Hessian-based Antenna Optimization with Automatic Differentiation

Recently [1], there has been a significant interest in the design of antenna systems driven by the emergence of new requirements in areas such as 5G and 6G communications, Internet of Things (IoT), remote sensing, and radar technologies. Antennas are expected to meet specifications such as higher radiation efficiency, agile beam steering, broader bandwidths, and multiband operation. Due to these increasingly challenging requirements, global optimization techniques such as the genetic algorithm (GA), particle swarm optimization (PSO), and simulated annealing (SA) are often employed to find an antenna design that satisfies the specified objectives. In all these methods, the information gathered by individual agents or swarms is shared with other elements, thereby enhancing their search capabilities. Global methods do not require an initial guess, making them ideal for problems lacking a good initial design. However, they suffer from poor computational efficiency, as they require a significant number of function evaluations to find an optimum solution.