Double Deep Q-Learning Approach for Tuning Microwave Cavity Filters using Locally Linear Embedding Technique

Abstract

From past two decades, Microwave Cavity Filter is the buzzword in communication arena. They are highly valued owing to their high-Q factor and robustness. Once assembled, the filter needs to be properly tuned to compensate for various design imperfections and manufacturing tolerances, which till date is carried out manually. To overcome these limitations a Computer-Aided Tuning (CAT) method has been proposed using Double Deep Q-Learning (DDQN) Algorithm to fine tune Microwave Cavity Filters. Owing to copious data obtained using a commercial filter, the researchers have used Locally-Linear Embedding (LLE) approach for dimension reduction. The algorithm has been tested via simulation of a 9 <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">th</sup> order filter. Furthermore, four screws were used for training and testing the algorithm. The proposed algorithm could tune the considered filter in 23 steps only. The quick tuning, received towards the end of the research, proves that the algorithm is effective in facilitating the tuning process.