Abstract:As a crucial component of modern communication networks and an important domain of international competition, satellite communication faces increasingly severe interference threats, with its security and reliability encountering serious challenges. Against this backdrop, interference awareness technology has become a critical means of ensuring communication reliability, holding significant practical importance and promising application prospects. This paper first conducts a comprehensive analysis of several types of satellite communication interference signals. On this basis, the current research status of interference awareness technologies for satellite communication is investigated from three perspectives: interference detection, identification, and localization, along with the advantages and limitations of various technologies in the current interference environment. Finally, a brief outlook on the future development directions of interference awareness technology is provided.

Abstract:With the rapid growth of satellite communication demands, the scarcity of spectrum resources has become increasingly prominent. To ensure full utilization of system bandwidth and reduce the performance gap between adjacent 2²ⁿ-order Quadrature Amplitude Modulation(QAM) superposition schemes, this paper investigates the application of a 2²ⁿ⁺¹-order-based superposed 32QAM scheme in Non-Orthogonal Multiple Access(NOMA) satellite networks. First, a 32QAM constellation formed by superposing a novel 8QAM and Quadrature Phase Shift Keying(QPSK) is designed, and joint encoding is employed between two users to achieve rotation of the 8QAM signal. Then, to obtain the optimal superposed constellation with the objective of maximizing Constellation-Constrained(CC) capacity, the optimal power allocation algorithm is simplified under high Signal-to-Noise Ratio(SNR) conditions. Simulation results demonstrate that when Mutual Information(MI) exceeds 4.98, the proposed scheme achieves a 0.29 wider range of power coefficient ratios compared to the square 8QAM+QPSK scheme, and its Bit Error Rate(BER) performance outperforms the other three schemes(square 8QAM+QPSK, conventional 8QAM+QPSK, and superposed 32PSK).

Abstract:Routing methods in mega-constellation networks are critical and challenging for network performance. Dynamic topology changes incur a large amount of state interaction and routing update overhead, and require more routing convergence time. In this paper, from the perspective of complex network theory, we propose a pinning control-based routing method applicable to mega-constellation networks, where a small number of key nodes are selected based on the Maximum Local Traffic(MLT) in a distributed control framework for mega-constellation networks, and scalable routing is realized by controlling the states of these nodes. Simulation results demonstrate that the proposed routing selection method achieves performance comparable to or surpassing existing routing methods while reducing routing information exchange overhead. This validates its effectiveness as a novel routing control paradigm.

Abstract:With the increasing demand for high spectral efficiency(>10 bps/Hz) and ultra?reliability (BER(Bit Error Rate)<10??) in 6G communications, this paper proposes a joint Orthogonal Time Frequency Space(OTFS) waveform design for terahertz(THz)?based Low?Earth Orbit(LEO) satellite communications. Aiming at challenges including 1.2 MHz Doppler shifts,120 μs delay spreads, and power amplifier nonlinearity(requiring PAR(Peak?to?Average Ratio)<4 dB) in satellite-terrestrial links, a Fractional Fourier Transform(FrFT)?based super?resolution channel estimation(error≤3 kHz) is developed,an Intelligent Reflecting Surface(IRS)?assisted 3D spatio-temporal-frequency beamforming architecture with 256 elements is established, and a Riemannian manifold?constrained constant envelope precoding algorithm is designed. Simulation results demonstrate that at 0.3 THz with 7 km/s relative velocity, the system achieves a BER of 1×10^-7, a PAR of 3.5 dB, and a spectral efficiency of 12.4 bps/Hz(40% improvement over Orthogonal Frequency Division Multiplexing(OFDM)), while boosting power amplifier efficiency from 12.7% to 78.4%. Validated by 3rd Generation Partnership Project(3GPP) Non-Terrestrial Network(NTN) channel models, this scheme provides a highly robust and power-efficient waveform solution for 6G space-terrestrial integrated communications.

Abstract:To address the limited transmission capability of power grid monitoring information in remote and disaster?affected areas, this paper proposes a Low Earth Orbit(LEO) satellite?based Internet of Things(IoT) transmission scheme tailored for emergency power grid monitoring applications. Firstly, a multi?level priority scheduling mechanism is developed by incorporating key transmission parameters such as the number of retransmissions and backoff time, thereby ensuring the accuracy and timeliness of information delivery according to different emergency levels. Secondly, in consideration of the highly dynamic characteristics of satellite/Unmanned Aerial Vehicle(UAV) communication channels, a closed?loop demodulation method is proposed to mitigate the impact of time?varying Doppler shifts. This method extracts dynamic Doppler frequency offset information by analyzing the relationship between the main and side peaks in Fast Fourier Transform(FFT)?based demodulation decisions, and introduces a feedback loop to dynamically compensate for the frequency offset in subsequent symbol demodulation. This significantly enhances the Bit Error Rate(BER) performance of IoT transmissions under high?mobility channel conditions. Finally, to validate the effectiveness of the proposed method, a comprehensive performance evaluation is conducted through Matlab simulations and Field?Programmable Gate Array(FPGA)?based hardware testing. Both simulation and FPGA results demonstrate that the proposed closed?loop demodulation scheme achieves a markedly better BER performance than the traditional method. At a BER of 10^?3, it provides an Signal-to-Noise Ratio(SNR) gain of about 2 dB at Spreading Factor(SF)=9, and the BER curve no longer shows a flooring effect as the SNR further increases.

