Abstract
This study presents a comprehensive comparative performance analysis of the Artificial Bee Colony (ABC) and the Surrogate-Assisted Differential Evolution (SADEA) algorithms for the design optimization of an 8-element linear dipole array operating at the 28 GHz 5G mmWave band. The optimization objective focuses on maximizing realized gain while maintaining a stringent impedance matching constraint of S11 < -15 dB. To ensure physical accuracy, a realistic loss model incorporating copper conductor losses is employed. Both methods are subjected to a restricted computational budget of 150 function evaluations to assess their convergence efficiency under limited resources. The results are validated against a classical design baseline derived from analytical formulations. Findings indicate that while both algorithms successfully satisfy the robust matching criteria, the ABC demonstrates a marginal superiority in achieving higher realized gain and enhanced impedance matching performance within this specific constrained mmWave design space.