Online First

Online First Papers are peer-reviewed and accepted for publication. Note that the papers under this directory, they are posted online prior to technical editing and author proofing. Please use with caution.
Display Method:

Coming soon.....

Vibration-Based Fault Diagnosis for Railway Point Machines Using VMD and Multiscale Fluctuation-Based Dispersion Entropy
SUN Yongkui, CAO Yuan, LI Peng, XIE Guo, WEN Tao, SU Shuai
, Available online  , doi: 10.23919/cje.2022.00.075
Abstract(528) HTML (257) PDF(142)
As one of the most important railway signaling equipment, railway point machines undertake the major task of ensuring train operation safety. Thus fault diagnosis for railway point machines becomes a hot topic. Considering the advantage of the anti-interference characteristics of vibration signals, this paper proposes an novel intelligent fault diagnosis method for railway point machines based on vibration signals. A feature extraction method combining variational mode decomposition (VMD) and multiscale fluctuation-based dispersion entropy is developed, which is verified a more effective tool for feature selection. Then, a two-stage feature selection method based on Fisher discrimination and ReliefF is proposed, which is validated more powerful than single feature selection methods. Finally, support vector machine is utilized for fault diagnosis. Experiment comparisons show that the proposed method performs best. The diagnosis accuracies of normal-reverse and reverse-normal switching processes reach 100% and 96.57% respectively. Especially, it is a try to use new means for fault diagnosis on railway point machines, which can also provide references for similar fields.
Sparse Homogeneous Learning: A New Approach for Sparse Learning
SHI Jiajie, YANG Zhi, LIU Jiafeng, SHI Hongli
, Available online  , doi: 10.23919/cje.2023.00.130
Abstract(70) HTML (35) PDF(10)
Many sparse representation problems boil down to address the underdetermined systems of linear equations subject to solution sparsity restriction. Many approaches have been proposed such as sparse Bayesian learning. In order to improve solution sparsity and effectiveness in a more intuitive way, a new approach is proposed, which starts from the general solution of the linear equation system. The general solution is decomposed into the particular and homogeneous solutions, where the homogeneous solution is designed to counteract as many elements of particular solution as possible to make the general solution sparse. First, construct a special system of linear equations to link the homogeneous solution with particular solution, which typically is an inconsistent system. Second, the largest consistent sub-system are extracted from the system so that as many corresponding elements of two solutions as possible cancel each other out. By improving implementation efficiency, the procedure can be accomplished with moderate computational time. The results of extensive experiments for sparse signal recovery and image reconstruction demonstrate the superiority of the proposed approach in terms of sparseness or recovery accuracy with acceptable computational burden.
Research Article
Federated Offline Reinforcement Learning with Proximal Policy Evaluation
YUE Sheng, DENG Yongheng, HUA Xingyuan, WANG Guanbo, REN Ju, ZHANG Yaoxue
, Available online  , doi: 10.23919/cje.2023.00.288
Abstract(6) HTML (3) PDF(0)
Offline reinforcement learning (RL) has gathered increasing attention in recent years, which seeks to learn policies from static datasets without active online exploration. However, the existing offline RL approaches often require a large amount of pre-collected data and hence are hardly implemented by a single agent in practice. Inspired by the advancement of federated learning (FL), this paper studies federated offline reinforcement learning (FORL), whereby multiple agents collaboratively carry out offline policy learning with no need to share their raw trajectories. Clearly, a straightforward solution is to simply retrofit the off-the-shelf offline RL methods for FL, whereas such an approach easily overfits individual datasets during local updating, leading to instability and subpar performance. To overcome this challenge, we propose a new FORL algorithm, named MF-FORL, that exploits novel “proximal local policy evaluation” to judiciously push up action values beyond local data support, enabling agents to capture the individual information without forgetting the aggregated knowledge. Further, we introduce a model-based variant, MB-FORL, capable of improving the generalization ability and computational efficiency via utilizing a learned dynamics model. We evaluate the proposed algorithms on a suite of complex and high-dimensional offline RL benchmarks, and the results demonstrate significant performance gains over the baselines.
Multi-scale Binocular Stereo Matching Based on Semantic Association
ZHENG Jin, JIANG Botao, PENG Wei, ZHANG Qiaohui
, Available online  , doi: 10.23919/cje.2022.00.338
Abstract(16) HTML (8) PDF(0)
Aiming at the low accuracy of existing binocular stereo matching and depth estimation methods, this paper proposes a multi-scale binocular stereo matching network based on semantic association. A semantic association module is designed to construct the contextual semantic association relationship among the pixels through semantic category and attention mechanism. The disparity of those regions where the disparity is easily estimated can be used to assist the disparity estimation of relatively difficult regions, so as to improve the accuracy of disparity estimation of the whole image. Simultaneously, a multi-scale cost volume computation module is proposed. Unlike the existing methods, which use a single cost volume, the proposed multi-scale cost volume computation module designs multiple cost volumes for features of different scales. The semantic association feature and multi-scale cost volume are aggregated, which fuses the high-level semantic information and the low-level local detailed information to enhance the feature representation for accurate stereo matching. We demonstrate the effectiveness of the proposed solutions on the KITTI2015 binocular stereo matching dataset, and our model achieves comparable or higher matching performance.
Comparative Analysis of Noise Margin between Pure SET-SET and Hybrid SET-PMOS Inverters
ZHANG Fan, LIU Yi, WANG Yibo, WU Minghu, HU Sheng, DONG Youli
, Available online  , doi: 10.23919/cje.2023.00.287
Abstract(55) HTML (28) PDF(3)
Single-electron transistor (SET) is considered as one of the promising candidates for future electronic devices due to its advantages of low power consumption and high integration. The comparative analysis of SET-based inverters, especially the noise margin, is carried out. Pure SET-SET and hybrid SET-PMOS inverters are designed for investigation. The effects of SET supply voltage, junction resistance and junction capacitance on noise tolerance and power consumption of inverters are studied. For hybrid SET-PMOS inverters, the noise margin high (NMH) is less than 60 mV under various conditions, which may become the bottleneck of its application. For pure SET-SET inverters, both NMH and NML could reach 300 mV at a supply voltage of 0.8 V. The minimum power consumption of pure SET-SET and hybrid SET-PMOS inverters is 2.85 nW and 58 nW, respectively. The pure SET-SET inverters have greater noise tolerance and lower power consumption, which is more conducive to large-scale integration. When junction capacitance $ C_J $ = 0.0273aF and junction resistance $ R_T \ge $ 1M in SET-SET inverters at a supply voltage of 0.8 V, the NMH and NML are not significantly affected by the junction resistance and the noise margin fluctuates at 300 mV.
