最新刊期
卷 52 ， 期 12 ， 2024
- 摘要：GaN has excellent material properties such as direct bandgap, high frequency, high power, and high electron mobility, making them have broad application prospects in fields such as power electronic devices and optoelectronic devices. Si substrates have advantages such as large size, low cost, and good process compatibility, so the GaN-on-Si has high research and commercial value. The crystal quality of GaN materials determines the performance of GaN based devices, but the crystal quality of GaN-on-Si is poor. So, researchers have proposed various methods to improve the crystal quality of GaN, but all of them have problems such as complex processes or high costs. Therefore, this research proposes a simple and low-cost induced nucleation technique to obtain high-quality Si based GaN materials. After completing the same dose of N ion implantation on Si substrate, rapid thermal annealing (RTA) treatment was carried out for different times, and epitaxial growth was carried out using metal organic chemical vapor deposition (MOCVD) method. The results showed that the sample with an annealing time of 6 minutes had the best crystal quality, surface morphology, and optical performance. Compared with the control sample, the screw dislocation density decreased by 14.7%, the edge dislocation density decreased by 34.4%, the total dislocation density decreased by 26.1%, and the two-dimensional electron gas (2DEG) surface density and electron mobility at the AlGaN/GaN heterojunction interface were improved, which improved the electrical performance of the AlGaN/GaN heterojunction.关键词：gallium nitride;induced nucleation;Rapid thermal annealing;metal organic chemical vapor deposition;dislocation density;optical performance;electrical performance68|14|1更新时间：2025-12-11
- 摘要：Current research on Chinese long text summarization based on deep learning has the following problems: (1) summarization models lack information guidance, fail to focus on keywords and sentences, leading to the problem of losing critical information under long-distance span; (2) the word lists of existing Chinese long text summarization models are often word-based and do not contain common Chinese words and punctuation, which is not conducive to extracting multi-grained semantic information. To solve the above problems, a Chinese long text summarization method with guided attention (CLSGA) is proposed in this paper. Firstly, for the long text summarization task, an extraction model is presented to extract the core words and sentences in the long text to construct the guided text, which can guide the generation model to focus on more important information in the encoding process. Secondly, the Chinese long text vocabulary is designed to changing the text structure from words statistics to phrases statistics, which is conducive to extracting richer multi-granularity features. Hierarchical location decomposition encoding is then introduced to efficiently extend location encoding of long text and accelerate network convergence. Finally, the local attention mechanism is combined with the guided attention mechanism to effectively capture the important information under the long text span and improve the accuracy of summarization. Experimental results on four public Chinese abstract datasets with different lengths, LCSTS, CNewSum, NLPCC2017 and SFZY2020, show that our proposed method has significant advantages over long text summarization and can effectively improve the value of ROUGE-1, ROUGE-2 and ROUGE-L.关键词：natural language processing;Chinese long text summarization;guided attention;hierarchical location decomposition coding;local attention35|15|0更新时间：2025-12-11
- 摘要：In the paper, a broadband and high gain metasurface antenna is proposed for meeting the demand of high performance antenna for C-band military equipment. The antenna uses the metasurface directly as the radiator and combines with the high-impedance-surface to achieve the high performance of broadband high gain with low profile and miniaturization. The equivalent circuit of the metasurface antenna is proposed. Characteristic mode analysis is applied to demonstrate the design process of an antenna. The processing cost, which antenna engineers concerned, is analyzed. The prototype antenna is processed and tested. The measured results show that the reflection coefficient is less than -10 dB and the peak gain is 9.9 dBi in the band from 5 GHz to 7.5 GHz (40%), and the measured results are in good agreement with the simulation results.关键词：characteristic mode;metasurface antenna;broadband;high gain;C band35|17|0更新时间：2025-12-11
