Compact Reconfigurable Multiband Patch Antenna with Machine Learning Approach for Wi-Fi, WLAN, and C-band Application

This work presents a novel compact wideband reconfigurable patch antenna with a machine learning (ML) approach. Reconfigurability is achieved using PIN Diode and operating in ON-ON, ON-OFF, OFF-ON, and OFF-OFF configurations. The resonance frequency characteristics and impedance bandwidth of the ON-ON configuration exhibit resonance at 2.1, 3.7, 6.1, 8.4, and 6.9 GHz, with respective bandwidths of 17.36%, 36%, and 49.25%. In the ON-OFF configuration, maximum resonance is observed at 2.4, 2.7, 3.7, 6.2, 6.9, 7.3, and 8.3 GHz, with corresponding bandwidths of 6.78%, 13.52%, 32.98%, and 51.83% respectively. Similarly, the OFF-ON condition resonance was observed at 2.4, 2.7, 3.7, 6.2, and 8.3 GHz, with bandwidths of 10.57%, 14.88%, 32.76%, and 51.47%. In the OFF-OFF configuration, peaks are observed at 3.1, 3.7, 6.2, 7.1, and 8.1 GHz, with bandwidths of 49.78% and 52.33% achieved, respectively. In all configurations, a gain of 5 dB is achieved. Extreme Gradient Boosting (XG Boost) algorithms provide the highest accuracy in 95.7%, 94.3%,95.6%, and 94.2% in ON-ON, ON-OFF, OFF-ON, and OFF-OFF configurations respectively. The proposed antenna is used for Wi-Fi, WLAN, and C-band applications.