimensional systems; Optimization methods; Probability; Sampling methods; Statistical distributions;
Conference_Titel :
Circuits and Systems, 1992. ISCAS '92. Proceedings., 1992 IEEE International Symposium on
Conference_Location :
San Diego, CA
Print_ISBN :
0-7803-0593-0
DOI :
10.1109/ISCAS.1992.230021
