19 July 2021 in IEEE Transactions on Antennas and Propagation
A cognitive antenna array, which is designed by using deep reinforcement learning (DRL) is proposed in this article to adapt to the complex electromagnetic environment. Specifically, the phased array antenna is utilized as the manipulatable component to achieve the characteristic of beam steering with the help of the DRL algorithm. We begin by establishing a DRL-based framework, which is comprised of a microprogrammed control unit, power divider, digital phase shifter, and the patch antenna array. In the DRL algorithm, the desired beam steering is obtained through trial-and-error interactions with the environment, which is required to observe predefined rewards based on the current state and action. Next, the system is trained to perform beam steering, and a set of hyperparameters of the deep neural network are obtained and stored for practical usage. A good agreement is achieved between the simulated and measured radiation patterns of the planar phased array antenna, which validates the DRL-based phase distribution regulation algorithm. Finally, the algorithm is implemented in the design process of a conformal phased array antenna, and it is shown that the measured radiation performance of the array is satisfactory for different beam scan angles.
11 January 2022 in IEEE Transactions on Antennas and Propagation
This article introduces an amplitude and phase independently adjustable transmitarray aperture with a thin profile and low cost to fabrication characteristics, and a transmitarray antenna (TA) featuring the high gain and low sidelobe level (SLL) is realized based on it. The TA is made up of a standard feed horn and a transmitarray aperture, each of which comprises a three-layer metasurface that is well-designed to tune the transmitted phase and amplitude on its own. As a result, the transmitarray aperture can cover the 360° phase range required for phase compensation while also realizing the amplitude distribution calculated by the Taylor design, contributing to the low SLLs. The advantage of independently controlling the amplitude and phase not only enables the proposed TA to be easy to realize and less cost on optimization but also improves the performance of high gain and low SLLs with a low profile and easy fabrication. Theoretical analyses and simulations are carried out, and a prototype is fabricated and measured. At 10 GHz, the measured peak gain is 21.16 dBi, with SLLs of −31.21 and −29.93 dB in the H- and E-planes, respectively. As a result, the tight agreements between measurement and simulation validate the design concept.
17 February 2022 in IEEE Transactions on Antennas and Propagation
This article investigates the pattern synthesis of the conformal phased array antenna (PAA) by using the deep deterministic policy gradient (DDPG) algorithm, and a nearly full solid angle for beam steering is realized. The beam steering capability of the planar and conformal PAAs is theoretically compared at first, and a conclusion is obtained that conformed to the conical-and-cylindrical structure can help to achieve ultrawide-angle beam steering. Next, a typical deep reinforcement learning algorithm, which is the DDPG algorithm, is utilized to deal with the fast beam steering problem of the conformal heterogeneous PAA. By virtue of the strong fitting ability of the DDPG algorithm for high-dimensional continuous nonlinear problems, the performance of fast beam steering is achieved within a wide-angle range within (−150°, 150°). Finally, a prototype of $1\times17$ conformal PAA is fabricated for measurement and verification, and the measured results are in good agreement with the simulation results.
2 March 2023 in IEEE Transactions on Antennas and Propagation
The array antenna will be inevitably deformed affected by the motion of the communication system or other special requirements, which will reduce the original good radiation performance, such as the gain dropping, the side lobes increasing, and even the beam splitting. In order to cope with the beam deterioration caused by the array antenna’s deformation, it is necessary to quickly manipulate the amplitude and phase distribution according to the current conformal array shape to stabilize the beam. This paper proposes a physical-method-driven deep-learning-based (PMDL-based) fast beam stabilization algorithm for the antenna array deformation. Firstly, we theoretically analyze the radiation pattern synthesis of the conformal array with the arbitrary surface to design the corresponding physical method, and we verify the accuracy of the method by the calculated and simulated results. Then, the deep neural network driven by the physical method is designed integrated with the aforementioned radiation pattern synthesis, whose training process is given. Finally, a 1×16 array antenna is taken as an example to verify the validity of the beam stabilization algorithm when the designed array is deformed. The simulated and measured results both indicate that the gain dropping and the sidelobe level rising are respectively less than 1.5 dB and 2 dB by applying the proposed PMDL-based fast beam stabilization when the array antenna is deformed within the range of 100° of the normal vector of each array element, while the beam stabilization time is less than 1.0 ms (testing time).
18 October 2022 in IEEE Transactions on Antennas and Propagation
Conformal array antennas (CAAs) can effectively broaden beam scanning range and improve radiation flexibility. Unfortunately, conformation will increase the difficulty of analysis and synthesis, and the stealth of CAAs is arduous. In this paper, we propose an innovative array optimization algorithm for stealthy CAAs based on the twin delayed deep deterministic policy gradient (TD3) algorithm, which combines the advantages of deep learning and reinforcement learning. Furthermore, we characterize the CAA as an electromagnetic model (EMM) and embed it into the TD3 algorithm as the environment to improve its stability and efficiency. Additionally, we independently excite the radiation and absorption modes to integrate radiation and stealth characteristics into the CAA. The simulated and measured results indicate that the designed CAA with EMM-TD3 algorithm features high-flexibility beam scanning within ±50° around 6 GHz and 10-dB radar cross section reduction in the frequency bands of 3.1~5.3 GHz and 6.5~11.2 GHz.