Perbandingan Antena Mikrostrip Patch Rectangular Berbasis Decision Tree dan Berbasis K-Nearest Neighbor pada Frekuensi 3,4 Ghz untuk Aplikasi 5G

Abstract

Communication network technology continues to experience rapid development, especially with the presence of 5G technology which is the newest generation with specifications that are much better than the previous generation. One important component in the implementation of 5G is the antenna, especially the microstrip antenna which is known to have high gain and directivity used in various applications including satellite navigation and telecommunications. In an effort to improve the performance of microstrip antennas for 5G applications at the 3.4 GHz frequency, this research proposes the use of a Decision Tree-based Machine Learning (ML) algorithm to optimize the design of rectangular patch microstrip antennas for 5G applications. Based on a comparison of the parameter results between the CST simulation with the Decision Tree and KNN algorithms obtained for the simulation using the CST Studio suite at a frequency of 3.4 GHz, the return loss value was -31.2561 dB, VSWR 1.05627, bandwidth 1.255 GHz, and gain 2.55 dBi. The prediction results from machine learning with the Decision Tree algorithm obtained a return loss value of -33.3429 dB, VSWR 1.02513, bandwidth 1.397 GHz, and gain 2,895 dBi. Meanwhile, using the KNN algorithm, the Return Loss value was -33.3429 dB, VSWR 1.02513, bandwidth 1.397 GHz, and gain 2,895 dBi.