Biomedical Antenna Design Optimization Using Multi-Objective Inverse Neural Networks

A new approach based on an Inverse Artificial Neural Network (IANN) for Multi-Objective Antenna design is presented in this paper. The network sets geometrical variables as the output and uses three antenna performances as inputs. The proposed ANN model is structured into two distinct parts: In the first part, three autonomous branches establish the correlation among the S parameters, gain, specific absorption rate (SAR), and geometric variables of the antenna. The outputs of these branches are used as input in the second part to derive a distinctive solution for these geometric variables. The proposed antenna dimensions are 20 × 24 × 1.58 mm³, and an ultra-wideband of 4.1 GHz to 8.7 GHz is achieved in free space and on human tissue, which coincides with the 5.8 GHz ISM band. Body temperature and specific absorption rate are simulated using the suggested rectangular patch antenna. The resulting optimized antenna holds promising potential for biomedical applications.