Design of Miniaturized Conformal Rasorber Using Machine Learning for Precision Intrusion Mitigation

Abstract

This letter introduces a miniaturized conformal single-layer frequency selective rasorber (FSR) optimized for A-A-T-A-A mode operation for precision intrusion mitigation. The designed FSR has a back-to-back arrangement to achieve absorption and transmission properties. Machine learning (ML) algorithms are used to optimize FSR designs and reduce computational time. The preliminary geometrical parameters from proposed FSR obtained from full-wave electromagnetic solver are used to train the ML model. The K-nearest neighbor (KNN), and multiple linear regression (MLR) algorithm utilized in this research works efficiently with a small mean square error (MSE) varying between 0.01 and 0.09. Furthermore, the MLR offers the highest correlation (${{R}^2})$ value of 0.99. The FSR optimized using the ML technique has a miniaturized size of 0.17λ_o and offers quadruple absorption band viz 6.9 GHz (f₁), 10.9 GHz (f₂), 18.3 GHz (f₃), and 20.8 GHz (f₄) for precision intrusion mitigation. The FSR provides transmission characteristics at 14.6 GHz to 15.7 GHz (as per FCC standards). Due to its conformal design, the proposed A-A-T-A-A FSR serves as an ideal solution for RADAR systems in defense environment.