最新刊期
卷 53 ， 期 7 ， 2025
- 摘要：In modern mobile communication systems, joint source-channel coding (JSCC) enables end-to-end optimization, improving spectral efficiency. Hybrid automatic repeat request (HARQ) improves link reliability, adaptability, and enhances the robustness of the system through its flexible retransmission mechanism. Joint optimization of JSCC and HARQ can further improve both reliability and spectral efficiency. In this paper, we propose a polarizing matrix extension (PME)-based JSCC in HARQ systems, referred to as PME-JSCC-HARQ. The PME-JSCC-HARQ scheme first performs source polar encoding, then extends the source polarizing matrix, placing the channel bits into the extended positions.During the retransmission process, the polarizing matrix is progressively extended, and retransmission bits are placed in the extended positions. Due to the lower triangular structure of the polarizing matrix, the transmission bits in each round of transmission do not affect the source bits or the previously transmitted encoded bits in the joint encoding process. To improve the decoding reliability of the PME-JSCC-HARQ scheme, the most reliable bit subchannels of the extended long polar code are selected as the information bit subchannels. Simulation results show that, when the list size is 32, compared to separate designs based on source polar coding and channel polar coding with retransmissions, the proposed scheme provides a performance gain of over 2.8 dB.关键词：polar codes;joint source-channel coding(JSCC);HARQ;polarizing matrix extension153|50|0更新时间：2025-12-10
- 摘要：5G new service scenarios require low latency and high reliability of transmission. Forward error coding (FEC) at the application layer can quickly recover lost data packets, but the decoding recovery performance in high burst packet loss channels is poor. To solve this problem, this paper proposes an interleaved streaming-random linear network coding (IS-RLNC) scheme, which encodes source packets at equal intervals and inserts encoded packets into source packets to send. This scheme improves the recovery performance in burst condition while keeping low decoding complexity and real-time performance. Experimental results show that IS-RLNC has a lower decoding failure probability than Reed-Solomon (RS) and Caterpillar RLNC (CRLNC) in high burst channel environments. IS-RLNC is able to recover more packets than RS codes in scenarios with tight delay constraints. Finally, experiments show that the optimal decoding performance for different burst channels has different interleaving depths.关键词：URLLC;FEC in application layer;RLNC;interleaved coding;stream coding111|38|0更新时间：2025-12-10
- 摘要：Wireless image transmission faces the dual challenges of bandwidth and computing resources, which is particularly prominent in application scenarios such as the Internet of Things where node computing power is limited. Joint source-channel coding (JSCC) can optimize both source and channel coding, and has gradually become an important research direction in wireless image transmission. In recent years, deep learning-based JSCC methods have received widespread attention, which achieve joint optimization of encoders and decoders through end-to-end training. However, most encoders of deep learning-based JSCC methods involve a large number of linear and nonlinear operations, resulting in high computational complexity and difficulty in application to devices with limited computing resources such as edge computing nodes in the Internet of Things. In order to achieve a lightweight coding process, this paper proposes a joint source channel coding method BCS-JSCC (Block Compressive Sensing-Joint Source Channel Coding) based on deep compressed sensing to achieve end-to-end optimization of the codec. This method designs a compressed sensing sampling of learnable scale binary measurement at the encoding end to realize a lightweight encoding method that matches the decoder in a noisy environment; at the decoding end, the linear inverse problem of measurement value transmission is solved based on the MMSE (Minimum Mean Squared Error) criterion to obtain the initial reconstruction sensitive to channel noise and suppress the influence of noise on the parameter reuse reconstruction network. Compared with the existing JSCC method based on deep learning, the BCS-JSCC method proposed in this paper can achieve better transmission performance under high signal-to-noise ratio conditions while keeping the number of floating point operations per pixel (FLOPs per pixel) at the encoding end the same. The advantage is more obvious under low computing power (0.10 K FLOPs/pixel). The encoder of the BCS-JSCC method proposed in this paper has a simple structure and low computational complexity, and is suitable for deployment on low computing power devices such as edge computing nodes of the Internet of Things.关键词：wireless image transmission;joint source-channel coding;deep learning;encoder;compressive sensing151|53|0更新时间：2025-12-10
- 摘要：Blind recognition of channel coding parameters, as an essential technology in non-cooperative communication and adaptive modulation and coding (AMC) systems, has attracted increasing attention in recent years. In the 5th generation mobile networks (5G), low-density parity-check (LDPC) codes are adopted as the forward error correction codes for data channels. However, due to the use of puncturing and padding, traditional blind recognition techniques are no longer applicable. We propose a novel scheme that leverages the belief propagation (BP) decoding iterative approach for blind recognition. Based on the conventional log-likelihood ratio (LLR) algorithm, this scheme incorporates the BP decoding concept to iteratively process punctured and padded bits, addressing the issue that traditional algorithms cannot recognize these bits. Simulation results demonstrate that the proposed algorithm outperforms existing related algorithms in terms of performance.关键词：adaptive modulation and coding;blind recognition;low-density parity-check codes;shortening and puncturing;belief propagation decoding70|25|0更新时间：2025-12-10