- 摘要：Due to its large bandwidth and high resolution in the range dimension, microwave photonic radar enables finer information extraction for important targets such as ships through inverse synthetic aperture radar (ISAR) imaging, which is crucial for maritime surveillance. However, under the ultra-high resolution characteristics, the three-dimensional spatially variant Doppler parameters caused by target rotation can lead to image defocusing. In two-dimensional echo domain imaging, regional compensation processing is necessary, but existing methods cannot achieve adaptive regional segmentation, making it difficult to achieve two-dimensional ultra-high resolution imaging. To address these issues, this paper proposes a high-precision microwave photonic ISAR imaging method based on spatially variant Doppler parameter clustering. Firstly, the echo model of microwave photonic ISAR ship target is established, and the three-dimensional spatial variability characteristics of Doppler parameters are derived. The necessity of two-dimensional regional compensation processing is analyzed. Then, by separating strong scattering points, adaptive estimation and interpolation of Doppler parameters, a mapping relationship between each scattering point of the target and the two-dimensional Doppler parameter is established. Clustering processing in the two-dimensional Doppler parameter domain is performed to achieve adaptive optimal segmentation of spatially variant Doppler parameters, laying the foundation for high-precision regional compensation processing. Finally, regional non-spatially variant two-dimensional phase consistency compensation processing is carried out to achieve ultra-high resolution imaging of microwave photonic ISAR. The effectiveness of this method is validated through simulation and experimental data processing.关键词：microwave photonic radar;Inverse Synthetic Aperture Radar (ISAR);ship imaging;three-dimensional spatial variability;Doppler parameter clustering36|15|0更新时间：2025-12-11
- 摘要：In this paper, we present a fast polynomial multiplication algorithm, the hybrid-basis number theoretic transform (NTT) and inverse NTT (INTT) algorithms. These algorithms can efficiently implement polynomial multiplication based on finite domain using NTT conversion. On this basis, the paper explores the computational structure of fast polynomial multiplication algorithms. Without adding extra computational components, it optimizes network connectivity and proposes an energy-efficient reconfigurable hybrid-basis polynomial multiplication acceleration network. This network can flexibly implement base-2, base-3, and base-4 NTT/INTT algorithms, while doubling the operational efficiency of base-3 and base-4 NTT. This paper studies the issue of memory access conflicts in the computation process of hybrid-basis NTT. It theoretically analyzes the causes of these conflicts and, based on this analysis, proposes an energy-efficient hybrid-basis memory management scheme, designing the corresponding address generation logic. The proposed memory access scheme is a form of in-place memory access, and once implemented in hardware, it can still manage memory for different polynomial multiplication algorithms. Experimental results show that, under the 55 nm CMOS process, completing polynomial multiplication with a dimension of 256 and modulus less than 216 requires only 0.785 μs. The maximum operating frequency can reach 476 MHz, with a power consumption of 83.6 mW and an area time product (ATP) of 152.604 kGE·μs. Compared to the existing research, the ATP value of the proposed structure in this paper is reduced by more than 40%.关键词：post-quantum cryptography algorithm;lattice-based;polynomial multiplication;NTT12|13|0更新时间：2025-12-11
- 摘要：A fundamental theory of novel ultra-wide band (UWB) bandpass (BP) negative group delay (NGD) topology is established in this paper. The microwave circuit under study consists of lossy transmission lines and stepped impedance resonators. The flat NGD topology is constructed using fully distributed elements. The ABCD- and S-parameter models are formulated to derive the NGD optimal values and bandwidth. In order to verify the theoretical feasibility, NGD prototypes are designed, fabricated, and measured. The flat BP-NGD microstrip circuit has a compact size of 11 mm × 81 mm (0.13 λg × 1.01 λg) with a NGD center frequency of fn=2.14 GHz. Excellent agreement has been observed between experimental and theoretical results, revealing ΔfNGD=1.28 GHz (BWNGD=61%fn) NGD bandwidth and tn=-0.52 ns NGD value. Furthermore, within the NGD frequency band, the flat BP-NGD prototype presents a good performance in terms of bandwidth about ΔfNGD =1.01 GHz, BWflat-NGD =48%fn with tn±0.05 ns group delay fluctuation. Compared with similar broadband flat NGD circuits, the flat NGD bandwidth of the SIR NGD circuit proposed in this article is increased by about 215%. The flat BP-NGD prototype return loss at the center frequency is better than 18.8 dB.关键词：negative group delay;group delay flatness;stepped impedance resonator;S parameter;design equations;ultra-wide band14|15|0更新时间：2025-12-11
