A design method of programmable transmitarray antenna with independent controls of transmission amplitude and phase is proposed in C-band. The unit cell with cascaded structures mainly consists of four parts, including the receiving antenna, reconfigurable attenuator with p-i-n diodes, reconfigurable phase shifter with varactors, and transmitting antenna. Correspondingly, various manipulations of spatial electromagnetic (EM) fields are achieved by varying the bias voltages of p-i-n diodes and varactors in the transmitarray antenna. The fabricated unit cell is measured in a standard waveguide, and the whole array with $8\times 8$ unit cells is measured with two horns for calibration of the programmable EM features. The experimental results show that the transmission amplitude can range from −16 to −3.6 dB, and the transmission phase achieves 270° coverage independently at the operating frequency of 5.67 GHz under 16 bit programmable control. To further exhibit the capability and functionality of the proposed transmitarray antenna, the waveform engineering of adaptive beamforming and power-allocation beamforming is separately realized in the experiment. The measured results have good agreements with our theoretical calculations, verifying the validity of our design method.

Due to the strong ability of recognizing electromagnetic (EM) environment and adaptively control of EM waves, the intelligent metasurfaces have received great attention recently. However, the intelligent metasurface with frequency recognition for adaptive manipulation of the EM waves has not been studied. Here, we propose a frequency‐recognition intelligent metasurface to precisely control the spatial EM waves under the agile frequencies with the help of a real‐time radio‐frequency sensor and an adaptive feedback control system. An active meta‐atom is presented to reach 2 bit phase coding and 1 bit amplitude coding capacities to control the amplitude and phase independently. Experimental results demonstrate that the metasurface can recognize different frequency of the incoming wave with very high resolution, and can adaptively realize the self‐defined multiple frequency agilities to manipulate the reflected EM waves without any human participation. As example, the intelligent metasurface with frequency recognition can adaptively operate wave absorption at 5.36 GHz, reflection to normal direction at 5.38 GHz, deflection to −30° at 5.40 GHz, random diffusion at 5.42 GHz, and deflection to +33° at 5.44 GHz by detecting the incoming frequency at the resolution of 0.02 GHz.

Metasurfaces have been developed as a promising approach for manipulating electromagnetic waves. Recently, deep learning algorithms have been introduced to design metasurfaces, but the network can only output one solution for each desired input and suffers from nonunique issue. To overcome the aforementioned challenges, a deep neural network model for inverse designs of anisotropic metasurfaces with full phase properties in ultrawideband is proposed. Given the target reflection spectra as inputs, the candidate metasurface patterns are generated through a generative adversarial network (GAN), and the corresponding predictions are simply achieved by the accurate forward neural network model to match the target spectra in the whole band with high fidelity. By training the generator and discriminator in GAN in an alternating order combined with setting a threshold of discriminator loss to trigger the phase prediction, the proposed method is much more efficient and consumes less time in the training process. Numerical simulations and experimental results demonstrate that the reflection phases of the generated meta‐atoms have excellent agreements with the given targets, providing an efficient way in automatically designing metasurfaces. The most important advantage of this approach over the previous schemes is to improve the design speed significantly with very good accuracy.

With the development of artificial intelligence and Internet of Things, hand gesture recognition techniques have attracted great attention owing to their excellent applications in developing human‐machine interaction (HMI). Here, the authors propose a non‐contact hand gesture recognition method based on intelligent metasurface. Owing to the advantage of dynamically controlling the electromagnetic (EM) focusing in the wavefront engineering, a transmissive programmable metasurface is presented to illuminate the forearm with more focusing spots and obtain comprehensive echo data, which can be processed under the machine learning technology to reach the non‐contact gesture recognition with high accuracy. Compared with the traditional passive antennas, unique variations of echo coefficients resulted from near fields perturbed by finger and wrist agonist muscles can be aquired through the programmable metasurface by switching the positions of EM focusing. The authors realize the gesture recognition using support vector machine algorithm based on five individual focusing spots data and all‐five‐spot data. The influences of the focusing spots on the gesture recognition are analyzed through linear discriminant analysis algorithm and Fisher score. Experimental verifications prove that the proposed metasurface‐based non‐contact wireless design can realize the classification of hand gesture recognition with higher accuracy than traditional passive antennas, and give an HMI solution.

In the past decades, metasurfaces have opened up a promising venue for manipulating lights and electromagnetic (EM) waves. In the field of nonlinearity, second‐harmonic generation (SHG) is a research focus due to its diverse applications. There have been many researches for realizing SHG in optical regime using nonlinear characteristics of optical materials, but its efficiency is low. In microwave frequencies, SHGs are basically studied in the guided‐wave systems. Here, high‐efficiency SHGs of spatial waves are presented in the microwave frequency using nonlinear metasurface loaded with active chips at the subwavelength scale. The nonlinear meta‐atom is composed of receiving antenna, transmitting antenna, and active circuit of frequency multiplier, which can realize strongly nonlinear response and link the EM signals from the receiving to transmitting antennas. Correspondingly, to achieve the function of spatial‐wave frequency multiplication, the working frequency of the transmitting antenna in the meta‐atom should be twice as that of the receiving antenna, and hence the active chip is well matched to obtain the signal transforming with high efficiency. Good performance of the spatial‐wave frequency multiplication is demonstrated in the proof‐of‐concept experiments with the best transform efficiency of 85.11% under normal incidence, validating the proposed method.

Hand gesture recognition tasks based on surface electromyography (sEMG) are vital in human-computer interaction, speech detection, robot control, and rehabilitation applications. However, existing models, whether traditional machine learnings (ML) or other state-of-the-arts, are limited in the number of movements. Targeting a large number of gesture classes, more data features such as temporal information should be persisted as much as possible. In the field of sEMG-based recognitions, the recurrent convolutional neural network (RCNN) is an advanced method due to the sequential characteristic of sEMG signals. However, the invariance of the pooling layer damages important temporal information. In the all convolutional neural network (ACNN), because of the feature-mixing convolution operation, a same output can be received from completely different inputs. This paper proposes a concatenate feature fusion (CFF) strategy and a novel concatenate feature fusion recurrent convolutional neural network (CFF-RCNN). In CFF-RCNN, a max-pooling layer and a 2-stride convolutional layer are concatenated together to replace the conventional simple dimensionality reduction layer. The featurewise pooling operation serves as a signal amplitude detector without using any parameter. The feature-mixing convolution operation calculates the contextual information. Complete evaluations are made on both the accuracy and convergence speed of the CFF-RCNN. Experiments are conducted using three sEMG benchmark databases named DB1, DB2 and DB4 from the NinaPro database. With more than 50 gestures, the classification accuracies of the CFF-RCNN are 88.87% on DB1, 99.51% on DB2, and 99.29% on DB4. These accuracies are the highest compared with reported accuracies of machine learnings and other state-of-the-art methods. To achieve accuracies of 86%, 99% and 98% for the RCNN, the training time are 2353.686 s, 816.173 s and 731.771 s, respectively. However, for the CFF-RCNN to reach the same accuracies, it needs only 1727.415 s, 542.245 s and 576.734 s, corresponding to a reduction of 26.61%, 33.56% and 21.19% in training time. We concluded that the CFF-RCNN is an improved method when classifying a large number of hand gestures. The CFF strategy significantly improved model performance with higher accuracy and faster convergence as compared to traditional RCNN.