An Improved YOLOv7-tiny Algorithm for Vehicle and Pedestrian Detection with Occlusion in Autonomous Driving
SU Jian, WANG Fang, ZHUANG Wei
, Available online  , doi: 10.23919/cje.2023.00.256
Abstract(98) HTML (50) PDF(18)
Future transportation is advancing in the direction of intelligent transportation systems (ITS), where an essential part is vehicle and pedestrian detection. Due to the complex urban traffic environment, vehicles and pedestrians in road monitoring have different forms of occlusion problems, resulting in the missed detection of objects. We design an improved YOLOv7-tiny algorithm for vehicle and pedestrian detection under occlusion, with the following four main improvements. In order to locate the object more accurately, 1×1 convolution and identity connection are added to the 3×3 convolution, and convolution reparameterization is used to enhance the inference speed of the network model. In view of the complex road background and more interference, the Coordinate Attention was added to the connection part of backbone and neck to enhance the network’s capacity to detect the object and lessen interference from other targets. At the same time, before being sent to the detection head, Global Attention Mechanism is added to improve the accuracy of model detection by capturing three-dimensional features. Considering the issue of imbalanced training samples, we propose Focal CIOU Loss instead of CIOU Loss to become the bounding box regression loss, so that the regression process attention to high-quality anchor boxes. Experiments show that the improved YOLOv7-tiny algorithm achieves 82.2% map@0.5 in PASCAL VOC dataset, which is 2.8% higher than before the improvement. The performance of map@0.5:0.95 is 5.2% better than the previous improvement. The proposed improved algorithm can availably to detect partial occlusion objects.
Self-Decoupled Square Patch Antenna Arrays by Exciting and Using Mixed Electric/Magnetic Coupling between Adjacent Radiators
LIU Qianwen, ZHU Lei, LU Wenjun
, Available online  , doi: 10.23919/cje.2023.00.222
Abstract(53) HTML (26) PDF(7)
This article presents and develops a simple decoupling method for the planar square patch antenna arrays by virtue of mixed electric and magnetic coupling property. Since the resonant modes of TM10 and TM01 are a pair of degenerate modes in the square patch radiator which are intrinsically orthogonal, a superposed mode of them can be generated to possess consistent field distributions along all the four sides of the patch by adjusting the feeding position. By employing such superposed mode, the mutual coupling between two horizontally adjacent patch elements will become identical to that between two vertical ones, indicating an expected possibility that the complex 2-D decoupling problem in a large-scale antenna patch array can be effectively facilitated and simplified to a 1-D one. Subsequently, metallic pins and connecting strip are properly loaded in each square patch resonator, such that appropriate electric and magnetic coupling strengths can be readily achieved and thus the mutual coupling can get highly decreased. A 1×2 antenna array with an edge-to-edge separation of 1mm, which corresponding to 0.0117λ0, is firstly discussed, simulated, and fabricated. The measured results show that the isolation can be highly improved from 4 dB to 17 dB across the entire passband. In final, 1×3, 2×2, and 4×4 antenna array prototypes are constructed and studied for verification of the expansibility and feasibility of the proposed decoupling method to both linear and 2-D antenna arrays.
A Distributed Self-tallying Electronic Voting System Using the Smart Contract
YAO Jingyu, WANG Tao, YANG Bo, ZHANG Wenzheng
, Available online  , doi: 10.23919/cje.2023.00.233
Abstract(82) HTML (39) PDF(15)
For electronic voting(e-voting) with a trusted authority, the ballots may be discarded or tampered, so it is attractive to eliminate the dependence on the trusted party. An e-voting protocol, where the final voting result can be calculated by any entity, is known as self-tallying e-voting protocol. To the best of our knowledge, addressing both abortive issue and adaptive issue simultaneously is still an open problem in self-tallying e-voting protocols. Combining Ethereum blockchain with cryptographic technologies, we present a decentralized self-tallying e-voting protocol. We solve the above problem efficiently by utilizing optimized Group Encryption Scheme and standard Exponential ElGamal Cryptosystem. We use zero-knowledge proof and homomorphic encryption to protect votes’ secrecy and achieve self-tallying. All ballots can be verified by anyone and the final voting result can be calculated by any entity. In addition, using the paradigm of score voting and “1-out-of-K” proof, our e-voting system is suitable for a wide range of application scenarios. Experiments show that our protocol is more competitive and more suitable for large-scale voting.
An Ultra-wideband Doubler Chain With 43-65 dBc Fundamental Rejection in Ku/K/Ka Band
WANG Long, CHEN Jixin, HOU Debin, XU Xiaojie, LI Zekun, TANG Dawei, ZHOU Rui, QI Hao, XIANG Yu
, Available online  , doi: 10.23919/cje.2023.00.157
Abstract(97) HTML (48) PDF(9)
In this paper, a double-balanced doubler chain with >43-dBc fundamental rejection over an ultra-wide bandwidth in 0.13-μm SiGe BiCMOS technology is proposed. To achieve high fundamental rejection, high output power, and high conversion gain over an ultra-wideband, a double-balanced doubler chain with pre- and post-drivers employs a bandwidth broadening technique and a ground shielding strategy. Analysis and comparison of single-balanced and double-balanced doublers were conducted, focusing on their fundamental rejection and circuit imbalance. Three doublers, a passive single-balanced doubler, an active single-balanced doubler, and a passive double-balanced doubler, were designed to verify the performance and characteristics of the single- and double-balanced doublers. Measurements show that the proposed double-balanced doubler chain has about 15-dB better fundamental rejection, and over more than twice the relative bandwidth compared to the single-balanced doubler chain fabricated with the same process. Over an 86.9% 3-dB bandwidth from 13.4 to 34 GHz, the double-balanced doubler chain delivers 14.7-dBm peak output power and >43-/33-dBc fundamental/third-harmonic rejection. To the authors’ best knowledge, the proposed double-balanced doubler chain shows the highest fundamental rejection over an ultra-wideband among silicon-based doublers at millimeter-wave frequency bands.