- 摘要：Guessing random additive noise decoding (GRAND) algorithms are a class of general decoding algorithms for linear codes. Among them, ORBGRAND (Ordered Reliability Bits GRAND) algorithm, a soft-decision GRAND algorithm that has attracted a lot of attention, is simple and easy to implement. However, there is a gap between the performance of ORBGRAND algorithm and soft-decision maximum likelihood decoding algorithm if the number of queries is limited. To tackle this problem, this paper proposes a hybrid decoding algorithm based on syndrome decoding (SD) algorithm and ORBGRAND algorithm. By introducing an assistant metric for ORBGRAND algorithm, the proposed algorithm performs SD algorithm and ORBGRAND algorithm in parallel for the received sequences and selects the codeword that has the maximum correlation value with the received soft-decision sequences as the decoding output. Simulation results indicate that the proposed hybrid decoding algorithm can outperform ORBGRAND algorithm at the expense of little complexity increase. As the signal-to-noise ratio increases, the performance improvement becomes more obvious. In addition, the proposed hybrid decoding algorithm maintains the generality of ORBGRAND algorithm and is suitable for decoding various types of linear codes.关键词：general decoding algorithm;guessing random additive noise decoding (GRAND);maximum likelihood decoding;syndrome decoding;hybrid decoding algorithm;correlation77|21|0更新时间：2025-12-10
- 摘要：To reduce the latency of the parity-check successive cancellation list (PC-SCL) decoding algorithm for parity-check polar codes, a fast parity-check successive cancellation list (Fast-PC-SCL) decoding algorithm is proposed. Firstly, the algorithm analyzes and studies two types of special nodes in parity-check polar (PC-Polar) codes: PC-Repetition (PC-REP) nodes and PC-single parity-check (PC-SPC) nodes, and theoretically proves that PC-REP nodes exhibit cyclic repetition of codeword sequences, while PC-SPC nodes have the property of the sum of codewords equals a specific value. Secondly, based on these properties, codeword list estimation methods are given for these two types of nodes. This enables the decoding of these nodes to be executed in parallel, significantly reducing decoding latency. Finally, by combining the codeword list estimation methods for these two types of nodes, the Fast-PC-SCL decoding algorithm is presented. This algorithm can decode without completely traversing the successive cancellation (SC) decoding tree, while fully retaining the effect of PC bit checks. Compared to the PC-SCL algorithm, it significantly reduces decoding latency without sacrificing performance. Experimental data shows that it can reduce latency by up to 55.13%.关键词：polar codes;successive cancellation list;parity-check;decoding latency;special nodes46|43|0更新时间：2025-12-10
- 摘要：Spatially-coupled quantum LDPC (SC-QLDPC) codes have significant applications in distributed quantum computing systems, yet achieving corresponding low-latency and efficient decoding remains a core challenge in their practical implementation. This paper focuses on the severe impact of chain reactions caused by error propagation during sliding window decoding of SC-QLDPC codes on decoding performance. We propose a dual-window backtracking optimization decoding strategy for quantum dual-window sliding decoding algorithms applicable to SC-QLDPC codes. A cost-performance metric method for window expansion during decoding is presented, and strategies for selecting optimal small backtracking windows during quantum sliding window backtracking are discussed. Experimental results demonstrate that the proposed quantum sliding window backtracking strategy achieves optimized balance between decoding complexity and performance, effectively addressing the significant performance degradation caused by improper window size selection and error propagation during quantum sliding window decoding processes.关键词：quantum computing;quantum error correction;spatially-coupled codes;sliding window decoding;error suppression;quantum LDPC codes52|48|0更新时间：2025-12-10
Advances and Future of Coding Theory and Technology
- 摘要：The detection of small targets in normal illumination conditions is challenging. In low-light environments, images suffer from severe information loss due to low brightness, low contrast, and low signal-to-noise ratio, which further weakens the feature information of small targets, making feature extraction more difficult. As a result, research on small target detection in low-light environments is scarce. To address this issue, this paper proposes a low-light small target detection method, MC-YOLO. The MC-YOLO includes four modules: multi-scale feature fusion enhancement (MFFE) module, detail feature extraction (DFE) module, Neck module and Head module. Firstly, the method uses the MFFE module to extract small target features in low-light environment through deformable convolutions and multi-scale feature fusion and to enhance global features through global average pooling making small target feature information more salient. Next, the DFE module fully utilizes contextual information to extract small target features while preserving the positional information of the small targets, which solves the problem of easy loss of detail feature information of small targets in low-light environments. Then, the neck module performs feature extraction and multi-scale feature fusion. Finally, the head module introduces a small target detection layer on the high-resolution feature map to detect small targets in low-light environments. Experimental results show that this method performs well in the accuracy of low-light small target detection, with the mAP of 83.2% on the self-made low-light small target dataset LLSOD, which is 3.6% higher than the current advanced target detection method YOLOv11.关键词：low-light small targets;multi-scale;feature fusion;Feature enhancement;detail features121|43|0更新时间：2025-12-10
- 摘要：Among image enhancement techniques, curve mapping-based retouching strategies have attracted significant research interest due to their ability to effectively retain the original content information of images. However, current curve-mapping methods primarily focus on the changes in color space before and after enhancement, often neglecting the influence of image content on the enhancement results. This limitation leads to suboptimal adjustments for images with similar colors but different content, resulting in less refined and natural enhancements. To address this issue, this paper proposes an image enhancement method based on content-aware multimodal fusion, which supplements image features by incorporating text features that describe the semantic perception of image content. By fusing features from both image and text modalities, the proposed approach captures multimodal content-aware semantics, enabling fine-grained adjustments tailored to different image content. Firstly, a multimodal large language model is employed to extract textual descriptions of image content, which are then used for multimodal prompt learning to guide the understanding of the image content. This method enables the model to leverage content-based text prompts for auxiliary image enhancement. Then, an attention mechanism is then applied to effectively integrate and fuse the textual and image features into a unified multimodal representation. Finally, this representation is used to construct a curve-mapping function, enabling content-specific image adjustments and enhancements. Experimental results on multiple public benchmark datasets demonstrate that the proposed method achieves state-of-the-art performance, highlighting the effectiveness and advantages of incorporating content-aware semantic information into image enhancement tasks.关键词：image enhancement;Text Generation;content-aware;multimodal fusion;curve mapping152|66|0更新时间：2025-12-10
