Locating Pareto Optimal Designs of Antenna Through Parallel Multiobjective Evolutionary Algorithm

Dawei Ding,Xianheng Ding,Haibo Zhang,Shaoxu Ling
DOI: https://doi.org/10.1002/mmce.22871
IF: 1.987
2021-01-01
International Journal of RF and Microwave Computer-Aided Engineering
Abstract:Population-based multiobjective metaheuristics are potential in microwave component designs because all electric properties could be optimized simultaneously to find several tradeoff designs in a single run. However, they often suffer extremely-expensive computational time caused by too many full-time objective function evaluations. In this paper, firstly, an efficient parallel implementation of MOEA/D (multiobjective evolutionary algorithm based on decomposition) is proposed to reduce the overall computational time. Secondly, optimization speed, approximation performance to the entire Pareto front, and algorithm robustness of the proposed parallel MOEA/D are investigated through extensive standard test instances. Then, a complete design procedure based on the proposed optimization algorithm is presented to locate multiple Pareto optimal designs of antenna. Finally, a compact broadband dual-polarized antenna operating at 3.3-5.0 GHz is designed for 5G base station applications. Comparisons are made to verify the great performance of the proposed method.
What problem does this paper attempt to address?