AOYOLO Algorithm Oriented Vehicle and Pedestrian Detection in Foggy Weather
SU Jian, MAO Shiang, ZHUANG Wei
, Available online  , doi: 10.23919/cje.2023.00.280
Abstract(152) HTML (76) PDF(18)
In the context of complex foggy environments, the acquired images often suffer from low visibility, high noise, and loss of detailed information. The direct application of general object detection methods fails to achieve satisfactory results. To address these issues, this paper proposes a foggy object detection method based on YOLOv8n, named AOYOLO. The AOD-Net, a lightweight defogging network, is employed for data augmentation. Additionally, the ResCNet module is introduced in the backbone to better extract features from low-illumination images. The GACSP module is proposed in the neck to capture multi-scale features and effectively utilize them, thereby generating discriminative features with different scales. The detection head is improved using WiseIoU, which enhances the accuracy of object localization. Experimental evaluations are conducted on the publicly available datasets RTTS and synthetic foggy KITTI dataset. The results demonstrate that the proposed AOYOLO algorithm outperforms the original YOLOv8n algorithm with an average mean average precision (mAP) improvement of 3.3% and 4.6% on the RTTS and KITTI datasets, respectively. The AOYOLO method effectively enhances the performance of object detection in foggy scenes. Due to its improved performance and stronger robustness, this experimental model provides a new perspective for foggy object detection.
Weighted Linear Loss Large Margin Distribution Machine for Pattern Classification
LIU Ling, CHU Maoxiang, GONG Rongfen, LIU Liming, YANG Yonghui
, Available online  , doi: 10.23919/cje.2022.00.156
Abstract(110) HTML (54) PDF(18)
Compared with support vector machine, large margin distribution machine (LDM) has better generalization performance. The central idea of LDM is to maximize the margin mean and minimize the margin variance simultaneously. But the computational complexity of LDM is high. In order to reduce the computational complexity of LDM, a weighted linear loss LDM (WLLDM) is proposed. The framework of WLLDM is built based on LDM and the weighted linear loss. The weighted linear loss is adopted instead of the hinge loss in WLLDM. This modification can transform the quadratic programming problem into a simple linear equation, resulting in lower computational complexity. Thus, WLLDM has the potential to deal with large-scale datasets. The WLLDM is similar in principle to the LDM algorithm, which can optimize the margin distribution and achieve better generalization performance. The WLLDM algorithm is compared with other models by conducting experiments on different datasets. The experimental results show that the proposed WLLDM has better generalization performance and faster training speed.
Robust Regularization Design of Graph Neural Networks Against Adversarial Attacks Based on Lyapunov Theory
YAN Wenjie, LI Ziqi, QI Yongjun
, Available online  , doi: 10.23919/cje.2022.00.342
Abstract(189) HTML (97) PDF(23)
The robustness of graph neural networks (GNNs) is a critical research topic in deep learning. Many researchers have designed regularization methods to enhance the robustness of neural networks, but there is a lack of theoretical analysis on the principle of robustness. In order to tackle the weakness of current robustness designing methods, this paper gives new insights into how to guarantee the robustness of GNNs. A novel regularization strategy named Lya-Reg is designed to guarantee the robustness of GNNs by Lyapunov theory. Our results give new insights into how regularization can mitigate the various adversarial effects on different graph signals. Extensive experiments on various public datasets demonstrate that the proposed regularization method is more robust than the state-of-the-art methods such as $ L_1 $-norm, $ L_2 $-norm, $ L_{21} $-norm, Pro-GNN, PA-GNN and GARNET against various types of graph adversarial attacks.
Expression Complementary Disentanglement Network for Facial Expression Recognition
WANG Shanmin, SHUAI Hui, ZHU Lei, LIU Qingshan
, Available online  , doi: 10.23919/cje.2022.00.351
Abstract(194) HTML (96) PDF(19)
Disentangling facial expressions from other disturbing facial attributes in face images is an essential topic for facial expression recognition. Previous methods only care about facial expression disentanglement (FED) itself, ignoring the negative effects of other facial attributes. Due to the annotations on limited facial attributes, it is difficult for existing FED solutions to disentangle all disturbance from the input face. To solve this issue, we propose an expression complementary disentanglement network (ECDNet). ECDNet proposes to finish the FED task during a face reconstruction process, so as to address all facial attributes during disentanglement. Different from traditional reconstruction models, ECDNet reconstructs face images by progressively generating and combining facial appearance and matching geometry. It designs the expression incentive (EIE) and expression inhibition (EIN) mechanisms, inducing the model to characterize the disentangled expression and complementary parts precisely. Facial geometry and appearance, generated in the reconstructed process, are dealt with to represent facial expressions and complementary parts, respectively. The combination of distinctive reconstruction model, EIE, and EIN mechanisms ensures the completeness and exactness of the FED task. Experimental results on RAF-DB, AffectNet, and CAER-S datasets have proven the effectiveness and superiority of ECDNet.
Towards Semi-supervised Classification of Abnormal Spectrum Signals Based on Deep Learning
JIANG Tao, CHEN Wanqing, ZHOU Hangping, HE Jinyang, QI Peihan
, Available online  , doi: 10.23919/cje.2022.00.395
Abstract(159) HTML (77) PDF(35)
In order to cope with the heavy labor cost challenge of the manual abnormal spectrum classification and improve the effectiveness of existing machine learning schemes for spectral datasets with interference-to-signal ratios, we proposes a semi-supervised classification of abnormal spectrum signals (SSC-ASS), aimed at addressing some of the challenges in abnormal spectrum signal (ASS) classification tasks. A significant advantage of SSC-ASS is that it does not require manual labeling of every abnormal data, but instead achieves high-precision classification of ASSs using only a small number of labeled data. Furthermore, the method can to some extent avoid the introduction of erroneous information resulting from the complex and variable nature of abnormal signals, thereby improving classification accuracy. Specifically, SSC-ASS uses a memory AutoEncoder module to efficiently extract features from abnormal spectrum signals by learning from the reconstruction error. Additionally, SSC-ASS combines convolutional neural network and the K-means using a DeepCluster framework to fully utilize the unlabeled data. Furthermore, SSC-ASS also utilizes pre-training, category mean memory module and replaces pseudo-labels to further improve the classification accuracy of ASSs. And we verify the classification effectiveness of SSC-ASS on synthetic spectrum datasets and real on-air spectrum dataset.
