Article Preview
TopIntroduction
Because of their intelligence and automation, evolutionary algorithms (EAs) have extensive use in a variety of fields (Lv et al., 2022; Behera et al., 2023; Rahman et al., 2022; Fatemidokht et al., 2021; Singh et al., 2024; Geng et al., 2022). Their effectiveness has also been validated in some antenna-design scenarios (Srivastava et al., 2022; Dong et al., 2018; Kaur & Sivia, 2022; Nouri et al., 2021). As the influence of semantic research increases (Hu et al., 2022; Nhi et al., 2022; Capuano et al., 2022), it is essential to design high-performance multiple-input and multiple-output (MIMO) antennas for semantic-based mobile phones to ensure quality semantic communication. Because those semantic-based mobiles are always becoming thinner while more components need to fit in them, there is limited space to distribute antenna elements, and the isolation of multiple antennas is greatly affected. While traditional MIMO antenna design depends mainly on the prior experience of the designer, some techniques (Zhang et al., 2010; Wang & Du, 2015; Zou et al., 2019; Wen et al., 2018; Li et al., 2022) are always used. These methods, however, are difficult to apply on semantic-based mobile MIMO antenna design and the design complexity will increase greatly, especially when there are multiple design objectives and limitations. Thus, EAs are considered as an alternative scheme for semantic-based mobile MIMO antenna design.
Those designs are often related to high-dimensional parameters and multi-objective optimization problems (MOPs). With a distinctive multi-objective decomposition operation, multi-objective EA based on decomposition (MOEA/D) (Zhang & Li, 2007) shows good optimization quality in those similar multi-objective algorithms. One drawback of MOEA/D is that its original evolution strategy cannot handle MOPs with complicated Pareto sets well (Li & Zhang, 2009). Thus, an appropriate strategy should replace the original one to improve its performance (Li & Zhang, 2009; Li et al., 2020a) and make MOEA/D more suitable for high-dimensional parameter MIMO antennas in semantic-based mobile applications.
In this paper, the black-hole (BH) algorithm (Hatamlou, 2013) is modified and then integrated into MOEA/D to design semantic-based mobile MIMO antennas. In MOEA/D-BH, a controllable absorption distance is introduced to strengthen its global-optimization ability and a neighborhood learning mechanism can adjust its local searching capability well. Thus, it can provide a required design with accelerated convergence speed for high-dimension parameters. As for the antenna-structure optimization, discrete fragment patches are used as the optimization design parameters (Han et al., 2015). Two MIMO antennas, which are working under 5G/sub-6 GHz bands, are optimized to prove the effectiveness of the proposed scheme. Moreover, an eight-port MIMO antenna is configured based on the above design. All antennas are fabricated and measured to verify their performance and its feasibility in semantic-based MIMO antenna design.