A partitioning approach to yield estimation for large circuits and systems

This paper describes an approach to yield estimation which has been developed specifically for large circuits and systems. The approach is based upon the partitioning of a given system into subsystems and is developed in terms of integral transforms. Efficient computational schemes are described which are developed from the properties of the transforms and some elementary ideas from mathematical statistics. The paper shows that the method is flexible, very efficient, and produces accurate results. For large systems involving many identical subsystems, the method offers orders of magnitude reductions in computation times; for large systems involving distinct (nonidentical) subsystems, considerable computational saving is also achieved. The method allows account to be taken of correlations among subsystems, and can be applied to a large class of nonlinear circuits and systems.