Friendship Inference Based on Interest Trajectory Similarity and Co-occurrence
TIAN Junfeng, HOU Zhengqi
, Available online  , doi: 10.23919/cje.2022.00.363
Abstract(148) HTML (74) PDF(8)
Most of the current research on user friendship speculation in location-based social networks is based on the co-occurrence characteristics of users, however, statistics find that co-occurrence is not common among all users; meanwhile, most of the existing work focuses on mining more features to improve the accuracy but ignoring the time complexity in practical applications. On this basis, a friendship inference model named ITSIC is proposed based on the similarity of user interest tracks and joint user location co-occurrence. By utilizing MeanShift clustering algorithm, ITSIC clustered and filtered user check-ins and divided the dataset into interesting, abnormal, and noise check-ins. User interest trajectories were constructed from user interest check-in data, which allows ITSIC to work efficiently even for users without co-occurrences. At the same time, by application of clustering, the single-moment multi-interest trajectory was further proposed, which increased the richness of the meaning of the trajectory moment. The extensive experiments on two real online social network datasets show that ITSIC outperforms existing methods in terms of AUC score and time efficiency compared to existing methods.
YOLO-Drone: A Scale-Aware Detector for Drone Vision
LI Yutong, MA Miao, LIU Shichang, YAO Chao, GUO Longjiang
, Available online  , doi: 10.23919/cje.2023.00.254
Abstract(113) HTML (55) PDF(16)
Object detection is an important task in drone vision. However, since the number of objects and their scales always vary greatly in the drone-captured video, small object-oriented feature becomes the bottleneck of model performance, and most existing object detectors tend to underperform in drone-vision scenes. To solve these problems, we propose a novel detector named YOLO-Drone in this paper. In the proposed detector, the backbone of YOLO is firstly replaced with ConvNeXt, which is the state-of-the-art one to extract more discriminative features. Then, a novel scale-aware attention (SAA) module is designed in detection head to solve the large disparity scale problem. A scale-sensitive loss (SSL) is also introduced to put more emphasis on object scale to enhance the discriminative ability of the proposed detector. Experimental results on the latest VisDrone 2022 test-challenge dataset (detection track) show that our detector can achieve an AP of 39.43%, which is tied with the previous SOTA, meanwhile, reducing 39.8% of the computational cost.
An Integrated External Archive Local Disturbance Mechanism for Multi-objective Snake Optimizer
GAO Leifu, LIU Zheng
, Available online  , doi: 10.23919/cje.2023.00.023
Abstract(155) HTML (77) PDF(23)
It is an interesting research direction to develop new multi-objective optimization algorithms based on meta-heuristics. The convergence accuracy and population diversity of existing methods are not satisfactory. This paper proposes an integrated external archive local disturbance mechanism for multi-objective snake optimizer (IMOSO) to overcome the above two points. There are two improved strategies. The adaptive mating between subpopulations strategy introduces the special mating behavior of snakes with multiple husbands and wives into the original snake optimizer. Some positions are updated according to the dominated relationships between the newly created individuals and the original individuals. The external archive local disturbance mechanism is used to re-search partial non-inferior solutions with poor diversities. The perturbed solutions are non-dominated sorting with the generated solutions by the next iteration to update the next external archive. The main purpose of this mechanism is to make full use of the non-inferior solution information to better guide the population evolution. The comparison results of the IMOSO and 7 state-of-the-art algorithms on WFG benchmark functions show that IMOSO has better convergence and population diversity.
Lightweight Object Detection Networks for UAV Aerial Images Based on YOLO
LI Yanshan, WANG Jiarong, ZHANG Kunhua, YI Jiawei, WEI Miaomiao, ZHENG Lirong, XIE Weixin
, Available online  , doi: 10.23919/cje.2022.00.300
Abstract(158) HTML (78) PDF(22)
Existing high-precision object detection algorithms for UAV aerial images often have a large number of parameters and heavy weight, which makes it difficult to be applied to mobile devices. We propose three YOLO-based lightweight object detection networks for UAVs, named YOLO-L, YOLO-S and YOLO-M respectively. In YOLO-L, we adopt a deconvolution approach to explore suitable upsampling rules during training to improve the detection accuracy. Besides, the Convolution-Batch Normalization- SiLU activation function (CBS) structure is replaced with Ghost CBS to reduce the number of parameters and weight, meanwhile Maxpool maximum pooling operation is proposed to replace the CBS structure to avoid generating parameters and weight. YOLO-S greatly reduces the weight of the network by directly introducing CSPGhostNeck residual structures, so that the parameters and weight are respectively decreased by about 15% at the expense of 2.4% mAP. And YOLO-M adopts the CSPGhostNeck residual structure and deconvolution to reduce parameters by 5.6% and weight by 5.7%, while mAP only by 1.8%. The results show that the three lightweight detection networks proposed in this paper have good performance in UAV aerial image object detection task.
Miniaturized, Shared Electric and Magnetic Dipole, Pattern Diversity IoT Antenna for Sub-6 GHz Applications
WANG Zhan, DONG Yuandan
, Available online  , doi: 10.23919/cje.2023.00.058
Abstract(124) HTML (60) PDF(15)
By using a novel meta-resonator structure, a miniaturized, surface mountable, shared aperture, and hybrid electric/magnetic dipole pattern diversity antenna is proposed for Internet of Things (IoT) applications. By exploring the shared electric (E-dipole) and magnetic dipole (M-dipole) structures, a novel T-shaped split-ring resonator (SRR) with even and odd modes is presented and studied by current distributions and equivalent circuits. Broadside and omnidirectional (monopole-like) patterns are achieved by exciting the M-dipole and E-dipole modes of the T-shaped SRR, respectively. To validate the proposed design, this shared aperture metamaterial-inspired pattern diversity antenna with a small size of 0.29 λ0 × 0.006 λ0 × 0.11 λ0 is fabricated and measured. The measured overlapped -10 dB bandwidth is from 3.40 to 3.63 GHz (7.0 %, covering the LTE B42 band) and the port isolation is greater than 23 dB. Both two modes achieve a good radiation efficiency better than -0.79 dB (> 83.0 %).
