Design and Optimization of Axial and Hybrid Corrugated Horn Antennas

Abstract

In the rapidly growing world of information technology and high data rate commu- nication, for any wireless system, antenna is an inevitable part. The most widely appreciated antenna for the feed of a re ector antenna is a horn antenna, specifically a corrugated horn antenna. In fact, the corrugated horn antennas are extensively being used for the satellite communication, remote sensing, radio astronomy, etc. In last one decade, the corrugated horn antennas have violated the limitations of conventional horn antennas in terms of radiation pattern and the gain requirements. Conventional methods for their designing and modeling are time-consuming and ex- hausting. As an alternative to these, soft computing techniques are being used for antenna design and optimization. There are many soft computing techniques. In the present work, a hybrid soft computing technique, known as Adaptive Neuro-Fuzzy Inference System (ANFIS) has been used which is having the best features of the Arti cial Neural Network (ANN) and the Fuzzy Inference System (FIS). This thesis work concentrates mainly on comprehensive study of axial and hybrid corrugated horn antennas, their designing and optimization using the ANFIS. Axial corrugated horn has proven to be the shortest profile possible while hybrid corru- gated horn antenna has proven to be the most effective feed. The proposed ANFIS based CAD tool takes peak gain, cross-polar level, side lobe level and edge taper value as inputs and gives the optimum values of the antenna design parameters like aperture radius, pitch, axial distance, number of corrugations, etc. For the training of the designed system, data has been generated using the com- mercially available full wave software High-Frequency Structure Simulator (HFSS). The proposed CAD tool has been designed for the center frequency of 55 GHz. For such high-frequency range, finite element method requires minimum 32 GB RAM storage and the hours of computational time. Compared to that, the trained ANFIS based CAD tool requires approximately 4 GB of RAM and can provide the optimum values of design parameters within a few seconds. Accuracy in data predicted by the proposed CAD tool has been noted having error rate less than 5%.