- 摘要：As an important part of signal parameter estimation, signal-to-noise ratio (SNR) estimation can provide prior information for power control, modulation classification, channel estimation, and dynamic mode switching, etc. Recently, high-order moments (HOMs) based algorithms have been widely concerned due to the advantages of low computational complexity and high real-time property. However, the estimation performance of the HOMs-based algorithms is still constrained in extremely low or high SNR regions. In this paper, a SNR estimation algorithm based on feature selection and linear combination of HOMs is designed, according to the distribution characteristics of HOMs. Firstly, the HOMs are screened by analyzing the relationship between different moments and SNR values. Based on this, we resort to the linear combination of the selected HOMs to estimate SNR. And the weights of linear combination are calculated by designing an optimization problem. The simulation results show that the proposed SNR estimation scheme makes a tradeoff among the estimation performance of high and low SNR regions. Compared with the existing HOMs-based algorithms, the proposed algorithm has a more comprehensive performance in the range of -10 dB to 20dB.关键词：estimation of signal-to-noise ratio;high-order moment;Nakagami-fading channel;feature selection;optimization of weights9|21|1更新时间：2025-12-11
- 摘要：A detection method based on interval autocorrelation coefficient (IAC) is proposed to address the difficulty of burst detecting frequency hopping/time hopping (FH/TH) communication signal under strong radar interference. On each hopping frequency sub channel, the maximum pulse width of the radar is used as the correlation length and interval unit, and two adjacent signals are extracted to calculate the IAC. In terms of the feature that the IAC of the FH/TH signal is much greater than the correlation coefficient of the radar pulse and random noise, the detection of the FH/TH signal can be achieved by confirming the burst-time and end-time of each sub channel, which are determined by point-by-point sliding and combined threshold method, and utilizing the cohesive relation of the burst time of each sub channel. The simulation results illustrate that, under the condition of strong interference, the proposed method can effectively overcome the radar interference and the burst detection probability of FH/TH signal is nearly close to 1. The real electromagnetic environment test results state that the method is less affected by interference of strong radar pulse, and is able to achieve relatively accurate detection.关键词：radar interference;FH/TH signal;burst detection;sub channel;interval autocorrelation coefficient;frequency hopping frequency;combination threshold19|19|1更新时间：2025-12-11
- 摘要：Considering the system sum rate decrease resulting from inter-user interference in the intelligent reflective surface (IRS)-assisted multi-user terahertz (THz) communications, a scheme combining sliding window and IRS is proposed to assist THz communications. Firstly, an array sliding window with an equal number of active users is constructed and connected to each active user in a one-to-one manner. Secondly, the mathematical expressions of the channel and rate provided by the sliding window to its corresponding associated active user are defined based on routing theory to derive the optimal active-passive beamforming and reformulate the beamforming problem into an equivalent graph optimization problem. Finally, given the constraints of the minimum user rate and total base station transmission power, alternating optimization algorithm is introduced to optimize sliding window parameters and maximize the system sum rate. Simulation results show that in the same channel scenarios, the system sum rate of our proposed scheme is increased by 5 bps/Hz compared to the conventional single IRS-assisted multi-user communications.关键词：intelligent reflecting surface;sliding window;terahertz;beamforming14|13|0更新时间：2025-12-11
- 摘要：To solve the problems of information transmission security caused by the dynamic openness of cognitive radio (CR), a beamforming optimization algorithm is proposed for STAR-RIS assisted CR system. Considering the constraints such as the transmit power of cognitive base station (CBS) and SU energy harvesting, variables such as SU power splitting (PS) coefficients are jointly optimized to maximize the SU worst secrecy rate. It is transformed into a convex problem by the SCA algorithm, and the coupling variables are solved by the SDR algorithm. The simulation results show that the proposed optimization algorithm has higher SU secrecy rate compared with others, and the introduction of artificial noise (AN) can further improve the security of the system.关键词：simultaneously transmitting and reflecting reconfigurable intelligent surface;cognitive radio;simultaneous wireless information and power transfer;physical layer security;secrecy rate21|20|0更新时间：2025-12-11