A Polarization Control Operator for Polarized Electromagnetic Wave Designing
CUI Shuo, LI Yaoyao, ZHANG Shijian, CHEN Ling, CAO Cheng, SU Donglin
, Available online  , doi: 10.23919/cje.2022.00.410
Abstract(281) HTML (140) PDF(34)
To describe and control the polarization state of electromagnetic waves, a polarization control operator of the complex vector form is proposed. Distinct from traditional descriptors, the proposed operator employs an angle parameter to configure the polarization state of the polarized wave. By setting the parameter in the proposed operator, the amplitude of the field components can be modified, resulting in changes in the magnitude and direction of the field vector, and thus realizing control of the polarization state of the electromagnetic wave. The physical meaning, orthogonal decomposition, and discrete property of the proposed operator are demonstrated through mathematical derivation. In the simulation examples, the polarization control operator with fixed and time-varying parameters is applied to the circularly polarized wave. The propagation waveform, the trajectory projection and the waveform cross section in different reception directions of the new electromagnetic waves are observed. The results show that complex electromagnetic waves with more flexible polarization states can be obtained with the aid of the polarization operator.
IP-Pealing: A Robust Network Flow Watermarking Method Based on IP Packet Sequence
FENG Wangxin, LUO Xiangyang, LI Tengyao, YANG Chunfang
, Available online  , doi: 10.23919/cje.2022.00.366
Abstract(244) HTML (121) PDF(34)
Network flow watermarking (NFW) is usually used for flow correlation. By actively modulating some features of the carrier traffic, NFW can establish the correspondence between different network nodes. In the face of strict demands of network traffic tracing, current watermarking methods cannot work efficiently due to the dependence on specific protocols, demand for large quantities of packets, weakness on resisting network channel interferences and so on. To this end, we propose a robust network flow watermarking method based on IP packet sequence, called as IP-Pealing. It is designed to utilize the packet sequence as watermark carrier with IP identification field which is insensitive to time jitter and suitable for all IP based traffic. To enhance the robustness against packet loss and packet reordering, the detection sequence set is constructed in terms of the variation range of packet sequence, correcting the possible errors caused by the network transmission. To improve the detection accuracy, the long watermark information is divided into several short sequences to embed in turn and assembled during detection. By a large number of experiments on the Internet, the overall detection rate and accuracy of IP-Pealing reach 99.91% and 99.42% respectively. In comparison with the classical network flow watermarking methods, such as PROFW, IBW, ICBW, WBIPD and SBTT, the accuracy of IP-Pealing is increased by 13.70% to 54.00%.
New Related-Tweakey Boomerang Attacks and Distinguishers on Deoxys-BC
LIU Jiamei, TAN Lin, XU Hong
, Available online  , doi: 10.23919/cje.2022.00.383
Abstract(228) HTML (109) PDF(29)
Deoxys-BC is the primitive tweakable block cipher of the Deoxys family of authenticated encryption schemes. Based on the existing related-tweakey boomerang distinguishers, this paper improves the boomerang attacks on 11-round Deoxys-BC-256 and 13-round Deoxys-BC-384 by the optimized key guessing and the precomputation technique. It transfers a part of subtweakey guess in the key-recovery phase to the precomputation resulting in a significant reduction of the overall time complexity. For 11-round Deoxys-BC-256, we give a related-tweakey boomerang attack with time/data/memory complexities of $2^{218.6}/2^{125.7}/2^{125.7}$, and give another attack with the less time complexity of $2^{215.8}$ and memory complexity of $2^{120}$ when the adversary has access to the full codebook. For 13-round Deoxys-BC-384, we give a related-tweakey boomerang attack with time/data/memory complexities of $2^{k-96}+2^{157.5}/2^{120.4}/2^{113}$. For the key size $k=256$, it reduces the time complexity by a factor of $2^{31}$ compared with the previous 13-round boomerang attack. In addition, we present two new related-tweakey boomerang distinguishers on 11-round Deoxys-BC-384 with the same probability as the best previous distinguisher.
QARF: A Novel Malicious Traffic Detection Approach via Online Active Learning for Evolving Traffic Streams
NIU Zequn, XUE Jingfeng, WANG Yong, LEI Tianwei, HAN Weijie, GAO Xianwei
, Available online  , doi: 10.23919/cje.2022.00.360
Abstract(153) HTML (78) PDF(22)
In practical abnormal traffic detection scenarios, traffic often appears as drift, imbalanced and rare labeled streams, and how to effectively identify malicious traffic in such complex situations has become a challenge for malicious traffic detection. Researchers have extensive studies on malicious traffic detection with single challenge, but the detection of complex traffic has not been widely noticed. Queried adaptive random forests (QARF) is proposed to detect traffic streams with concept drift, imbalance and lack of labeled instances. QARF is an online active learning based approach which combines Adaptive Random Forests method and adaptive margin sampling strategy. QARF achieves querying a small number of instances from unlabeled traffic streams to obtain effective training. We conduct experiments using the NSL-KDD dataset to evaluate the performance of QARF. QARF is compared with other state-of-the-art methods. The experimental results show that QARF obtains 98.20% accuracy on the NSL-KDD dataset. QARF performs better than other state-of-the-art methods in comparisons.