Abstract:Coded-aided carrier synchronization can fully utilize the soft information of data symbols from the decoding process for carrier parameter estimation, demonstrating excellent performance under low Signal-to-Noise Ratio(SNR) conditions. Classical coded-aided carrier synchronization methods mainly solve nonlinear equations containing hidden variables through the Expectation-Maximization (EM) algorithm or its series expansion approximations, but suffer from high complexity or limited estimation accuracy. This paper exploits the characteristic that the Jacobian matrix of the log-likelihood function derivative in polar form is easy to derive, and uses the Newton-Raphson iterative algorithm to solve the nonlinear function. Under the condition of maintaining parameter estimation performance without degradation, the computational complexity of coded-aided carrier synchronization parameter estimation is significantly reduced. Computer simulations show that the proposed algorithm can achieve Bit Error Rate(BER) performance comparable to the EM algorithm while significantly reducing computational complexity, approaching that of series expansion approximation algorithms. It is applicable to single carrier linear modulation systems such as Binary Phase Shift Keying(BPSK), Quadrature Phase Shift Keying(QPSK), and Quadrature Amplitude Modulation(QAM).

Abstract:Located between the microwave millimeter and infrared frequency bands, terahertz waves possessing both electronic and photonic properties have been extensively researched and utilized in various fields such as high-speed communication, radar imaging, atmospheric remote sensing, and astronomical observation. The terahertz frequency source is a crucial component of the terahertz system, playing a vital role in its overall performance. In recent decades, solid-state terahertz frequency sources based on semiconductor devices have gained widespread attention due to their compactness, high degree of integration, and low cost of production at scale. Among nonlinear semiconductors, the Schottky diode offers several advantages, including a high cutoff frequency, low power consumption, and good noise performance. These features make it the preferred choice for low-cost and high-performance frequency sources in various terahertz applications. As a result, Schottky diode-based multiplied sources have become an important part of solid-state terahertz frequency sources. This paper reviews the development history of Schottky diodes and the latest research progress on solid-state THz frequency multipliers based on Schottky diodes, and discusses the future development prospects of solid-state Schottky-based THz frequency sources based on this foundation.

Abstract:With the advancement of information technology, information security has garnered significant attention, and image encryption technologies have consequently emerged. Metasurfaces, capable of flexibly manipulating electromagnetic waves, hold substantial potential for application in the field of image encryption. Currently, research on metasurface-based image encryption primarily focuses on the optical frequency band. To address this, a chiral metasurface operating in the terahertz frequency band has been developed. This metasurface exhibits completely opposite absorption characteristics for circularly polarized waves in its original and mirror configurations. To better elucidate the chirality principle of the designed metasurface, a multiple-interference model is introduced for theoretical validation. By optimizing the structural parameters, an image-encoding structure based on the chiral metasurface is constructed. It is demonstrated that distinct barcode images can be obtained under the incidence of terahertz waves with different polarization states. This work highlights the significant potential of chiral metasurfaces for image encryption applications.

Abstract:Near-field microwave antennas play a crucial role in the precise control of quantum spins, but traditional microwave antennas often struggle to achieve a balance between miniaturization, high radiation efficiency, and large bandwidth. To address this challenge, this paper proposes a circular reconfigurable antenna based on circular patch antenna principles. Compared with traditional copper loop antennas, the Optically Detected Magnetic Resonance(ODMR) fluorescence contrast of the circular reconfigurable antenna is improved by 1.5 times. By adjusting the tuning voltage, the circular reconfigurable antenna achieves a linear tuning bandwidth of 1.25 GHz without compromising microwave field uniformity. This innovative design solves the problem of narrow operating range in high-radiation-efficiency antennas while simultaneously achieving antenna miniaturization and load frequency offset correction. Experimental results show that using the circular reconfigurable antenna, the fluorescence contrast of the diamond Nitrogen Vacancies(NV) center signal reaches 8.5%, the linear bandwidth is increased from 2.25 GHz to 3.5 GHz, and the sensitivity reaches 78 pT/Hz^1/2, verifying the excellent performance of the designed antenna. These characteristics make the circular reconfigurable antenna a promising ideal candidate for integrated diamond NV center detector technology in large dynamic range magnetic field and microwave detection.