- 摘要：The combination of visible light communication (VLC) and non-orthogonal multiple access (NOMA) technology is an important method to meet the needs of indoor high speed communication and broadband data access. Due to the increasing demand for indoor dense communication, inter cell interference (ICI) and resource scarcity are becoming increasingly severe. In order to reduce the ICI, further enhance the quality of experience (QoE) of indoor user’s high-speed communication, and improve the spectrum resource utilization for VLC, a collaborative multi-point user access and power allocation algorithm is proposed for indoor VLC-NOMA. In the user access stage, a user access algorithm based on user needs of QoE and coordinated multi-point (UA-UEQCM) accessing VLC is designed. We can evaluate the QoE and equivalent mean opinion score (MOS) in terms of user’s satisfaction value, according to the minimum rate required by user, the recommended rate value required by user, and the achievable rate provided by VLC-NOMA. Then, based on the MOS user required and the available resource of coordinated access point in VLC-NOMA, a utility function of NOMA group is put forward to optimize the bidirectional selection between user equipment and coordinated multiple access points of VLC for improving MOS in VLC-NOMA network. In the power allocation stage, a power allocation algorithm by using improved teaching and learning (PA-ITL) is adopted to optimize the power allocation of the NOMA group. We design the total MOS value of VLC-NOMA as the fitness function value, and use adaptive updating and learning modes to optimize the NOMA group’s power allocation factor. The simulation results show that, compared with the comparison user access algorithm and power allocation, when the number of users is 26, the proposed user access algorithm and power allocation can improve the total MOS value by up to 18.06% and 8.60% for indoor VLC-NOMA networks, respectively.关键词：visible light communication;non-orthogonal multiple access;coordinated multi-point;user access;power allocation;quality of experience10|14|0更新时间：2025-12-11
- 摘要：Unmanned aerial vehicle (UAV) can be deployed as aerial base station (BS) or relays to provide wireless transmission services for ground user (GU) leveraging their advantages of low cost, high flexibility, and maneuverability. In scenarios where direct transmission between the BSs and the GUs may be unavailable, UAVs can be deployed as aerial relays which forward data packets for the GUs. In this paper, we address the UAV deployment and resource allocation strategies in a UAV-assisted communication system with the knowledge of statistical GU positions. We first formulate the joint UAV deployment, GU association and power allocation problem as a constrained average energy consumption minimization problem. To solve the formulated problem, we first propose a circle packing-based initial UAV deployment algorithm, then transform the original optimization problem into three subproblems, which are solved by applying an alternating iterative algorithm. Specifically, based on the given UAV deployment and GU association strategy, we propose a power allocation strategy by applying the Lagrange dual method. Additionally, given UAV deployment and power allocation strategy, the GU association strategy is designed iteratively based on Voronoi diagram. Furthermore, based on locally optimal power allocation and GU association strategy, we design the UAV deployment strategy by using quadratic transformation and the first-order Taylor expansion. The subproblems are solved iteratively until the algorithm reaches convergence, and the joint optimization strategy can be obtained. Simulation results demonstrate the effectiveness of the proposed algorithms.关键词：random user characteristics;UAV deployment;GU association;resource allocation;average system energy consumption25|15|0更新时间：2025-12-11
- 摘要：As a non-destructive electromagnetic detection technology, ground penetrating radar (GPR) has been widely used in municipal engineering, transportation, military and other fields. In the complex underground environment, the propagation law of electromagnetic wave becomes complicated, and the dielectric constant of background medium is difficult to be accurately obtained. Back projection (BP) imaging algorithm needs to predict the relative dielectric constant of the background media and calculate the scattering intensity of each imaging grid one by one, so the calculation efficiency is low. This paper puts forward the imaging method of deep learning based auto focusing BP (DABP). Firstly, a region of interest (ROI) detection module is designed. Based on the sparse space characteristics of underground targets, by combining YOLOX network and BP imaging mechanism, the potential target region is quickly detected, and only the imaging grid in the region is processed, which avoids the global back projection calculation and greatly reduces the amount of computation. Secondly, aiming at the problem that BP imaging is difficult to focus when the dielectric constant is unknown, an auto focusing BP (AF-BP) imaging module is designed, BS-YOLOv5 network and corresponding data set are constructed, and dielectric constant estimation and auto focusing imaging of underground media based on improved dichotomy are realized. Then, an artifact suppression based on double threshold and integral focusing (DTIF) module is designed to further improve the focusing degree of imaging results. Compared with BP imaging method, ISLR index of simulation data decreased by 250% and SCR index increased by 131%. ISLR index of measured data imaging results decreased by 322%, SCR index increased by 72%, and imaging speed of simulation experiment and measured experiment increased by 300%, which verified the effectiveness of the proposed method in improving GPR imaging efficiency and imaging quality.关键词：ground penetrating radar;deep learning;back projection;auto focusing imaging;artifact suppression13|16|0更新时间：2025-12-11
