Rotman Lens Design Using Factorization Machine and Simulated Annealing

This study introduces an Artificial Intelligence (AI)-based optimization method for the Rotman lens, an electromagnetic structure for antenna beamforming. Traditionally, Rotman lenses have been designed using analytical formulas, followed by classical optimization algorithms for fine-tuning. However, the presence of numerous local minima in the design space complicates the optimization process, requiring extensive iterations. To address this challenge, we employ an AI-based optimization technique—factorization machine combined with simulated annealing—known for its effectiveness in navigating large design spaces. Our results show that, out of a $2^{49}$ design space, an optimal design was found in just 602 iterations, achieving performance that surpasses that of a conventional Genetic Algorithm. These findings highlight the potential of AI-driven optimization in improving the performance of electromagnetic systems.