Asian Journal of Current Research

The rapid development of wireless communication systems has created an increasing demand for compact, broadband, high-performance, and application-specific antennas. Conventional antenna design approaches generally rely on analytical formulations, empirical relationships, parameter sweeps, and repeated full-wave electromagnetic (EM) simulations, which become computationally expensive as the number of design variables increases. To address these challenges, metaheuristic optimisation techniques such as Genetic Algorithm (GA) and Particle Swarm Optimisation (PSO) have been widely investigated for exploring complex antenna design spaces. More recently, machine learning (ML) and deep learning (DL) approaches have been introduced for antenna performance prediction, inverse design, and surrogate-assisted optimisation. These approaches can reduce the computational burden associated with repeated EM simulations; however, their effectiveness often depends on sufficiently large and representative simulation datasets. Hybrid ML–metaheuristic approaches further improve the optimisation process by combining predictive models with intelligent search algorithms, while Active Learning enables the selective acquisition of informative simulation samples. This review presents the evolution of computational antenna design from conventional and metaheuristic approaches to ML-based, hybrid, and Active Learning-driven methodologies. The reviewed studies are critically examined with respect to their design strategies, optimisation techniques, simulation requirements, data dependence, and computational efficiency. Particular emphasis is placed on the role of surrogate modelling and adaptive sample selection in reducing expensive full-wave simulations. The review identifies the limitations of existing approaches and highlights the need for autonomous, data-efficient, and simulation-efficient frameworks for high-dimensional microstrip antenna optimisation.