부산대학교 도서관

An innovative massive multi-objective design procedure is proposed for the synthesis of next-generation antennas for 5G base stations. The 5G antenna design problem is formulated by jointly considering several contrasting requirements in terms of bandwidth, directivity, half-power beamwidth, polarization, and neighbor element isolation. Towards this end, a finite-array model is developed which enables the simulation of a set of adjacent elements during the design process. Thanks to such an approach, the obtained design can be directly included in 5G antenna arrays without further re-optimization to compensate for mutual coupling effects. The resulting massive multi-objective problem is recast as a multi-objective one by suitably clustering the cost function terms according to their physical features, and ad-hoc global search techniques are customized and applied in order to address with the obtained highly non-linear optimization problem. Preliminary numerical results concerning a Pareto-optimal tradeoff solution are presented to validate the proposed approach.