Machine Learning Approach to Design Fractal Antenna for 5G Applications

The objective of this paper is to design a rectangular fractal based patch antenna using micro strip line feeding techniques adopting machine learning technique for a desired frequency of 3.6 GHz suitable for 5G applications. The machine learning model is trained with 40 sample data set for the design of fractal based antenna for required applications. The gain and frequency response were observed by changing the input parameters. To design a fractal based patch antenna at a frequency of 3.5 GHz to 5.5 GHz band and the data set is developed by changing in width and length of the rectangle patch antenna. The obtained data set is used in the machine learning models and the preferred value is attained for required frequency from the earlier output data by using machine learning, and the error is minimized. In this paper different ML techniques are adopted and proposed a suitable model to predict the dimensions for required applications. Machine learning approach is best suitable to makes decision form the learning data. This paper designs a fractal based micro-strip antenna with three square iterations with the length of L/4, L/8, L/16 and design and analysis is done by using high frequency structure simulator (HFSS). It uses FR4 epoxy material as a substrate which is of low cost and size and it has a high relative permittivity. The cubic SVM regression model gave a best results in compared with different ML techniques. It offers R2 of 0.95, RMSE value 0.1163 and MAE value is 0.2507. The proposed model has achieved reasonable gains and good radiation characteristics and it resonates at 3.6 GHz with the gain of 4.1dB. The proposed antenna is suitable for IoT and 5G applications.