Design of an on-chip Hilbert fractal inductor using an improved feed forward neural network for Si RFIC

Abstract

This paper presents an efficient modeling of Hilbert fractal inductors by improved feed forward neuralnetwork trained hybrid particle swarm optimization and gravitational search algorithm (FNNPSOGSA). The proposedmodel computes the effective inductance value (L) and quality factor (Q) of Hilbert fractal inductors with metal tracewidth, effective fractal length, frequency, and oxide thickness as input parameters. In contrast to the traditional feedforward neural network, the proposed FNNPSOGSA has been designed with fewer hidden neurons with much-enhancedlearning and generalization capabilities. As a consequence, the proposed model achieves better speed and is as accurateas electromagnetic simulations. From the simulation results, it is proved that the proposed model is a good alternativefor complex fractal inductor design