Communication signal modulation recognition faces challenges including label scarcity, high dimensionality, nonlinearity, and real-time constraints. In recent years, deep learning-based modulation recognition has emerged as a prominent research focus; however, insufficient sample sizes frequently lead to overfitting, poor generalization, and the curse of dimensionality. This paper proposes a modulation recognition framework based on the Multi-Domain Feature Fusion and Tabular Prior-Data Fitted Network (MD-TabPFN). Adopting an optimization paradigm of general model + domain adaptation, this study represents the first attempt to transfer the tabular foundation model TabPFN to the field of communication signal recognition, enabling high-precision signal classification with only a single forward pass. The framework leverages deep multi-domain feature fusion to construct a tailored dataset that supports model inference, and introduces a Domain Attention (DA) module to dynamically allocate feature weights, thereby focusing on critical domain-specific information. Extensive experiments demonstrate that in scenarios with limited labeled samples (1–200 samples per class(SPC)), the proposed MD-TabPFN achieves state-of-the-art modulation recognition accuracy while maintaining highly efficient real-time inference speed, significantly outperforming comparative methods.

To address the practical challenges of satellite-assisted mobile communication for Internet-of-Things (IoT) devices—such as the needs for low-profile antennas, wide-angle beam steering under strict size and power constraints, and the limited tuning flexibility of conventional reconfigurable structures—this work proposes a compact dual-polarized antenna with digitally controlled pattern steering. The method combines mirror-image theory with transmission-optics-based phase regulation, enabling beam steering through a reconfigurable digital meta-surface that replaces the conventional ground plane. A planar dual-polarized dipole is positioned above the meta-surface, and beam steering is achieved by digitally switching the PIN-diode states to alter the discrete reflection-phase distribution. The prototype, fabricated and characterized at 12–14 GHz, demonstrates wide-angle beam steering with stable gain, low cross-polarization, and FPGA-based real-time digital control. The proposed approach provides a compact and energy-efficient solution for next-generation satellite-assisted IoT terminals requiring flexible and robust beam steering.

The cross-polarization backscatter positioning system based on the information-energy decoupling mechanism can achieve high gain and obstacle avoidance advantages while reducing the number of positioning antennas. A novel dual-band quad-polarized antenna design with in-band radar cross section (RCS) reduction using polarization conversion metasurface (PCM) for IoT anti-multipath indoor positioning is first proposed in this article. At first, a dual-band reflective miniaturization PCM unit is designed to analyze the relationship between the polarization conversion band and in-phase reflection band under u- and v- polarized incident waves. This provides theoretical support for PCM to simultaneously achieve radiation enhancement and in-band RCS reduction. Then, a dual-band shaping dipole antenna is designed as a radiator. Finally, a dual-band Wilkinson power divider and Butler matrix are designed to feed the$2\times 2$dipole array to achieve quad-polarization. In the overall design of the antenna, according to the principle of in-phase reflection in the dual-band, every$4\times 4$subarray of PCM is used as a reflector for every dipole antenna in the v direction to realize unidirectional radiation and low-profile design. The quad-polarization of X, Y, left-hand circular polarization (LHCP), and right-hand circular polarization (RHCP) is achieved within the bandwidths of 0.85–0.93 GHz and 2.38–2.5 GHz. Based on the principle of reflection phase cancellation in the dual-band, an$8\times 8$arrangement of PCM units with their mirror units is designed to achieve in-band monostatic RCS reduction in the dual-band range of 0.745–1.02 GHz and 2.1–2.67 GHz. Therefore, this antenna is highly suitable for the application of item-level high-precision cross-polarization positioning systems in multiconductor environments with indoor deployment of multiple IoT nodes.

A few recent developments have demonstrated that deep-learning-based solutions can outperform traditional algorithms for automated pavement crack detection. In this paper, an efficient deep network called CrackNet-V is proposed for automated pixel-level crack detection on 3D asphalt pavement images. Compared with the original CrackNet, CrackNet-V has a deeper architecture but fewer parameters, resulting in improved accuracy and computation efficiency. Inspired by CrackNet, CrackNet-V uses invariant spatial size through all layers such that supervised learning can be conducted at pixel level. Following the VGG network, CrackNet-V uses 3 × 3 size of filters for the first six convolutional layers and stacks several 3 × 3 convolutional layers together for deep abstraction, resulting in reduced number of parameters and efficient feature extraction. CrackNet-V has 64113 parameters and consists of ten layers, including one pre-process layer, eight convolutional layers, and one output layer. A new activation function leaky rectified tanh is proposed in this paper for higher accuracy in detecting shallow cracks. The training of CrackNet-V was completed after 3000 iterations, which took only one day on a GeForce GTX 1080Ti device. According to the experimental results on 500 testing images, CrackNet-V achieves a high performance with a Precision of 84.31%, Recall of 90.12%, and an F-1 score of 87.12%. It is shown that CrackNet-V yields better overall performance particularly in detecting fine cracks compared with CrackNet. The efficiency of CrackNet-V further reveals the advantages of deep learning techniques for automated pixel-level pavement crack detection.

In order to cope with the increasing complexity of wireless networks, dynamic spectrum access, and varying application needs, fast and flexible reconfiguration of protocol stack is desired. Since Medium Access Control (MAC) layer plays a pivotal role in providing efficient spectrum sharing and utilization, rapid on-the-fly reconfigurability of MAC protocols is highly important. We have designed and implemented TRUMP: a Toolchain for RUntiMe Protocol realization. Based on the component-oriented design approach, TRUMP allows runtime realization, reconfiguration, and optimization of MAC layer according to the varying application requirements, spectral environmental, and network conditions. TRUMP, with a platform-independent MAC meta-language, also enables rapid MAC prototyping. In this article, we carry out a detailed performance evaluation of MAC schemes realized through TRUMP on WARP SDR platform. Our results indicate that TRUMP allows reconfiguration of MAC schemes in the order of a few microseconds, thus meeting the strict timeliness requirements of MAC processing. We also present application examples to highlight capabilities of TRUMP.

To adapt to variation of polarization dependent loss (PDL) in depolarization channel, an adaptive polarization modulation scheme which can improve the spectral efficiency of polarization modulation system is proposed. In the proposed scheme, the transmitter adapts the optimal constellation size to the variation of the channel due to PDL while fulfilling a fixed target BER constraint, which is based on the received feedback information of channel state information estimated at the receiver. Our simulation results and analysis show that, adaptive polarization modulation (APM) can achieve higher spectral efficiency (SE) than nonadaptive polarization modulation, subject to the same fixed target BER constraint.

We present a 3D object relighting technique for multiview-multi-lighting (MVML) image sets. Our relighting technique is a fusion of multi-view stereo (MVS) technique and image based relighting (IBL) technique. The MVML dataset consists of multiple camera view with each view filmed under multiple time-multiplex illumination modes. A multi-view 3D reconstruction algorithm is first applied using traditional multi-view stereo algorithm. After this, the reconstructed model is relighted through an image based relighting scheme for each camera view, followed with view-independent texture mapping procedure. Interactive relighting results demonstrate our high quality reconstruction accuracy, realistic relighting effects and real-time relighting performance. Moreover, our relighting technique is suitable for dynamic 3D object relighting.