Related-Key Zero-Correlation Linear Attacks on Block Ciphers with Linear Key Schedules
, Available online  , doi: 10.23919/cje.2022.00.419
Abstract(243) HTML (123) PDF(23)
Related-key model is a favourable approach to improve attacks on block ciphers with a simple key schedule. However, to the best of our knowledge, there are a few results in which zero-correlation linear attacks take advantage of the related-key model. We ascribe this phenomenon to the lack of consideration of the key input in zero-correlation linear attacks. Concentrating on the linear key schedule of a block cipher, we generalize the zero-correlation linear attack by using a related-key setting. Specifically, we propose the creation of generalized linear hulls (GLHs) when the key input is involved; moreover, we indicate the links between GLHs and conventional linear hulls (CLHs). Then, we prove that the existence of zero-correlation GLHs is completely determined by the corresponding CLHs and the linear key schedule. In addition, we introduce a method to construct zero-correlation GLHs by CLHs and transform them into an integral distinguisher. The correctness is verified by applying it to SIMON16/16, a SIMON-like toy cipher. Based on our method, we find 12/13/14/15/15/17/20/22-round related-key zero-correlation linear distinguishers of SIMON32/64, SIMON48/72, SIMON48/96, SIMON64/96, SIMON64/128, SIMON96/144, SIMON128/192 and SIMON128/256, respectively. As far as we know, these distinguishers are one, two, or three rounds longer than current best zero-correlation linear distinguishers of SIMON.
New Algebraic Attacks on Grendel with the Strategy of Bypassing SPN Steps
QIAO Wenxiao, SUN Siwei, HU Lei
, Available online  , doi: 10.23919/cje.2023.00.127
Abstract(188) HTML (93) PDF(40)
The rapid development of modern cryptographic applications such as zero-knowledge, secure multi-party computation, fully homomorphic encryption has motivated the design of new so-called arithmetization-oriented symmetric primitives. As designing ciphers in this domain is relatively new and not well-understood, the security of these new ciphers remains to be completely assessed. In this paper, we revisit the security analysis of arithmetization-oriented cipher Grendel. Grendel uses the Legendre symbol as a component, which is tailored specifically for the use in zero-knowledge and efficiently-varifiable proof systems. At FSE 2022, the first preimage attack on some original full GrendelHash instances was proposed. As a countermeasure, the designer adds this attack into the security analysis and updates the formula to derive the secure number of rounds. In our work, we present new algebraic attacks on GrendelHash. For the preimage attack, we can reduce the complexity or attack one more round than previous attacks for some instances. In addition, we present the first collision attack on some round-reduced instances by solving the constrained input/constrained output problem for the underlying permutations.
Method and Practice of Trusted Embedded Computing and Data Transmission Protection Architecture Based on Android
WANG Yichuan, GAO Wen, HEI Xinhong, DU Yanning
, Available online  , doi: 10.23919/cje.2022.00.196
Abstract(217) HTML (104) PDF(9)
In recent years, the rapid development of Internet technology has constantly enriched people’s daily life and gradually changed from the traditional computer terminal to the mobile terminal. But with it comes the security problems brought by the mobile terminal. Especially for Android system, due to its open source nature, malicious applications continue to emerge, which greatly threatens the data security of users. Therefore, this paper proposes a method of trusted embedded static measurement and data transmission protection architecture based on Android to reduce the risk of data leakage in the process of terminal storage and transmission. We conducted detailed data and feasibility analysis of the proposed method from the aspects of time consumption, storage overhead and security. The experimental results show that this method can detect Android system layer attacks such as self-booting of the malicious module and improve the security of data encryption and transmission process effectively. Compared with the native system, the additional performance overhead is small.
A Lattice-Based Method for Recovering the Unknown Parameters of Truncated Multiple Recursive Generators with Constant
YU Hanbing, ZHENG Qunxiong
, Available online  , doi: 10.23919/cje.2022.00.387
Abstract(217) HTML (107) PDF(17)
Multiple recursive generators with constant, the high-order extension of linear congruence generators, are an important class of pseudorandom number generators that are widely used in cryptography. The predictability of truncated sequences output by multiple recursive generators with constant that predicting the whole sequences by the truncated high-order bits of the sequences is a cryptographically crucial problem. This paper studies the predictability of truncated multiple recursive generators with constant. Given a few truncated digits of high-order bits output by a multiple recursive generator with constant, we first convert the multiple recursive generator with constant to multiple recursive generator and then adopt the method we proposed recently to recover the modulus, the coefficients, and the differences of initial state. In particular, we give an estimation of the number of truncated digits required for recovering the differences of initial state by using the expected norm of target vector. We prove by exponential sums that the number of truncated digits required for uniquely determining both the initial state and the constant is finite and give an upper bound. Extensive experiments confirm the correctness of our method.
Investigating the Effects of V2C MXene on Improving the Switching Stability and Reducing the Operation Voltages of TiO2-Based Memristors
, Available online  , doi: 10.23919/cje.2022.00.327
Abstract(177) HTML (89) PDF(26)
Three-atoms-type V2C MXene, an emerging class of transition metal carbides, has attracted tremendous attention in the fabrication of advanced memristive devices due to its excellent electrochemical properties. However, the inserted and behind physical effects of inserting V2C on traditional TiO2-based memristors have not been clearly explored. In this work, exhaustive electrical characterizations of the V2C/TiO2-based devices exhibit enhanced performance (e.g., improved switching stability and lower operating voltages) compared to the TiO2-based counterparts. In addition, the advantaged influences of the inserted V2C have also been studied by means of first-principles calculations, confirming that V2C MXene enables controllable internal ionic process and facilitated formation mechanism of the Ag conductive filaments. This work demonstrates a way to combine experimental and theoretical investigations to reveal the positive effects of introducing V2C MXene on memristor, which is beneficial for fabricating performance-enhanced memristors.
Energy-Efficient Driving Strategy for High-Speed Trains with Considering the Checkpoints
ZHANG Zixuan, CAO Yuan, SU Shuai
, Available online  , doi: 10.23919/cje.2022.00.174
Abstract(387) HTML (191) PDF(142)
With rising energy prices and concerns about environmental issues, energy-efficient driving strategies (EEDS) for high-speed trains have received a substantial amount of attention. In particular, energy-saving schemes play a huge role in reducing the energy and operating costs of trains. This article studies the EEDS of high-speed trains at a given time. A well-posed model is formulated, in which the constraints of the checkpoints, in addition to the speed limits, vehicle dynamics, and discrete control throttle, are first considered. For a given control sequence, the Karush-Kuhn-Tucker (KKT) conditions are used to obtain the necessary conditions for an EEDS. According to several key equations of the necessary conditions, the checkpoint constraints are satisfied. Some case studies are conducted based on the data of the Beijing-Shanghai high-speed line to illustrate the effectiveness of the proposed approach.