- 摘要：Object detection for high-resolution remote sensing images has become a key research area in intelligent remote sensing information processing, with extensive application scenarios and significant practical value. Unlike natural images, remote sensing images present unique challenges such as arbitrary object orientations, multi-scale variations, complex backgrounds, and densely arranged targets. To further improve the performance of high-resolution remote sensing image object detection, this paper proposes EIMYOLO, a novel rotated object detection algorithm based on YOLOv11, which incorporates innovative feature fusion and enhancement strategies. The proposed method integrates three plug-and-play modules designed to enhance feature representation and fusion. First, the Edge Feature Reinforcement Block improves orientation sensitivity and feature discriminability by extracting fine-grained edge information, especially in complex scenes. Second, the Interlayer Feature Enhancement Extractor boosts intralayer feature representation, particularly benefiting the detection of dense and elongated objects. Third, the Multi-Scale Attention Dynamic Fusion enhances inter-layer feature integration through adaptive attention mechanisms. Extensive experiments conducted on two benchmark datasets, HRSC2016 and DIOR-R, demonstrate the effectiveness of our approach, achieving state-of-the-art mean Average Precision (mAP) scores of 90.80% and 72.40%, respectively. These results confirm the superior performance of EIMYOLO over existing baseline and comparative methods.关键词：high-resolution remote sensing images;deep learning;feature fusion;Feature enhancement;attention mechanism144|55|0更新时间：2025-12-10
- 摘要：Detection transformer (DETR) has been emerging as a hotspot in computer vision, multimodal learning and other fields. However, its performance is heavily affted by the learning feature bias transmission between decoder layers, and the same reference points used by the cross-attention of different decoder layers, and the semantic vagueness of the encoder output features. To address these deficiencies, this paper employs Conditional DETR as the baseline and decouples its cross-attention mechanism into weights and values, then proposes an inter-layer adaptive attention weight refinement (IAAWR), with the aim of dynamically adjusting the cross-attention weights of different layers of the decoder, with a review to weakening the inter-layer transfer of learning bias. In addition, an adaptive feature enhancement (AFE) method is proposed utilizing divide and conquer idea, with the aim of improving the feature extraction capability of each layer of the encoder for the local region of the target, resulting in the enhancement of semantics in the output features. Furthermore, the strategy of parameter-free iterative reference point refinement (IRPR) is proposed to achieve dynamic update of the reference points of the prediction box, enhancing the flexibility and fineness of regression prediction.These three innovations have been integrated into the baseline model Conditional DETR, resulting in an adaptive and interpretable DETR model referred to adaptive and interpretable DETR (AI-DETR).This AI-DETR defeats the Conditional DETR in terms of average precision (AP) on the publicly available dataset microsoft common objects in context (MS-COCO) with 1.8 percentage points and on the very challenging real-world datasets Butterfly_2018 and Butterfly_2023 datasets with 1.3 and 0.8 percent points, respectively. The qualitative and quantitative analyses, in conjunction with visualisations of the results, elucidate and validate the individual contribution of each innovation within the AI-DETR.关键词：object detection;butterfly detection;deep learning;detection transformer (DETR);cross-attention mechanism99|37|0更新时间：2025-12-10
- 摘要：Existing unsupervised video object segmentation (UVOS) methods often employ pixel-level dense matching strategies to enchance model performance by aligning and fusing features among multiple frames or between a single frame and its corresponding optical flow. However, in challenging scenarios such as occlusion, camera shak, and motion blur, optical flow estimation errors can easily generate numerous erroneous matches, leading to overfitting of the fused spatio-temporal representations to motion noise. To address this issue, we propose a motion-prompts guided adaptive learning UVOS framework. By designing an unsupervised motion-prompts generation algorithm, the dense motion information encoded by optical flow is transformed into sparse point and box prompts. With the help of prompt learning, the segment anything model (SAM) is guided to adaptively learn through two lightweight adapters designed in this paper, thereby obtaining more robust spatio-temporal representations and enhancing the model’s noise resistance capability. To obtain effective prompts, we design an unsupervised motion-prompt generation algorithm. This algorithm calculates a series of statistical measures from the optical flow features to identify salient regions, then utilizes motion edge information and an adaptive threshold to eliminate pseudo-salient regions, ultimately generating the point and box coordinates that highlight the locations of motion-salient objects. To enhance the generalization ability of SAM in downstream UVOS tasks, an adaptive representations learning SAM model is proposed. By incorporating two light-weight feature adapters, the model adaptively extracts knowledge relevant to the downstream UVOS task from SAM’s general knowledge base, enabling accurate coarse localization of objects. To overcome the lack of attention to details in pure Transformer-based SAM, a convolutional neural networks (CNN)-based feature focusing refinement module guided by the location map is designed. The localization attention map generated by SAM progressively guides the refinement process, shifting the model’s focus from global coarse localization to local refinement, and ultimately producing more accurate segmentation masks. Our method has been thoroughly validated on three mainstream datasets: DAVIS 2016 (DAVIS16), financial and business management system (FBMS), and YouTube-Objects (YTOBJ). Compared with current state-of-the-art methods, our approach achieves improvements of 1.8%, 1.6%, and 2.6% in the region similarity metric, respectively, thereby fully demonstrating the effectiveness of our proposed method.关键词：unsupervised video object segmentation (UVOS);optical flow noise;segment anything model (SAM);prompt learning;adaptive representation learning;decouple appearance-motion learning;multi-modality40|51|0更新时间：2025-12-10
