Machine Learning-Assisted Modeling in Antenna Array Design

With the fast development in modern wireless communication, the complexity of the modern antenna array design increases rapidly with the increasing element number. Conventional antenna array design methodologies introduce knowledge-guided perceptions and assumptions to achieve highly effective antenna array design, without considering mutual coupling and platform effects. Such effects could worsen beam quality, especially for arrays with small or medium array size. Data-driven methodologies including machine learning-assisted modeling method provide powerful tools to consider those affections within the design procedure. In this article, the machine learning methods are introduced in several scenarios in antenna array designs, including the modeling of active element patterns under mutual coupling, the decoupling performance of defected ground and the base element performance in series-fed antenna array. It can be seen that the introduction of machine learning-assisted modeling could be of great help in antenna array designs.