Multiple-input multiple-output millimeter-wave array utilizing very short wavelength and large physical aperture provides a key path to achieve high-resolution real-time imagery. However, large-scale array and short wavelength make the system very sensitive to non-systematic, non-ideal and unpredictable errors. The errors manifest as significant phase and amplitude perturbations that become dynamically during continuous operation, resulting in inconsistent responses among the transceivers. It will degrade the imaging performance in terms of poor resolution and increased side-lobes. To this end, a novel adaptive auto-calibration algorithm, termed Amplitude and Phase Errors auto-Calibration based on Dynamic Bayesian Learning (APEC-DBL), is proposed. The algorithm formulates phase and amplitude perturbations as random multiplicative factors in the echo data domain. It uses dynamic Bayesian learning to solve the perturbation function and obtains a closed-form analytical solution. The entire process enables auto-calibration without human intervention. Within the Bayesian framework, the algorithm incorporates prior information about the target. It builds a joint probability model for the perturbations, noise, and target scattering coefficients. To avoid the complicated high-dimensional integration of the full posterior distribution, it uses variational Bayesian inference to compute the complete posterior distribution of the target scattering coefficients. The scheme turns the calibration process into a learnable inference problem. It unifies the perturbation auto-calibration and high-resolution imagery within an end to end framework. Simulation results show the algorithm maintains an 8 mm imaging resolution at 0 dB SNR and achieves the lowest PSLR and ISLR compared with conventional methods. Furthermore, effectiveness and superiority of the proposed algorithm are validated using raw echo data from an experimental prototype and security screening system.
The non-stationary ionosphere induces Doppler spreading of sea clutter in skywave over-the-horizon radar (OTHR), which severely degrades target detection. This paper develops a modeling and simulation framework to explain two typical types of anomalous spread sea clutter: striplike and clumplike patterns. Ray tracing is used to determine multi-hop propagation paths of radio waves, while a localized multiple phase screen is combined with ray tracing to describe the scattering, diffraction, and drifting effects of finite-size ionospheric irregularities. The affected ray set is explicitly mapped to the range-Doppler (RD) domain, enabling analysis of clutter morphology. Simulation results reproduce the striplike and clumplike patterns observed in real OTHR data. The results show that rangeambiguity folding of multi-hop echoes, together with time-varying ionospheric phase, produces striplike spreading, which is further enhanced by medium-scale traveling ionospheric disturbances. In contrast, clumplike spreading is mainly caused by localized drifting irregularities that selectively decorrelate only part of the sea-clutter rays.
This paper proposes a dual-channel multiplexing circularly polarized (CP) folded transmitarray (FTA) capable of generating Bessel orbital angular momentum (OAM) vortex beams. A linearly polarized (LP) to CP element with full 360° geometric phase (GP) coverage is firstly developed for the top TA. Next, an LP-to-LP polarization torsion element is constructed for the bottom reflectarray (RA) to achieve orthogonal LP conversion. Based on the proposed LP-to-CP phase-shifting element, the phase distribution of the TA is constructed, which can simultaneously record the phase information of two non-diverging vortex OAM beams with different modes. The proposed design enables channel multiplexing of non-diverging vortex OAM beams with l1 = +1 and l2 = -1 modes. To validate the performance of the proposed design, a prototype is fabricated and tested. Measured results indicate that the joint bandwidth satisfying -10 dB impedance and 3-dB axial ratio (AR) reaches 24 GHz to 32 GHz (28.6%). The resultant OAM vortex beams exhibit mitigated beam divergence, yielding a long non‑diffractive propagation distance up to 800 mm (73 λ0). Moreover, the measured mode purities for both l1 = +1 and l2 = -1 channels exceed 76.8%.
This paper presents an AI-driven automated design framework that directly maps prescribed monostatic radar cross section reduction (RCSR) objectives to the final designs within a unified design process. It avoids the repeated expert-dependent multi-step redesign generally required by conventional approaches when the prescribed objectives change. A convolutional neural network (CNN)-based surrogate model is constructed for fast electromagnetic response prediction of generated units, while a long short-term memory (LSTM)-based reinforcement learning policy sequentially generates the unit topologies and array arrangements according to the prescribed RCSR objective. Furthermore, mirror-symmetric unit-cell configurations and the corresponding paired arrangement are incorporated to provide polarization-insensitive characteristics of generated designs. Finally, two RCS reduction plates with different prescribed objectives (denoted as RCSR-MS1 and RCSR-MS2) are generated and validated through full-wave simulations, fabrication, and measurements. Under normally incident LP waves, RCSR-MS1 achieves 15-dB RCSR over 9.6–21.2 GHz, while RCSR-MS2 achieves 10-dB RCSR over 7.0–22.8 GHz (106.1% fractional bandwidth). Under normally incident CP waves, RCSR-MS1 achieves approximately 15-dB RCSR over 9.5–21 GHz, while RCSR-MS2 achieves approximately 10-dB RCSR over 7–23 GHz (106.7% fractional bandwidth). Moreover, the generated designs also exhibit a certain degree of angular stability, with the prescribed RCS reduction performance generally maintained within the incidence angles of approximately 30°. Both generated designs satisfy their respective prescribed RCSR objectives, thereby validating the effectiveness of the proposed framework and its requirement-to-complete-design capability for different RCS reduction objectives.
In the microwave low-frequency bands such as the P-and L-bands, metasurface unit cells inherently exhibit high resonant quality factors (Q-factors), leading to narrowband absorption and poor tunability. To overcome this limitation, we propose a resistive-capacitive co-regulation mechanism that enables coordinated regulation of the Q-factor and resonance configuration. By introducing tunable resistive and capacitive branches within a windowed architecture, the design produces multiple zero-crossings of the input admittance’s imaginary part, thereby generating multiple resonances and significantly broadening the absorption bandwidth. Equivalent circuit modeling establishes the theoretical link between Q-factor regulation and resonance formation, while simulations and experiments validate the effectiveness of the approach. The proposed strategy provides a robust framework for broadband and programmable metasurface absorbers in low-frequency microwave applications.