- 摘要：Sky-wave over-the-horizon radar (OTHR) effectiveness is limited by the operation environment. When the ionospheric state is bad or the operating parameters are unsuitable, the radar signal will not illuminate the scheduled area. Hence, the fact that the land-sea clutter (LSC) is normal or abnormal directly reflects the working status of OTHR. To address the scarcity and imbalance of OTHR clutter signals, a data enhancement method based on generative adversarial network is proposed for clutter range-Doppler image enhancement. A lightweight ResNet18 model is used for real-time identification of the radar images. Further, an LSC anomaly detector (LSCAD) is designed to achieve automatic identification of the radar LSC situation. The LSCAD extracts the high-amplitude region from the radar range-Doppler map, classifies it by the classification network based on the augmented dataset, and feds back to the radar operator. Simulation results show that the LSC data enhancement increases the LSC classifier accuracy by 25.26%. The LSCAD can make a correct judgement on the LSC status of the real data and literature images. Therefore, the LSCAD can be used as an extended module of the OTHR and provides automatic detection and warning about the LSC anomaly, which helps OTHR improving the degree of automation.关键词：sky-wave over-the-horizon radar;land-sea clutter;generative adversarial network;data enhancement;clutter detection36|16|0更新时间：2025-12-11
- 摘要：Information metamaterial is an artificial structure that can customize its equivalent material and media properties by designing unit parameters and arrangement, and realize free control of electromagnetic fields and electromagnetic waves, thereby bringing new physical phenomena. Information Metamaterial Aperture-based Microwave Computational Imaging (IMA-MCI) technology can achieve high-resolution imaging of targets within the beam without relying on the relative motion between the radar platform and the target. In microwave imaging, due to the limitations of the fabrication process of information metamaterial antennas, phase errors may be caused, and it is still challenging for IMA-MCI to reconstruct the target scene under the condition of phase error. To solve this problem, a microwave computational imaging model based on reflective information metamaterial antenna is constructed, and an imaging technology based on the combination of deep unfolding network and phase retrieval algorithm is proposed. Based on the phase retrieval algorithm, the algorithm introduces a dynamic super network to generate damping factors for the original network, and introduces a recurrent neural network, which can generate damping factors online according to the model, and still has good performance when the parameters of the system change. Experimental results show that the proposed method has good imaging performance and robustness.关键词：microwave computational imaging;information metamaterials;phase retrieval;deep learning;Hypernetwork10|20|0更新时间：2025-12-11
- 摘要：Accurately locating aviation targets in three dimensions presents formidable challenges owing to their high-speed movements and expansive maneuverability compared to ground targets. This paper presents a three-dimensional positioning algorithm for aerial targets based on a multi-sensor network. Using two high-altitude unmanned airships, each equipped with an optical sensor device, and UAV-shipborne bistatic dual-coordinate radars, the study focuses on achieving precise positioning of aerial targets through multi-platform collaboration. This approach addresses the limitations of traditional methods, which often fail due to the incomplete measurement dimensions of individual sensors, preventing accurate independent determination of the target’s 3D spatial position. Our approach begins with the introduction of a two-level point-trace correlation algorithm grounded in spatial alignment, leveraging both angle and distance to effectively correlate non-dimensional measurements from multiple sensors. Subsequently, we establish the initial positioning points of the target on each spatial measurement model through the construction of target guidance points, employing techniques such as ellipsoidal space Nelder-Mead Euclidean distance optimization and azimuth space projection. Finally, precise positioning points of the target are obtained through the utilization of unscented transform and homologous data compression.Simulation results demonstrate the efficacy of our algorithm in achieving stable correlation of radar-dual optical measurement data, even in scenarios where dimensionality is lacking. Notably, our algorithm achieves an optimal positioning error for aviation targets as low as 115.7 meters.关键词：multi-sensor networking;target positioning;trace points associated;space ellipsoid;Nelder-Mead algorithm;unscented transform24|13|0更新时间：2025-12-11
