EFFICIENT ISOLATION MODELLING FOR TWO-PORT MIMO ANTENNA BY GAUSSIAN PROCESS REGRESSION

This article presents a synthesis modelling of isolation in a diamond-shaped fractal electromagnetic band gag (DSFEBG) based two-port multiple input and multiple-output (MIMO) antenna using the Gaussian Process Regression (GPR). A compact two-port MIMO antenna with 0.140[lambda] inter-element spacing is considered for isolation improvement. To predict mutual coupling in two-port MIMO antenna supervised learning-based regression technique of GPR, the model is trained with 50 samples and tested on 125 samples. The model performs result prediction in less than 1 second with RMSE less than 0.0001%. For a better understanding of the isolation between elements of the MIMO antenna, the automatic relevance determination property of GPR is presented. The proposed model does faster computation and is efficient in predicting isolation in DSFEBG based antennas for lower and high-frequency bands of 5G communication system.