This letter proposes a new design of millimeter-wave (mm-Wave) array antenna package with beam steering characteristic for the fifth-generation (5G) mobile applications. In order to achieve a broad three-dimensional scanning coverage of the space with high-gain beams, three identical subarrays of patch antennas have been compactly arranged along the edge region of the mobile phone printed circuit board (PCB) to form the antenna package. By switching the feeding to one of the subarrays, the desired direction of coverage can be achieved. The proposed design has >10-dB gain in the upper spherical space, good directivity, and efficiency, which is suitable for 5G mobile communications. In addition, the impact of the user's hand on the antenna performance has been investigated.

A pattern-reconfigurable Yagi–Uda antenna based on liquid metal is presented. The antenna consists of a balun-fed active dipole and a pair of stretchable passive parasitic dipoles, which are implemented by eutectic gallium-indium (EGaIn) alloy embedded in microfluidic channels. The parasitic dipoles are driven at each end by two low-cost, three-dimensional printed media rods. Afterward, spinning the rods at different angles leads to varying degrees of stretching upon the stretchable dipoles. Note the length of passive parasitic dipoles is a vital factor for antenna radiation reconfiguration. The antenna exhibits bidirectional radiation if the parasitic dipoles are equal in length. Otherwise, directional radiation toward the shorter parasitic dipole direction can be obtained. Based on the above-mentioned working principle, a pattern-reconfigurable antenna working in wireless local area network (WLAN) band is fabricated and measured. Apart from the reconfigurable capability, the proposed antenna keeps operating in the WLAN band of 2.4–2.48 GHz during the whole shape deformation.

In this article, we propose an equivalent circuit theory-assisted deep learning approach to accelerate the design of metasurfaces. By combining the filter equivalent circuit theory and a sophisticated deep learning model, designers can achieve efficient metasurface designs. Compared with most existing metasurface generative design methods that rely on arbitrarily generated training dataset (TDS), the proposed method can adaptively produce highly relevant and low-noise training samples under the guidance of filter equivalent circuit theory, resulting in a significantly narrowed target solution space and improved model training efficiency. Furthermore, we select the variational autoencoder (VAE) as a generative model, which can compress the raw training samples into a lower-dimensional latent space where optimization methods, such as genetic algorithm, can be more efficiently executed to find the optimal design than a brute-force search. To verify the effectiveness of the proposed method, we apply it in the creation of three examples of frequency selective surfaces (FSSs), presenting wide-band, dual-band, and band-stop responses. Experimental results show that the proposed method can realize much faster and more stable convergence than deep learning design methods without domain knowledge.

Modern electromagnetic (EM) device design generally relies on extensive iterative optimizations by designers using simulation software (e.g. CST), which is a very time-consuming and tedious process. To relieve human engineers and boost productivity, we proposed a machine learning framework to solve the problem of automated design for EM tasks. The proposed approach combines advanced reinforcement learning (RL) algorithms and deep neural networks (DNNs) in an attempt to simulate the decision-making process of human designers to realize automation learning. Specifically, the RL-based agent can interact with the EM design software without engaging human designers, allowing for automated design. Besides, the data accumulated during EM software simulation in the early design stage are reused as training data to build a DNN surrogate model to replace the time-consuming EM simulation and further accelerate the training of RL to achieve better optimization of EM design. Two types of antenna array decoupling including 1×2 and 1×4 arrays working at 3.5 GHz are used as test vehicles to validate the proposed method. The decoupling metasurfaces designed by the proposed fully automated method based on RL showed satisfactory results comparable to the results achievable by human designers. This indicates that the proposed method can be used to build powerful tools to boost the design efficiency of EM devices.

A low-cost single-layer 2-bit programmable metasurface (PM) is proposed in this article. This PM is processed on a single-layer printed circuit board (PCB) with the microwave structure, ground, and bias lines, enabled by a folded ground. The folded ground design moves a part of the ground from the bottom to the top layer to leave an area for bias lines. Vias are used to connect the ground in top and bottom layers to form a whole. The bias line in the microwave structure area is grounded by two capacitances to isolate microwave current with the bias line area. An asymmetric patch is designed in microwave area as the 2-bit microwave structure. To verify the microwave performance of this PM, functions of generating double beams, electromagnetic (EM) waves with orbital angular momentum (OAM), and a high gain beam are simulated. The far-field pattern of a 16 $\times$ 16 PM antenna in beam scanning is measured. The measured results show that the PM antenna has a beam scanning capability of $\pm$ 60 $^{\circ}$ and a maximum aperture efficiency of 44.6%. Finally, to illustrate the universality of the folded ground design, the varactor-loaded continuous phase adjustable PM and variable polarization 1-bit PM are discussed.