- 摘要：When small robots perform real-time positioning and mapping in complex environments, there are problems such as insufficient computing resources of the onboard CPU (Central Processing Unit), poor mapping accuracy, and low exploration efficiency. This paper proposes a real-time 3D reconstruction method based on simultaneous localization and mapping (SLAM) and truncated signed distance function (TSDF). This method obtains the RGB (Red Green Blue) image and depth image of the reconstructed target and scene based on a depth camera or a binocular camera, and obtains the pose information based on ORB_SLAM2. The surface reconstruction algorithm TSDF based on feature point cloud data is combined with the depth map to achieve a real-time 3D scene reconstruction. In order to reduce the error between the 3D reconstructed model and the real scene, a method of fusing feature points using light collision detection is proposed, and the error between the light projection distance and the distance from the voxel to the object surface is reduced by combining the optimization strategy. The integrity of the reconstructed scene is guaranteed by the optimized TSDF value. In ASL (Autonomous Systems Lab) the system is used in an open source dataset of the 3D reconstruction model of the proposed system. Compared with Voxblox, Voxfield and VDBblox, the root mean square error of the system's 3D reconstruction model is reduced by 55.6%, 47.11% and 21.7% respectively. Compared with Voxblox and Voxfield, the system map update rate is increased by 9.7% and 12.9% respectively. Finally, the system is used in indoor scene experiments, and the average map update rate is 7.35 ms/frame, which verifies the feasibility and effectiveness of the proposed system.关键词：3D reconstruction;TSDF;simultaneous localization and mapping;ORB_SLAM2;ray collision detection13|12|0更新时间：2025-12-11
- 摘要：Edge computing extends some tasks of center cloud server to the edge of the network, which can effectively alleviate the problems of high computation overhead and long processing latency caused by massive devices and data in the era of Internet of Everything. In edge computing environment, edge nodes and terminal devices are usually deployed in the unattended and open places, making them vulnerable to physical attacks while facing traditional security threats. To achieve secure communication in edge computing, several multi-receiver signcryption schemes with high communication efficiency have been proposed. However, there are still two issues with the application of existing schemes in areas with high security requirements: (1) no prevention mechanism for physical attacks is provided; (2) the anonymity protection for the senders has not been implemented. To fill this gap, we propose an efficient multi-receiver and multi-message signcryption scheme based on the hardware security primitive physical unclonable function (PUF) in this paper. Combining PUF with certificateless public key cryptography (CL-PKC) on elliptic curve, the proposed scheme does not need to use bilinear pairings with high computational complexity and is free from the key escrow problem. The security analysis shows that the scheme can effectively prevent physical attacks while possessing security attributes including confidentiality, unforgeability, and anonymity. Compared with related schemes, the proposed scheme achieves higher security with lower communication overhead without significantly increasing the computation overhead, which can better meet the requirements of secure communication in edge computing.关键词：edge computing;signcryption;Physical Unclonable Function (PUF);multi-receiver;multi-message;anonymity32|17|0更新时间：2025-12-11
- 摘要：With the widespread of smart contracts, the business logic has become more complex, causing a large number of security vulnerabilities. In order to avoid huge losses caused by potential vulnerabilities, a series of smart contract vulnerability detection methods were proposed. However, existing methods cannot comprehensively represent semantic and structural features of the contract, making it difficult to accurately detect potential vulnerabilities and security risks in smart contracts. To address this issue, this paper proposes a smart contract vulnerability detection method based on graph attention networks, named SCG-Detector (Smart Contract Graph Detector). Firstly, an abstract syntax tree (AST) is constructed by parsing the contract source code to represent the contract’s syntactic structure information. Data dependency relationships and control dependency relationships, which represent semantic information, are added to the AST to construct a smart contract graph (SCG) that characterizes the contract’s syntactic structure and semantic information. Secondly, using the SCG as input, the graph attention network model is trained with an attention mechanism to learn the features of vulnerabilities in the contract. Finally, the trained graph attention network model is used to detect whether there are vulnerabilities in the contract and the types of vulnerabilities present. Experiments are conducted on 12 616 smart contracts to compare with 8 widely used methods, including sFuzz, Conkas, ConFuzzius, Mythril, Osiris, Slither, Oyente, and MANDO-GURU. The experimental results shows that the
