Alexandria Engineering Journal (Aug 2023)
CSRR inspired antenna using artificial neural network for sub 6 GHz 5G applications
Abstract
The article investigates a compact low profile guitar shaped antenna loaded with co-directional complementary split ring resonator(CO-CSRR) based on negative permittivity metamaterial for sub-6 GHz 5G applications.The inclusion of metamaterial on the radiator produces two popular frequency bands n78 5G NR (3.13–4.01) GHz and n79 5G NR (4.98–6.19) GHz with gain of 5.75 dBi and 6.3 dBi respectively. The design is able to achieve high impedance bandwidth of 700 MHz and 1040 MHz at 3.6 GHz and 5.2 GHz with low return loss (S11 < -15 dB). The antenna offers simulated radiation efficiency of 93.9% at 3.6 GHz and 91.8% at 5.2 GHz making it suitable for 5G communication demands.The equivalent circuit of guitar shaped antenna is presented and the values of lumped elements are calculated. The metamaterial characteristic is verified by Nicholson Ross weir method. A neural network model is developed both in Regression predictive modelling and classification predictive modelling for the proposed guitar shaped antenna using feed forward Levenberg-Marquardt algorithm to predict return lossfor various dimensions of antenna. The model predictsthe data with 90.9% accuracy. The prototype of the antenna is designed with a compact size of 28 × 28 × 1.6 mm3, fabricated on FR4 substrate and experimentally verified.