- 摘要：Transformer-based image restoration methods have demonstrated remarkable performance in single image super-resolution tasks, owing to their self-attention (SA) mechanism, which effectively captures non-local information, thereby achieving higher-quality high-resolution image reconstruction. However, the matrix multiplication operations in the self-attention mechanism consume substantial computational resources, making Transformer-based models generally challenging to deploy on low-power devices with limited computational capabilities and memory. Additionally, the low-pass characteristics of the SA mechanism restrict its ability to capture high-frequency local details, leading to overly smooth reconstruction results. To address these issues, we propose a multi-scale feature adaptive modulation network (MFAMNet) for single image super-resolution, whose core is the multi-scale feature adaptive modulation (MFAM) module. This module obtains low-frequency content at different scales through downsampling operations, computes the global variance of the input features to modulate the processed low-frequency features, and then adaptively aggregates the input features using the modulated features, thereby achieving efficient modeling of non-local information. After aggregating the input features, we introduce a channel attention mechanism to refine the fused features from the channel dimension, enhancing the extraction of shared information across all channels while dynamically reallocating cross-channel weights. Furthermore, since MFAM processes input features from a long-range perspective, it is necessary to supplement local contextual information. To this end, we also design a spatial enhancement module (SEM) as an effective alternative to complex self-attention mechanisms, significantly improving spatial local aggregation capabilities and further refining the features output from MFAM in both spatial and channel dimensions. Extensive experiments demonstrate that the proposed MFAMNet achieves a better trade-off between reconstruction performance and computational efficiency on public benchmark datasets. Notably, in 4× super-resolution, self-modulation feature aggregation network (MFAMNet) improves the average performance by 0.15 dB compared to the state-of-the-art self-modulation feature aggregation network (SMFANet) on five public test sets, while maintaining nearly the same model complexity, e.g., floating-point operations per second (FLOPs).关键词：single image super-resolution (SISR);lightweight;self-attention (SA);feature modulation;multi-scale feature75|41|0更新时间：2025-12-10
- 摘要：Evolutionary multi-objective feature selection algorithms face challenges such as high computational cost and slow convergence when addressing high-dimensional classification datasets. Multi-task optimization has emerged as an effective paradigm to reduce search dimensionality and improve efficiency, and has been increasingly applied to this domain. Nevertheless, existing approaches predominantly focus on feature importance while neglecting redundancy relationships among features, which may compromise optimization performance. To overcome this limitation, this study proposes a novel evolutionary multi-task multi-objective feature selection algorithm based on feature redundancy analysis, referred to as MTGA. The proposed method first clusters all features according to their redundancy metrics, dividing the high-dimensional space into low-redundancy clusters. Then, different features are selected from each cluster to construct multiple subtasks, thereby preserving key information while eliminating redundancy. For each subtask, a new reproduction operator is designed based on feature importance. Additionally, a knowledge transfer mechanism facilitates the sharing of important features across subtasks, mitigating the risk of premature convergence. To validate the proposed algorithm, extensive experiments are conducted on fourteen high-dimensional UCI benchmark datasets. The results demonstrate that MTGA outperforms multiple classical feature selection methods, exhibiting excellent performance.关键词：high-dimensional feature selection;feature redundancy;multi-objective evolutionary algorithm;multi-task optimization;knowledge transfer mechanism79|28|0更新时间：2025-12-10
- 摘要：In the context of multi-agent path finding (MAPF), the priority-based search (PBS) algorithm integrates a priority mechanism with the node expansion framework of conflict-based search (CBS), achieving notable efficiency in path planning. However, the greedy strategy in PBS, which prioritizes path cost, often leads to slow conflict resolution during the expansion of the priority tree (PT). To address this limitation, this paper proposes an improved PBS algorithm, improved priority-based search with conflict cost bayesian weight (IPBS-ccbw). Building upon path cost, the proposed approach incorporates conflict counts to construct a composite metric that balances path cost and conflict frequency. During the planning and expansion process, Bayesian updates are applied to the conflict cost weights of child nodes, effectively balancing conflicts and path costs. In addition, the algorithm introduces conflict monitoring and strategy reconstruction mechanisms to prevent the algorithm from falling into deep search traps. The results of simulation comparison experiments on Benchmark standard test maps as well as small-scale physical experiments show that the IPBS-ccbw algorithm exhibits superior path optimization capabilities in different environments. Compared with the PBS algorithm, the IPBS-ccbw algorithm demonstrates stronger conflict mitigation capability and higher solving efficiency in large-scale dense scenarios. The solution time can be reduced by 27.3% to 91.9%, and the solution success rate can be improved by 40% to 85% when the number of intelligences reaches the maximum.关键词：multi-agent;path planning;bayesian update;dynamic weight;conflict reduction43|25|0更新时间：2025-12-10
