Application of Genetic Algorithm and Deep Neural Network for UWB Antennas

We present a novel approach for designing, and modeling an ultra-wideband (UWB) printed antenna employing genetic algorithm (GA) and deep neural network (DNN), respectively. The antenna configuration features defected radiating and partial ground planes optimized by GA. The optimum antenna design shows an operating bandwidth from 3.11 to 10.8 GHz with fractional bandwidth of 110%. The peak gain of the antenna varies between 1.5 and 3 dBi with stable radiation characteristics across the operating bandwidth. DNN is implemented for inverse modeling of the optimized antenna to efficiently compute the geometrical parameters based on desired electrical performance parameters. The optimum performance of the proposed UWB antenna and the DNN model validates the effectiveness of the integrated GA-DNN based antenna optimization and modeling methodology.