Radiation Pattern Extrapolation through Generative Adversarial Network
Lida Kouhalvandi, Sercan Aygün, Ladislau Matekovits, M. Hassan Najafi · 2024
Antenna designs play a crucial role in wireless communication systems where high-performance specifications are greatly required. The radiation pattern (RP) specification in both the E-plane and H-plane is important, as it connects the antenna gain along a given direction. This performance is calculated in the entire bandwidth for various frequencies and is time-consuming. To speed up these simulations, a new approach with the help of a generative adversarial network (GAN) is presented, leading to the prediction of the expected radiation pattern outcomes for the determined frequencies. This method is verified for two previously optimized antennas, one operating between 8.8-10.1 GHz and the other working in the 11.3-13.16 G Hz band. The experimental simulation results prove that the mean absolute error is less than 0.35, which yields suitable accuracy for RP predictions.