Special Issue:Multi-dimensional QoS Provision of Intelligent Edge Computing for IoT
Multi-dimensional QoS Evaluation and Optimization of Mobile Edge Computing for IoT: A Survey
HUANG Jiwei, LIU Fangzheng, ZHANG Jianbin
, Available online  , doi: 10.23919/cje.2023.00.264
Abstract(33) HTML (17) PDF(9)
With the evolvement of the Internet of things (IoT), mobile edge computing (MEC) has emerged as a promising computing paradigm to support IoT data analysis and processing. In MEC for IoT, the differentiated requirements on quality of service (QoS) have been growing rapidly, making QoS a multi-dimensional concept including several attributes, such as performance, dependability, energy efficiency, economic factors, etc. To guarantee the QoS of IoT applications, theories and techniques of multi-dimensional QoS evaluation and optimization have become important theoretical foundations and supporting technologies for the research and application of MEC for IoT, which have attracted significant attention from both academia and industry. This paper aims to survey the existing studies on multi-dimensional QoS evaluation and optimization of MEC for IoT, and provide insights and guidance for future research in this field. This paper summarizes the multidimensional and multi-attribute QoS metrics in IoT scenarios, and then several QoS evaluation methods are presented. For QoS optimization, the main research problems in this field are summarized, and optimization models as well as their corresponding solutions are elaborated. We take notice of the booming of edge intelligence in AI-empowered IoT scenarios, and illustrate the new research topics and the state-of-the-art approaches related to QoS evaluation and optimization. We discuss the challenges and future research directions.
An Intelligent Privacy Protection Scheme for Efficient Edge Computation Offloading in IoV
YAO Liang, XU Xiaolong, DOU Wanchun, Bilal Muhammad
, Available online  , doi: 10.23919/cje.2023.00.111
Abstract(59) HTML (31) PDF(7)
As a pivotal enabler of intelligent transportation system (ITS), internet of vehicles (IoV) has aroused extensive attention from academia and industry. The exponential growth of computation-intensive, latency-sensitive and privacy-aware vehicular applications in IoV result in the transformation from cloud computing to edge computing, which enables tasks to be offloaded to edge nodes (ENs) closer to vehicles for efficient execution. In ITS environment, however, due to dynamic and stochastic computation offloading requests, it is challenging to efficiently orchestrate offloading decisions for application requirements. How to accomplish complex computation offloading of vehicles while ensuring data privacy remains challenging. In this paper, we propose an intelligent computation offloading with privacy protection scheme, named COPP. In particular, an Advanced Encryption Standard-based encryption method is utilized to implement privacy protection. Furthermore, an online offloading scheme is proposed to find optimal offloading policies. Finally, experimental results demonstrate that COPP significantly outperforms benchmark schemes in the performance of both delay and energy consumption.
A Deep Deterministic Policy Gradient-based Method for Enforcing Service Fault-tolerance in MEC
LONG Tingyan, CHEN Peng, XIA Yunni, MA Yong, SUN Xiaoning, ZHAO Jiale, LYU Yifei
, Available online  , doi: 10.23919/cje.2023.00.105
Abstract(101) HTML (49) PDF(14)
Mobile edge computing (MEC) provides edge services to users in a distributed and on-demand way. Due to the heterogeneity of edge applications, it is a key challenge for service providers to deploy latency and resource-intensive applications on resource-constrained devices. This is especially true when underlying edge infrastructures are fault and error-prone. In this paper, we propose a fault tolerance approach named DFGP, for enforcing mobile service fault-tolerance in MEC. It synthesizes a generative optimization network (GON) model for predicting resource failure and a deep deterministic policy gradient (DDPG) model for yielding preemptive migration decisions. We show through extensive simulation experiments that DFGP is more effective in fault detection and guaranteeing quality of service (QoS), in terms of fault detection accuracy, migration efficiency, task migration time, task scheduling time, and energy consumption than other existing method.
SPECIAL ISSUE: Explainability, Robustness, and Security in AI Systems
Data Reconstruction Attacks Against Highly Compressed Gradients
XIANG Kunlan, YANG Haomiao, GE Mengyu, WANG Xiaofen, DAI Hongning
, Available online  , doi: 10.23919/cje.2022.00.457
Abstract(37) HTML (18) PDF(3)
Federated learning (FL) exchanges gradients instead of local training data and is therefore considered to protect data privacy. However, recent studies have shown that these gradients can be used to perform a data reconstruction attack (DRA). Nevertheless, none of these attacks can be applicable to highly compressed gradients, while most practical FL systems share highly compressed gradients for communication bandwidth reduction. In this work, we find that during the Top-K gradient compression, some rows of the fully-connected layer gradient with the same index as the ground-truth labels are larger in absolute value than the rest of the gradients, so they are not compressed (we call this phenomenon Label-gradient-remain as in the following). Building upon the Label-gradient-remain phenomenon, we introduce a DRA method termed highly compressed gradient leakage attack (HLA) designed specifically for highly compressed gradients. Especially, we design two new initialization methods for this attack, Init-Attribute and Init-Feature. Compared with the commonly used initialization method using noise, Init-Attribute can compensate for the information loss caused by high gradient compression, thus improving the effectiveness of DRA. Specifically, Init-Attribute first infers attributes from gradients and then finds the most similar data from the auxiliary dataset as the initialization data according to the inferred attributes. To ensure Init-Attribute works effectively, the auxiliary dataset requires extensive manual annotations, so we further develop Init-Feature, which generates initialization data directly by decoding gradients, thereby eliminating the need for manual annotation. Experiments on multiple benchmark datasets show that our proposed method is still effective even if 99.9% of the gradients are compressed to zero (i.e., a compression ratio of 0.1%).
