Design of Microstrip Patch Antenna Using Support Vector Regression

Microstrip patch antennas are popular due to their compact size, lightweight nature, cost-effectiveness, and ease of fabrication, making them suitable for applications like mobile devices and satellite communications where size, performance, and cost are crucial. However, optimizing these antennas for desired performance metrics such as bandwidth, gain, and efficiency presents challenges, as traditional design methods often require time-consuming iterative processes and extensive simulations. To address these challenges, Support Vector Regression is employed to predict the design parameters, specifically the antenna's length and width, to achieve a desired resonant frequency and gain. In this approach, data is gathered using an HFSS simulation tool by varying the patch's width and length. This simulation data serves as the training set for the SVR model, which is then evaluated for its predictive accuracy. The trained SVR model is applied to predict the patch dimensions at a resonant frequency of 2.35 GHz. These predicted dimensions are subsequently used to design the antenna in HFSS, where simulation results show a return loss of -18.3 dB and a gain of 3.2 dB at 2.35 GHz. Comparing the SVR model's predictions with the HFSS simulation outcomes reveals that while the gain values are relatively close, the HFSS simulation yields a significantly lower return loss, indicating better performance than the SVR model predictions.