Abstract:As a vital part of the feed system for monopulse phased array antennas, the sum-difference beamforming network's amplitude and phase consistency directly determines the key performance metrics such as the null depth of the difference beam, beam pointing accuracy, and sidelobe suppression. This paper proposes and designs a K-band sum-difference beamforming network composed of a broadband sum-difference device and a power divider network. The network is fabricated using a "sandwich" structure, where the entire stripline-based beamforming network is automatically press-bonded with upper and lower structural cavities after processing. Compared with traditional manual welding of microstrip lines and structural cavities, this approach offers significantly improved manufacturing consistency, reduced production and assembly difficulty, higher production efficiency, and superior assembly yield. Through design, fabrication, and practical testing, the following performance metrics are achieved within the Fl~Fh frequency range:VSWR≤2 for all ports;port amplitude consistency(Root Mean Square(RMS) statistic)≤0.21 dB; port phase consistency(RMS statistic)≤2.8°. The test results demonstrate that this form of sum-difference beamforming network exhibits excellent performance and can be widely applied to engineering mass production tasks.

Abstract:Large Language Models(LLMs)are machine learning models characterized by massive parameters and computational resources, enabling them to better capture patterns and regularities in data when processing complex tasks. However, since LLMs are typically trained through self-supervised learning on vast amounts of unlabeled data, this can result in opaque decision-making processes, increased difficulties in controlling outputs, and potential risks of generating harmful content. In response to these issues, this paper provides a systematic review of existing LLM evaluation methods. Firstly, we summarize, analyze, and compare evaluation approaches across five dimensions: reasoning capabilities, factuality of generated content, tool usage and agent capabilities, robustness, and bias in generated content. Secondly, we present a detailed survey of datasets employed in the evaluation process. Finally, we outline the major challenges in current LLM evaluation and discuss potential future research directions.

Abstract:Radar maneuvering target tracking is a process of estimating the target's position state through measurement information. Traditional Kalman filtering algorithms exhibit significant degradation in tracking accuracy when confronted with nonlinear motion equations. While intelligent algorithms leverage attention mechanisms to capture long-range dependencies of features and enable precise modeling of unknown motions, they suffer from insufficient local information extraction capabilities, leading to decreased tracking accuracy when the target undergoes local maneuvers. To address this issue, this paper proposes an end-to-end filtering and prediction algorithm for radar maneuvering targets based on Conformer-Multi-Token Prediction(Conformer-MTP). The model employs a Conformer-structured encoder with an additional convolutional module to enhance the attention mechanism, achieving an organic integration of global modeling and local information extraction. Simultaneously, the model utilizes an MTP-structured decoder that unifies the filtering and prediction processes, allowing them to share partial decoder parameters, thereby reducing the overall computational complexity and parameter count of the system. Comparative tests conducted on datasets comprising five motion types demonstrate that the Conformer-MTP algorithm achieves higher filtering and prediction accuracy than the Interacting Multiple Model(IMM) algorithm, with filtering error reduced by 21.62% and prediction error reduced by 16.64%. Furthermore, ablation studies reveal that compared with the standard Transformer algorithm, the proposed algorithm reduces filtering error by 58.52% and prediction error by 39.28%, while decreasing the total parameter count by 36.56% and average processing time by 10.9%. These results indicate that the Conformer-MTP structure can effectively enhance model training performance and improve overall accuracy.

Abstract:General signal detection models typically feature large parameter sizes and high computational complexity, making them difficult to deploy directly on resource-constrained edge devices. Consequently, model lightweighting has become one of the active research topics in this field. To address this challenge, this paper proposes a lightweight communication signal detection algorithm named ISR-YOLO(I-Shufflenet Rec-SPPF(Spatial Pyramid Pooling-Fast)-You Only Look Once). A lightweight network, IShufflenet, is designed as the backbone feature extraction network, where the feature extraction modules are reconstructed to improve signal detection accuracy. In the neck network, a feature fusion module called Rec-SPPF is designed based on signal geometric characteristics to further enhance the integrity of signal bounding boxes. Additionally, the Convolutional Block Attention Module(CBAM) is incorporated to improve the sensitivity of the lightweighted network to elongated signals. Furthermore, the Enhanced Intersection over Union(EIoU) Loss, which separately calculates the height and width of bounding boxes, is adopted as the new bounding box loss function to improve the network's localization accuracy. Experiments conducted on the IEEE SPAW2021 public dataset and simulated datasets demonstrate that ISR-YOLO achieves performance almost identical to YOLOv5 while reducing computational complexity and model size by 87.34% and 82.64%, respectively, significantly decreasing the dependency on high-computing-power hardware at the edge.