Patch Antenna Design using Artificial Intelligence Methods for 4G/5G Applications

The advent of fifth generation wireless communications (5G), highlights the need for new priorities in research. Especially, the growing mobile networks require abilities for massive connectivity, advanced security, increased network capacity and extremely low-latency. Antennas play a crucial part of this ecosystem, thus, their design is a challenging task. In this work, a Boundary - Seeking Generative Adversarial Network is trained to serve as a surrogate model for the design of a patch antenna with random slots and two operating frequencies at 2.6 GHz and 3.5 GHz, making it suitable for 4G/5G applications. The outcome of the trained model is used as an initial population for further optimization using the biogeography-based optimization algorithm. The final design achieves S-11 = -37 dB and S-11 = -33 dB at 2.6 GHz and 3.5 GHz respectively.