- 摘要：The Android system occupies over 70% of the market share of mobile operating systems, making it a key platform for malicious actors to distribute malware. Repackaged malware embeds a small amount of malicious code into legitimate software, masking malicious activities with a majority of benign behaviors to evade traditional malware detection methods. However, academic research on repackaged malware remains relatively limited. Existing detection methods based on partitioning function call graphs often lack generalizability and fail to fully capture the semantic features of malicious behavior associated with sensitive API(Application Programming Interface) centrality. To solve these problems, we propose Partdroid, a detection method for Android repackaged malware. The method analyzes manifest files and smali code to extract application component information and generate component function call graphs. It combines graphs of components with sensitive APIs and uses taint analysis to uncover inter-component relationships, forming a sensitive component function call graph to overcome partitioning limitations. Additionally, Partdroid highlights malicious behavior by exploring the relationships between sensitive APIs, entry functions, and interaction functions. It also integrates centrality algorithms to calculate the importance of sensitive APIs comprehensively, addressing the limitations of directly using centrality algorithms for feature extraction. Experimental results demonstrate that Partdroid outperforms other tools in detecting Android repackaged malware, achieving an F1 score of 91.34% and accuracy of 91.93% with a random forest classifier, and 91.63% and 92.15% with a voting algorithm. Moreover, Partdroid performs outstandingly in detecting new malware, identifying 3 suspicious software among 2 000 randomly selected applications from the Google Play Store.关键词：Android repackaged malware;function call graph;sensitive API;malicious behavior;machine learning72|43|0更新时间：2025-12-10
- 摘要：Bimodal action recognition aims to enhance recognition performance in complex scenarios by leveraging complementary information across different data modalities to overcome the limitations of single-modal approaches. Existing methods typically adopt independent backbone networks to extract features from each modality separately before performing feature fusion. However, they often fail to adequately address semantic discrepancies between modalities, such as cross-modal feature misalignment and representational inconsistency, which can introduce noise during the fusion process and degrade recognition accuracy. To address these issues, this paper proposes a spatiotemporal adaptive fusion framework for bimodal action recognition. Specifically, a temporal keyframe selection module is introduced to identify and emphasize informative frames through a competitive mechanism. Simultaneously, a spatial salient region selection module adaptively filters discriminative regions across modalities, suppressing irrelevant information and guiding the network to learn more robust spatiotemporal representations. In addition, a self-distillation mechanism is employed to reinforce the network’s focus on action-relevant features, incorporating both prediction distribution loss and region-level distillation loss to facilitate fine-grained feature optimization. To further improve the fusion quality, an adaptive mask fusion module is proposed, which attenuates the influence of uninformative regions by applying learnable masks within the multi-head self-attention and multi-layer perceptron computations. Experimental results on the InfRA and NTU RGB+D datasets demonstrate that the proposed method achieves Top-1 accuracy improvements of 3.75% and 3.49%, respectively, compared to baseline models, validating the effectiveness of the proposed framework in adaptively selecting and integrating bimodal features for improved action recognition.关键词：bimodal action recognition;key frame;salient region;self-distillation;adaptive fusion61|32|0更新时间：2025-12-10
-
Multi-Angle ISAR Recognition Method for Space Targets Based on Feature Correlation of Key Components
摘要：The geosynchronous space situational awareness program (GSSAP) satellite of the USA has repeatedly orbited and detected our satellites in recent years, which is a great threat. For this kind of orbiting maneuvering spacecraft, its continuous spin, spin axis change and orbiting motion vector constitute a three-dimensional rotation, and the scattering characteristics between adjacent inverse synthetic aperture radar (ISAR) image frames are greatly different, making it difficult to recognize it. To this end, this paper proposes a multi-angle ISAR recognition model for space targets based on feature correlation of key components. A contrast learning module based on self-supervised learning strategy is constructed to reduce the impact of parameter changes in imaging and target attitude on image recognition. A key component feature correlation module is constructed to mine local correlation information between key components of the images using graph information reasoning methods. Finally, a complex-valued transformer layer extracts global contextual features between image blocks and achieves effective expression of the target through feature fusion. Experimental results based on real radar data show that the proposed method can significantly improve the effect of multi-angle recognition. Under the same recognition condition of data volume, the recognition rate is increased by 5.58% compared with the existing recognition method, verifying the performance of multi-angle recognition.关键词：inverse synthetic aperture radar;space target;maneuvering target;complex-valued network;target recognition61|24|0更新时间：2025-12-10 - 摘要：The static and dynamic agents, road structures, and interactions among various elements in driving scenarios are typically complex and rapidly change across time and space. Consequently, motion prediction for autonomous vehicles remains a challenging task, especially with the open problem of efficiently representing and integrating multi-modal scene information, including road conditions, various agent states, and historical interaction information. Current approaches often rely on independently designed modules to process each modality in parallel. However, this approach tends to result in limited system flexibility, challenging adjustments, and, frequently, high computational redundancy, which reduces overall system efficiency. Furthermore, decoding the spatiotemporal information from autonomous driving scenarios to generate safe driving commands is inherently challenging. This paper proposes an autonomous driving motion planning method based on a spatiotemporal attention Transformer, comprising a phased multi-modal scene encoder and a spatiotemporal fusion decoder. This model progressively constructs a multi-modal scene representation and predicts the future safe trajectory of the autonomous vehicle under spatiotemporal fusion. The proposed approach establishes a new baseline on the large-scale nuScenes autonomous driving dataset, achieving competitive results.关键词：autonomous driving motion prediction;phased multimodal encoder;spatiotemporal fusion decoder;Transformer;new baseline74|35|0更新时间：2025-12-10
