Modeling of drain current characteristics of SOI MOSFETs using neural networks

This paper presents neural network approaches for modeling the I-V characteristics of Silion-on-Insulator (SOI) MOSFETs. The modeling approach is fast and accurate making it suitable for circuit simulators. In our modeling, two different neural network architectures, MultiLayer Perceptron (MLP) and Generalized Radial Basis Function (GRBF), are used and compared. To increase the training efficiency of the neural network, a modular method has been utilized. The drain current characteristic obtained by the neural network model is compared to simulation data. The comparison shows excellent agreements with relative errors of less than 1.1% over a wide range of drain and gate voltages and channel lengths.