Artificial Neural Network and Its Benefit in Modeling and Efficient Yield Analysis of Antennas

Abstract

This paper presents yield-aware antenna sizing and modelling through artificial neural network (ANN) leading to an accurate and efficient synthesis process. In the proposed methodology, the suitable amount of data is provided and then Quicksort algorithm is employed for categorizing the most/least effectiveness of each parameter. Finally, the ANN is trained and used for optimizing the design parameters leading to find optimal output specifications. The performance is named as yield analysis, and it accelerates in predicting the effects of variations in the antenna's dimension and it leads to find the optimal design parameters in a shortest time without any further simulation effort. The developed process results in accurate yield predictions of antenna dimensions with reduced time. To validate the efficiency of the proposed method, three antennas from the recently published literature are considered, and then our presented method is employed for optimizing them and quantify the computational time in modelling. The simulation results demonstrate that using the proposed method, there is approximately 25% speed-up in modelling, sizing, and optimizing antennas.