Neuro-genetic design centering of millimeter wave oscillators

In this paper, a new technique for design centering and the yield enhancement of millimeter-wave (MMW) circuits is presented using neural networks for circuit modeling and genetic algorithms for parametric yield optimization. A Monte Carlo based method is developed for the yield estimation utilizing the neural network models. The neuro-genetic methodology has been used for the design centering of 30 GHz cross-coupled VCO as well as a fixed-frequency 60 GHz oscillator. The results display significant yield enhancement i.e. 8% to 91% for 30 GHz VCO and 7% to 70% for the 60 GHz oscillator

[1]  Andrzej J. Strojwas.  Design for Manufacturability and Yield , 1989, 26th ACM/IEEE Design Automation Conference.

[2]  Stephen W. Director,et al.  A new methodology for the design centering of IC fabrication processes , 1991, IEEE Trans. Comput. Aided Des. Integr. Circuits Syst..

[3]  M. Meehan.  Understanding and maximising yield through design centering (microwave circuits) , 1991 .

[4]  Michel Nakhla,et al.  A neural network modeling approach to circuit optimization and statistical design , 1995 .

[5]  Qi-Jun Zhang,et al.  Neural Networks for RF and Microwave Design , 2000 .

[6]  J. Laskar,et al.  Millimeter wave RF front end design using neuro-genetic algorithms , 2005, Proceedings Electronic Components and Technology, 2005. ECTC '05..

[7]  A.L. Sangiovanni-Vincentelli,et al.  A survey of optimization techniques for integrated-circuit design , 1981, Proceedings of the IEEE.

[8]  Ronald L. Iman.  Latin Hypercube Sampling , 2008 .

[9]  K. Krishna,et al.  Optimization Of Parametric Yield: A Tutorial , 1992, 1992 Proceedings of the IEEE Custom Integrated Circuits Conference.