- 摘要：With the deep sub-micron technology in integrated circuit develops, Miller effect becomes non-negligible and the increasing interconnection resistivity increases, leading to increased delay time prediction inaccuracy. A composite current source stage delay calculation method based on dynamic capacitance is proposed in this paper. A voltage-based interpolation method is firstly introduced to support the delay calculation of composite current source. A cell delay calculation method with a Π model load is then established and a multi-threshold analysis is used to improves the effective capacitance, derives the dynamic capacitance and realizes the iterative calculation process. The dynamic capacitance concept is applied to the stage delay calculation, and used to realize the stage delay calculation method with distributed RC network as the load. By using machine learning, the interconnect wire delay is further optimized. Based on the ASAP 7 nm predictive PDK (ASAP 7) technology, the stage delay calculation method proposed in this paper achieves an average relative error of 1.49%, 3.16%, 1.70%, and 0.88% for the cell delay, cell transition time, interconnect line delay, and interconnect line transition time, respectively, compared with simulation program with integrated circuit emphasis (SPICE) simulation, the stage delay calculation method reaches convergence with about 4~5 iterations.关键词：delay model;stage delay calculation;composite current source model;miller effect;dynamic capacitance31|35|0更新时间：2025-12-10
- 摘要：The functionality of traditional metasurface is relatively simple, and it is difficult to meet the complex and changeable practical application requirements. To solve this problem, a multifunctional tunable terahertz metasurface based on vanadium dioxide (VO2) is proposed, which consists of two double split rings made of VO2 and metal, a dielectric layer, and a metal substrate. Based on the phase transformation characteristics of VO2 materials and the pancharatnam-berry phase principle, the metasurface unit can present different phase distributions in the two frequency bands of low frequency (0.8~1.2 THz) and high frequency (1.2~1.8 THz), and realize the independent regulation of the reflected circularly polarized wave. On this basis, three terahertz metasurface arrays with different functions are arranged to generate vortex beams and deflected vortex beams with multi-topological charges in the broadband range, as well as to realize the functional switching between the planar focused beams and the vortex beams at 1.2 THz.The proposed metasurface exhibits excellent performance in dynamic regulation, broadband response, and multi-functional integration, and provides new possibilities for miniaturization and multi-functionalization of optical components. It has potential application value in terahertz communication, adaptive optical systems, etc.关键词：terahertz metasurface;multifunctional;vortex beams;deflected beams;focused beams;vanadium dioxide55|15|0更新时间：2025-12-10
- 摘要：In this paper, a decoupling method for a ±45°-polarized crisscross-shaped dipole antenna array is studied based on a collinear antenna array with a standing wave ratio of less than 1.5 in the bandwidth range of 1.7 GHz to 2.3 GHz. The mutual coupling between two ±45°-polarized collinear crisscross-shaped dipole antennas can be decomposed into three coupling types: E-plane co-polarized coupling (E coupling), H-plane co-polarized coupling (H coupling) and cross-polarized coupling. Firstly, the E-plane single-polarized antenna array is studied, and a self-decoupling zero can be introduced by bending the antenna’s radiating arm vertically to a certain length. A vertical metal branch is then introduced between the two antennas to adjust the position of the self-decoupling zero. As the distance between the ends of the horizontal radiating arms is relatively close between two collinear horizontally arranged crisscross-shaped dipole antennas, the E coupling has great influence on the port isolation. The 45° port isolation and -45° port isolation between the two antennas can be improved simultaneously by bending the radiating arm and introducing vertical metal branch. To suppress the cross-polarized coupling between the 45° port and the -45° port, an additional U-shaped metal strip is introduced. By bending the radiation arms, combining the vertical metal branches and U-shaped metal bands, the isolation between the two crisscross-shaped dipole antennas can be increased from 16 dB to more than 24 dB. In addition, since parasitic structures such as metamaterials, dielectric layers, and multi-layer metasurfaces with large sizes and thicknesses do not need to be loaded above the array, the overall profile and volume of the antenna array do not increase significantly.关键词：antenna array;base-station antenna;decoupling;dipole antenna;dual-polarized antenna;self-decoupling41|33|0更新时间：2025-12-10
- 摘要：Turbo codes have been widely used in communication systems such as 3G (3rd Generation Mobile Communication Technology) and 4G (4th Generation Mobile Communication Technology). To improve the efficiency of channel coding, puncturing techniques are commonly employed in practical applications. Due to the absence of some parity bits, the blind recognition of punctured Turbo codes becomes more challenging, and there is currently limited research on interleaver identification for such codes. This paper addresses the identification of random interleavers in punctured Turbo codes by leveraging the concept of logarithmic conformity, and proposes utilizing the soft output viterbi algorithm (SOVA) to update a posteriori information, thereby compensating for the performance loss caused by the approximation in logarithmic conformity computation. Simulation results demonstrate that the proposed algorithm outperforms existing related algorithms in terms of performance. Moreover, both the logarithmic conformity and SOVA have relatively low computational complexity, making the proposed algorithm highly suitable for real-time applications.关键词：punctured Turbo code;random interleaver;logarithmic conformity;soft output viterbi algorithm;soft decision33|21|0更新时间：2025-12-10
