Behavioral-level performance modeling of analog and mixed-signal systems using support vector machines

This paper presents a novel behavioral-level analog and mixed-signal (AMS) system performance modeling methodology using support vector machines (SVM). The method relies on linearly graded sub-spaces to model complex multi-dimensional performance spaces. A detailed evaluation of the method has been carried out for the purpose of potential use for AMS synthesis. The method has been applied to a complex non-ideal 2
^nd
order sigma-delta modulator (SDM) and results show good accuracy of performance modeling and numerical efficiency