- 摘要：Addressing the issue of low spatial resolution in light field image, cross feature updating-based network for light field image super resolution is built in this paper to generate a higher quality array of light field sub-aperture images. In this work, a 3-branch structure is adopted for shallow feature extraction. In order to extract spatial and angular features from different forms of light field data, parallel residual block is designed. A cross feature update structure is used for extract deep features, and feature alignment interaction module, self-attention feature interaction module, and spatial feature enhancement module are designed to achieve step-by-step fusion and updating of spatial and angular features. In the data reconstruction part, the updated spatial information is fused by using multi-scale residual block and channel attention block alternately and finally super-resolution images are obtained through data upsampling. On the basis of fully exploring and supplementing spatial and angular features, the proposed network adopts a step-by-step fusion, update, and enhancement mechanism to gather spatial information at different levels, leading to superior super-resolution results. Comparative experiments demonstrate the effectiveness of the proposed method, with the network achieving an average peak signal-to-noise ratio(PSNR) value of 32.31 dB for 4× tasks on 5 public light field datasets, surpassing the performance of existing networks.关键词：4D light field;super-resolution reconstruction;cross feature updating;feature alignment interaction;self-attention feature interaction7|15|0更新时间：2025-12-11
- 摘要：Ensemble learning is an important branch and research hotspot in machine learning. The current main paradigm of ensemble learning algorithms is to obtain multiple sample subsets based on the original sample set, then to train the base classifiers separately and integrate the base classifier results. The main problem of this paradigm is that the diversity among subsets is significantly reduced since all subsets are derived from the original sample set. This problem is especially serious when the data size of the original sample set is small, the sampling ratio is large, and the degree of imbalance is high. In addition, the improvement in the divisibility of the sample subsets obtained by resampling is also limited when the divisibility of the original sample set is low. In order to solve this problem, this paper proposes a manifold nearest neighbor sample envelope and hierarchical multitype transformation algorithm for ensemble learning. It aims to improve the diversity and divisibility of the sample subset by transforming the original sample set into a hierarchical enveloped sample set with differentiation through the envelopment mechanism and the multitype sample transformation. First, the manifold nearest neighbor sample envelope mechanism is designed to transform the original samples into sample envelopes. Second, a multitype sample transformation is performed on the sample envelope to reconstruct and generate hierarchical envelope samples. Third, the inter-layer consistency preservation mechanism based on joint structure domain adaptation is designed to preserve the distribution consistency of the samples before and after the transformation. Thus, improving the high representation ability of the envelope samples to the original samples. Four, feature dimensionality reduction and basic classifier training are performed separately for each layer of the envelope sample set. Finally, the final classification results are obtained using the two dimensional decision fusion mechanism. More than ten datasets and several representative algorithms are used in the experimental part for validation. The results show that compared with the original sample set, the proposed algorithm improves the diversity of the sample subsets, which improves the ensemble learning performance with up to 18.56% accuracy improvement. Compared with related ensemble learning algorithms, the accuracy of this paper’s algorithm has been improved by up to 7.56%. This paper provides a new idea for the improvement of existing ensemble learning algorithms, and it is valuable to transform the paradigm of “ensemble learning directly based on original samples” into a new paradigm of “ensemble learning based on hierarchical envelope samples”.关键词：ensemble learning;envelope learning;sample transformation;nearest neighbor sample enveloping;domain adaptation;classifier ensemble15|16|0更新时间：2025-12-11