- 摘要：Intelligent reflecting surface (IRS) aided communication is regarded as one of the most promising techniques in the future wireless networks. When electromagnetic interference exists, the system performance of IRS aided communication degrades significantly due to the fact that IRS can amplify both the desired signal and the interference signal. Ultra reliable and low latency communication (URLLC) is an important application scenario in the future wireless networks. The impact of electromagnetic interference on IRS aided short-packet communication (SPC), which accomplishes URLLC, was analyzed in this paper. Under the assumption that small-scale fading followed Rician distribution and IRS used the statistical channel state information (CSI) to adjust the phase shifts, the closed-form expression for average block error rate (BLER) at the user is firstly derived by using the central limit theorem (CLT), moment matching method, and SPC related theory. Then, an optimization algorithm based on bisection search algorithm is proposed to minimize the system transmission latency under the reliability constraint of the user. Finally, simulation is used to validate the theoretical analysis. The simulation results show that, under the user’s reliability constraint, the presence of electromagnetic interference obviously degrades the capability to lower the transmission latency by increasing the number of elements of IRS compared to the environment without electromagnetic interference.关键词：electromagnetic interference;intelligent reflecting surface (IRS);ultra-reliable and low-latency communication (URLLC);short packet communication (SPC);statistical channel state information (CSI)61|31|0更新时间：2025-12-10
- 摘要：Federated learning can build a distributed and secure computing environment at the cloud-edge-terminal for application scenarios with high data privacy protection and real-time requirements. As a cross-device distributed learning, client heterogeneity and privacy security are two critical issues. Firstly, under the conditions of client data heterogeneity and device heterogeneity, there are large differences in response speed and data distribution, which can lead to lag between clients and greatly affect the performance of federated learning. Secondly, in terms of privacy security, federated learning still has security problems such as single-point attack on the central server, untrustworthy clients, and inference attacks. In this paper, we designed a hierarchical federated learning system FATChain to solve the above problems. Firstly, for the problem of client heterogeneity, an efficient client selection mechanism is proposed to group the selected clients according to their response speeds, and cluster sampling based on the representative gradient is used for each group of clients to ensure that clients with unique data distributions are selected, and synchronous and asynchronous training are combined through hierarchical bridging, which reduces the pressure caused by global synchronization while solving the problem of data and device heterogeneity. At the same time, a weighted aggregation algorithm based on the influence function is designed to improve the aggregation weight of high-quality local models, to solve the problem that global accuracy is affected by the high weight of low-quality local models due to data heterogeneity, to accelerate the convergence of the global model, and to improve the accuracy of model training. Secondly, to address the privacy and security issues, the federated learning algorithm is combined with the blockchain to achieve decentralization and solve the problem of single-point attack. A poisoning attack detection module is set up in the system to filter out the unqualified local updates before aggregation, solving the problem of poisoning attack. And the approach that participant nodes in the blockchain grouping only upload the updates without generating the blocks is utilized, which effectively prevents the inference attack caused by the malicious participant. The analysis shows that the proposed federated learning system well achieves privacy security protection for all parties, while the performance is greatly improved compared to similar schemes with good scalability. And it is suitable for large-scale application scenarios with high requirements for privacy protection.关键词：blockchain;federated learning;privacy protection;cluster sampling;model aggregation;Cloud-Edge-Terminal59|48|0更新时间：2025-12-10
- 摘要：Reconfigurable intelligent surface (RIS) has the ability to reconstruct the wireless environment and is expected to provide a new paradigm for future communication networks. However, the widely studied centralized RIS has many limitations when served in multi-user systems, such as poor coverage ability and low multiplexing gain. To this end, we propose a distributed RIS deployment architecture based on Poisson point process (PPP) to enhance the communication performance of multi-user millimeter wave (mmWave) massive multi-input multi-output (MIMO) systems. Moreover, we design the active beamforming and passive beamforming based on the array responses of the maximum gain path in mmWave channels, and derive the closed-form expressions of the ergodic spectral efficiency to facilitate the analysis of the power scaling laws. The expressions show that the spectral efficiency is proportional to the number of transmitting/receiving antennas, the square of the number of elements on one RIS, and the density of distributed RISs. Simulation results verify the accuracy of the theoretical analyses, indicating that the RIS outperforms other active devices in terms of performance improvement.关键词：reconfigurable intelligent surface;mmWave communication;massive MIMO;Poisson point process;spectral efficiency61|26|0更新时间：2025-12-10
- 摘要：Medium Earth orbit synthetic aperture radar (MEO SAR) can achieve medium spatial resolution with a wide observable swath and short revisit time. The combination of MEO SAR illuminator and airborne SAR receiver could overcome the coverage limitations of traditional low Earth orbit (LEO) airborne BiSAR systems, demonstrating broad application prospects. However, due to the high orbital altitude, long propagation delay, and significant trajectory curvature of the MEO platform, the imaging geometry becomes more complex, and the conventional LEO-airborne bistatic SAR imaging algorithms become invalid. Although Geosynchronous Orbit (GEO) SAR systems also suffer from the phase errors induced by the “stop-go” assumption and trajectory curvature, their high altitude, low speed, and minimal acceleration allow the slant range variation to be approximated by second-order models. In contrast, MEO SAR has much higher velocity and acceleration, which aggravates high-order nonlinear phase errors, making conventional GEO slant range models inadequate for MEO scenarios. To address these challenges, this paper proposes an improved monostatic equivalent imaging algorithm for MEO-airborne bistatic SAR. This method establishes an MEO-airborne bistatic geometric model based on the orbit characteristics of MEO SAR, innovatively introduces an orbital curvature compensation factor, and proposes an improved “non-stop-go” equivalent slant range model. It overcomes the modelling limitations of traditional equivalent monostatic methods for high-maneuverability and curved trajectories. Furthermore, this paper employs series inversion to derive two-dimensional spectra while considering azimuth space-variant effects in the scene, ultimately developing an imaging algorithm suitable for MEO-airborne bistatic SAR. Experimental results demonstrate that compared with traditional methods, the proposed algorithm achieves peak sidelobe ratio, integrated sidelobe ratio, and resolution close to theoretical values for point targets. In extended target imaging of the proposed algorithm, the image entropy is reduced by 30% and contrast improved by 20% than the traditional equivalent monostatic method, effectively suppressing defocusing caused by MEO SAR’s complex orbital characteristics.关键词：medium-Earth-Orbit synthetic aperture radar (MEO SAR);improved monostatic equivalent algorithm;bistatic SAR;“non-stop-go” model;curved trajectory41|24|0更新时间：2025-12-10
