Machine learning-driven design and performance analysis of microstrip antennas for sub-6GHz/mmWave 5G networks

Abstract

In   the   realm  of   modern   communication   systems,   antennas  are   crucial components, with the microstrip patch antenna being particularly notable for its   low   profile   and   seamless   integration.   Despite   its   widespread   use, designing this antenna involves complex simulations to optimize parameters, requiring significant expertise and consuming considerable time and energy. To  streamline  this  process, machine  learning  (ML) algorithms  are  being utilized.  This  paper  introduces  an  innovativeapproach  that  employs MLtechniques to design a rectangular microstrip patch antenna operating within the  sub-6GHz  frequency  range  (1-6  GHz)  and  the  millimeter  frequency range   (28-40   GHz).   The   antenna   design   maintains   consistent   patch dimensions   positioned   strategically   at   the   center,   with   a   thorough examination of patch length and width to enhance performance. Datasets are meticulously  prepared,  covering  output  parameters  such  as beam  area, directivity,  gain,  and  radiation  efficiency across  the  specified  frequency ranges.   By   employing   various MLalgorithms,   this   study   conducts   a comprehensive   analysis   to   identify   the   most   effective   algorithm   for accurately predicting antenna characteristics. The K-nearest neighbor(KNN)algorithm  achieved  high  accuracy  across  all  parameters: gainat  94.23% under sub-6GHz and 95.93% under millimeter frequency range, directivity at  99.02%  and  98.59%,  radiation  efficiency  at  93.94%  and  94.28%,  and beam  area  at  99.07%  and  98.59%  respectively.  These  results  optimize microstrip  antenna  designs  and  enhance  understanding  of  the  relationship between design parameters and performance outcomes with ML.