Title :
Performance modeling of analog circuits via neural networks: the design process view
Author :
Sobecks, Brian ; Nevin, Joseph ; Helmicki, Arthur
Author_Institution :
Motorola Inc., Schaumburg, IL, USA
Abstract :
This paper presents a novel framework for modeling the performance of analog circuits and a methodology for constructing the corresponding models. The methodology uses simulator data to iteratively train neural networks in order to produce very accurate multivariate nonlinear models, which can be instantaneously evaluated. Results are presented for circuits at various levels of abstraction
Keywords :
analogue integrated circuits; circuit simulation; hardware description languages; integrated circuit design; integrated circuit modelling; neural nets; abstraction levels; analog circuits; design process view; iterative training; multivariate nonlinear models; neural networks; performance modeling; simulator data; Analog circuits; Circuit simulation; Circuit synthesis; Circuit topology; Measurement; Neural networks; Predictive models; Process design; Signal design; Vehicles;
Conference_Titel :
Circuits and Systems, 1998. Proceedings. 1998 Midwest Symposium on
Conference_Location :
Notre Dame, IN
Print_ISBN :
0-8186-8914-5
DOI :
10.1109/MWSCAS.1998.759428