- 摘要：In distributed electromagnetic countermeasure scenarios, our jamming transmissions directed against adversarial targets may inadvertently interfere with our own co-frequency electromagnetic systems. To address this issue, a distributed co-frequency jamming cancellation architecture and method are proposed. The method models the time-delay errors and residual frequency offsets between the distributed jamming nodes and the authorized receiving nodes, and solves for the model coefficients using the recursive least squares algorithm. This approach effectively accounts for high-dimensional channel and time-frequency variations. The reference jamming signal is then used for the reconstruction and cancellation of the distributed jamming. Simulation results demonstrate that under normalized timing errors of 0.2 and normalized residual frequency offsets of 10-4, the proposed distributed co-frequency jamming cancellation method achieves a 13.1 dB improvement in jamming cancellation capability over the single-node approach, with residual jamming approaching the noise floor.关键词：distributed jamming;jamming cancellation;regional symbiosis;spectrum management48|27|0更新时间：2025-12-10
PAPERS
- 摘要：Electromagnetic non-destructive testing (NDT) is a non-invasive method for detecting and evaluating the internal information of various media. It is widely used in various fields, including biomedical detection, localization of internal damage in buildings, and identification and imaging of targets within underground spaces and other media. During the testing process, reflections at the air-material under test (MUT) interface before entering the interior of the MUT. The reflected signals not only carry no useful information but also attenuate the total energy of the detection signal. Therefore, eliminating the reflections at the air-MUT interface through impedance matching can enhance the signal transmission efficiency, thereby effectively increasing the strength of the NDT echo signals. Metasurfaces, a novel type of impedance-matching structures, have been widely applied in NDT signal enhancement scenarios. This paper reviews the principles of impedance matching and its development and applications in various fields, including biomedical detection, signal transmission in and out of water bodies, indoor information penetration through walls, and enhanced detection of underground pipelines. It also summarizes the current requirements, challenges, and difficulties associated with the practical application of metasurface-based impedance matching layers. This work aims to provide references and insights for the advancement of impedance matching techniques in the field of NDT and signal transmission enhancement.关键词：non-destructive testing (NDT);Impedance matching;metasurfaces;signal enhancement;ground penetrating radar;target recognition84|40|0更新时间：2025-12-10
- 摘要：With the continuous development of quantum computing technology, various application systems relying on the three major functions of traditional public key cryptography (key agreement/digital signatures/public key encryption) will no longer be secure. In response to the quantum threat, international standardization organizations led by the United States NIST (National Institute of Standards and Technology) are actively soliciting and deploying the standardization work of post-quantum cryptography (PQC) algorithms, aiming to complete the migration from traditional public key cryptography algorithms to PQC algorithms before truly practical quantum computers emerge. Crystals-Dilithium is one of the lattice-based digital signature algorithms in the NIST-PQC standard, which features high security and fast computation speed, making it an important path to implement digital signature algorithms resistant to quantum attacks. This paper commences with the theoretical foundations of the mainstream Crystals-Dilithium digital signature algorithm. It delves into optimization methods for underlying key components and overall hardware architecture design, focusing on hardware resource optimization and performance enhancement. The paper contrasts and analyzes existing methods and outcomes, aiming to clarify the direction for subsequent research. It aspires to provide a robust reference for the design of post-quantum digital signature cryptographic chips that balance performance with hardware resource efficiency.关键词：post quantum cryptography;lattice-based cryptography;Crystals-Dilithium digital signature;Hardware implementation;optimization scheme59|22|0更新时间：2025-12-10
- 摘要：The blockchain has some scalability issues such as high-resource consumption and low-level throughput, which seriously affects the application of blockchain technology. Sharding technology provides a feasible solution to the scalability issues of blockchain. In this paper, the various scalability solutions based on logic architecture of blockchain are introduced firstly, then, the sharding technology is summarized from four aspects: sharding hierarchy, system operation, key problems, and functional components. The design of sharding blockchain is decomposed into nine functional components, on this basis, the existing works of sharding blockchain are analyzed from the perspective of functional components, and the details of these typical sharding solutions are presented. Finally, the current research challenges faced by sharding technology are discussed from the perspectives of security, performance, and balance. Also, the future research directions of development process and simulation are provided.关键词：sharding scheme;blockchain;functional component;scalability;logic architecture81|26|0更新时间：2025-12-10
SURVEYS AND REVIEWS
- Technical support is provided by Beijing Founder electronics co., LTD 京ICP备09064830号-19
京公网安备11010802024621 - It is recommended to read the content of this site in Chrome&IE9+. Please switch to extreme mode in browser 360.
- Cookies We use cookies to help provide and enhance our service and tailor content. By continuing, you agree to the use of cookies.
0