Machine Learning-Based Matching Medium Design for Implant Communications

Abstract

Matching media are used in various applications to increase the power transmitted into the human body. The selection of the optimum matching medium permittivity is not a straightforward task, as the optimum value maximizing the transmitted power depends on the thickness of the matching medium and the electromagnetic properties of the target tissue. In this paper, a computationally heavy empirical approach and a machine learning-based approach are utilized for the selection of the matching medium. The empirical approach demonstrates that the matching medium can increase the |S 21 | values up to 8 dB, which is validated with measurements. Next, a machine learning-based tool is proposed to predict the optimum matching medium permittivity for any target tissue and any matching medium thickness. A one-dimensional convolutional neural network followed by a multi-layer perceptron is trained with the simulated average Poynting vector magnitudes for muscle and fat as target tissues. The average Poynting vector magnitude and the dipole length for given system parameters are predicted by the trained artificial neural network. The accuracy is calculated by comparison with the results of the empirical analysis and found to be 1% and 12.3% mean absolute percentage error for dipole length and average Poynting vector magnitude, respectively. The proposed tool decreases the time required to milliseconds.