Machine learning enabled dual to wideband frequency agile $$\:{\varvec{A}\varvec{l}}_{2}{\varvec{O}}_{3}\:$$ceramic-based dielectric MIMO antenna for 5G new radio applications

This article presents a dual-band to wideband Frequency Agile (FA) rectangular dielectric resonator (DR) based hybrid MIMO antenna for 5G New Radio (NR) application with connected ground. The DR is made of Al₂O₃ (ε_r = 9.8) ceramic material. The FA is achieved through the PIN Diode switches. When the PIN Diode is in an "ON" state, it provides dual bands due to the excitation of TE₁₁₁ mode. When the PIN Diode is in an "OFF" state, it provides wideband characteristics due to the excitation of TE₁₁₁ and TE₂₁₁ modes in the rectangular DR. The isolation and gain are achieved by 20 dB and 4.3 dBi, respectively. The maximum tuning range is 49.36. The MIMO performance characteristics are achieved within the allowable range. A good agreement is achieved between the simulated and measured results. The suggested MIMO antenna is optimized through the various ML algorithms in which Random Forest (RF) ML algorithms achieved the highest accuracy more than 99% compared to other ML algorithms for S-parameters prediction. Hence, it is suitable for 5G NR applications.