- 摘要：To solve the long-standing problems of the great scale variation in target sizes and blurred boundaries that make segmentation difficult in medical image segmentation, we propose a novel dual-branch hybrid network framework based on feature encoding and gated decoder based on multi-scale feature for accurate multi-organ segmentation. In order to fully exploit the strengths of convolutional neural network (CNN) in local information extraction and transformers in modeling long-range dependency, we employ U-Net and Swin-Unet to construct the dual-branch network. The innovation of this method lies in the shuffling operation of high-dimensional features extracted at multiple stages from different branches of the network. It efficiently integrates local and global information by means of a dual-branch channel cross-fusion, enhancing information interaction between the dual-branch network at different stages. This addresses the limitation in segmentation accuracy caused by the blurring of object contours in images. Additionally, to address the challenge of great scale variation among multiple organs, we introduce a new gated decoder based on multi-scale feature (GDMF) to extract multi-scale high-dimensional features at different stages of the network and perform adaptive feature enhancement, and adopts the attention mechanisms and feature mappings to assist in acquiring accurate target information. The experimental results on automated cardiac diagnosis challenge (ACDC) and fast and low GPU memory abdominal organ segmentation challenge 2021 (FLARE21) datasets demonstrate that our proposed method outperforms existing mainstream medical image segmentation methods and effectively solves the problems of the great scale variation in target sizes and blurred boundary in medical images.关键词：medical image segmentation;CNN-Transformer architecture;shuffle feature encoding;gated decoding9|15|0更新时间：2025-12-11
- 摘要：Continuous electrocardiogram (ECG) monitoring is crucial for effectively preventing and diagnosing cardiovascular diseases. However, existing ECG monitoring methods are limited by their reliance on expensive equipment unavailable to common users, the stringent requirements of the monitoring process, and confined application scenarios, making them insufficient to meet the urgent need for long-term continuous ECG monitoring of the general population in their daily lives. Given these limitations, this study proposes a motion-robust ECG signal sensing method based on modified non-negative matrix factorization (NMF). The basic idea is to leverage a gyroscope embedded into a low-cost wrist-worn wearable to characterize cardiac activities encoded into body vibrations and interpret them to generate fine-grained ECG signals accurately. As eliminating body motion interference is inherently hard, this work innovatively employs modified NMF to tackle the problem; this can effectively handle body motion interference, even if untrained, and extract the cardiogenic body vibrations from noisy gyroscope data. Due to the lack of clear pattern of cardiogenic body vibrations in each cardiac cycles, current cardiac cycle segmentation solutions cannot be applied. Thus, this work deeply analyses the morphological features of cardiogenic body vibrations and utilizes machine learning techniques for the identification of spike points for segmentation. Finally, cycle generative adversarial network (CycleGAN) framework is employed to construct a correlation mapping model between the cardiogenic body vibrations and the ECG signals. With innovative construction, this model can accurate generation of the ECG signals without the need for a huge amount of training data. Extensive experiments with 18 volunteers confirm the effectiveness of the proposed method, with the average amplitude errors of 7.92% and 9.02% for stationary and moving scenarios, respectively. These values fall well within the acceptable range of medical standards for error tolerance of less than 10%.关键词：electrocardiogram;wrist-worn devices;cardiogenic body vibrations;non-negative matrix factorization;cycle generative adversarial network19|12|0更新时间：2025-12-11
- 摘要：Yaw stability is an important problem in the stability study of distributed drive electric vehicles (DDEV). In order to solve the problems of complex structure and strong coupling in yaw stability, a control system based on multi-objective parallel chaos optimization is proposed for DDEV in this paper. This control system consists of two parts. The upper controller employs a multi-objective optimization strategy, leveraging a multi-objective parallel chaotic optimization algorithm, to determine the optimal yaw rate and desired slip rate essential for maintaining lateral stability in a distributed drive electric vehicle. The lower controller is execution unit, two fuzzy logic controllers are adopted to correct front wheel steering angle and distribute drive/brake torque respectively, according to the optimal desired variables computed by the upper controller. The modeling and simulation process is completed on the Matlab/Simulink platform, and the results show that the control system optimally coordinates the active steering angle of the front wheels and the driving/braking torque to ensure the lateral stability of the DDEV.关键词：electric vehicle;multi-objective optimization;fuzzy control;distributed drive;yaw stability24|12|0更新时间：2025-12-11
PAPERS
- 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