ERSinAI2023+ Lightweight Steganography Detection Method Based on Multiple Residual Structures and Transformer
LI Hao, ZHANG Yi, WANG Jinwei, ZHANG Weiming, LUO Xiangyang
, Available online  , doi: 10.23919/cje.2022.00.452
Abstract(55) HTML (27) PDF(13)
Existing deep learning-based steganography detection methods utilize convolution to automatically capture and learn steganographic features, yielding higher detection efficiency compared to manually designed steganography detection methods. Detection methods based on convolutional neural network frameworks can extract global features by increasing the network’s depth and width. These frameworks are not highly sensitive to global features and can lead to significant resource consumption. This manuscript proposes a lightweight steganography detection method based on multiple residual structures and Transformer (ResFormer). A multi-residuals block based on channel rearrangement is designed in the preprocessing layer. Multiple residuals are used to enrich the residual features and channel shuffle is used to enhance the feature representation capability. A lightweight convolutional and Transformer feature extraction backbone is constructed, which reduces the computational and parameter complexity of the network by employing depth-wise separable convolutions. This backbone integrates local and global image features through the fusion of convolutional layers and Transformer, enhancing the network’s ability to learn global features and effectively enriching feature diversity. An effective weighted loss function is introduced for learning both local and global features, BiasLoss loss function is used to give full play to the role of feature diversity in classification, and cross-entropy loss function and contrast loss function are organically combined to enhance the expression ability of features. Based on BossBase-1.01, BOWS2 and ALASKA#2, extensive experiments are conducted on the stego images generated by spatial and JPEG domain adaptive steganographic algorithms, employing both classical and state-of-the-art steganalysis techniques. The experimental results demonstrate that compared to the SRM, SRNet, SiaStegNet, CSANet, LWENet, and SiaIRNet methods, the proposed ResFormer method achieves the highest reduction in the parameter, up to 91.82%. It achieves the highest improvement in detection accuracy, up to 5.10%. Compared to the SRNet and EWNet methods, the proposed ResFormer method achieves an improvement in detection accuracy for the J-UNIWARD algorithm by 5.78% and 6.24%, respectively.
Efficient Untargeted White-box Adversarial Attacks Based on Simple Initialization
ZHOU Yunyi, GAO Haichang, HE Jianping, ZHANG Shudong, WU Zihui
, Available online  , doi: 10.23919/cje.2022.00.449
Abstract(214) HTML (105) PDF(41)
Adversarial examples (AEs) are an additive amalgamation of clean examples and artificially malicious perturbations. Attackers often leverage random noise and multiple random restarts to initialize perturbation starting points, thereby increasing the diversity of AEs. However, given the non-convex nature of the loss function, employing randomness to augment the attack’s success rate may lead to considerable computational overhead. To overcome this challenge, we introduce the one-hot mean square error (MSE) loss to guide the initialization. This loss is combined with the strongest first-order attack, the projected gradient descent (PGD), alongside a dynamic attack step size adjustment strategy to form a comprehensive attack process. Through experimental validation, we demonstrate that our method outperforms baseline attacks in constrained attack budget scenarios and regular experimental settings. This establishes it as a reliable measure for assessing the robustness of deep learning models. Additionally, we explore the broader application of this initialization strategy in enhancing the defense impact of few-shot classification models. We aspire to provide valuable insights for the community in designing attack and defense mechanisms.
Knowledge Graph Completion Method of Combining Structural Information with Semantic Information
HU Binhao, ZHANG Jianpeng, CHEN Hongchang
, Available online  , doi: 10.23919/cje.2022.00.299
Abstract(23) HTML (12) PDF(9)
With the development of knowledge graphs, a series of applications based on knowledge graphs have emerged. The incompleteness of knowledge graphs makes the effect of the downstream applications and affected by the quality of the knowledge graphs. To improve the quality of knowledge graphs, translation-based graph embeddings, such as TransE, learn structural information by rep-resenting triples as low-dimensional dense vectors. However, is difficult to generalize to unseen entities, Entities that are not observed during training but appear during testing. The other methods use the powerful representational ability of pre-trained language models to learn entity descriptions and contextual representation of triples. Although they are robust to incompleteness, but they need to calculate the score of all candidate entities for each triple during inference. We consider combining two models to enhance the robustness of unseen entities by semantic information, and prevent combined explosion by reducing inference overhead through structured information. We use a pre-training language model to code triples and learn the semantic information within them, and then use a hyperbolic space-based distance model to learn structural information and integrate the two types of information together. We evaluate our model by performing link prediction experi-ments in standard datasets. In experiments, our model achieves better performances than state-of-the-art meth-ods on two standard datasets.
Joint Optimization of Trajectory and Task Offloading for Cellular-Connected Multi-UAV Mobile Edge Computing
XIA Jingming, LIU Yufeng, TAN Ling
, Available online  , doi: 10.23919/cje.2022.00.159
Abstract(422) HTML (211) PDF(44)
Since the computing capacity and battery energy of unmanned aerial vehicle (UAV) are constrained, UAV as aerial user is hard to handle the high computational complexity and time-sensitive applications. This paper investigates a cellular-connected multi-UAV network supported by mobile edge computing (MEC). Multiple UAVs carrying tasks fly from a given initial position to a termination position within a specified time. To handle the large number of tasks carried by UAVs, we propose a energy cost of all UAVs based problem to determine how many tasks should be offloaded to high-altitude balloons (HABs) for computing, which UAV-HAB association, the trajectory of UAV, and calculation task splitting are jointly optimized. However, the formulated problem has nonconvex structure. Hence, an efficient iterative algorithm by applying successive convex approximation (SCA) and the block coordinate descent (BCD) methods is put forward. Specifically, in each iteration, the UAV-HAB association, calculation task splitting, and UAV trajectory are alternately optimized. Especially, for the nonconvex UAV trajectory optimization problem, an approximate convex optimization problem is settled. The numerical results indicate that the scheme of this paper proposed is guaranteed to converge and also significantly reduces the entire power consumption of all UAVs compared to the benchmark schemes.
A Secure Communicating While Jamming Approach for End-to-End Multi-Hop Wireless Communication Network
MA Xiao, LI Dan, WANG Liang, HAN Weijia, ZHAO Nan
, Available online  , doi: 10.23919/cje.2022.00.448
Abstract(142) HTML (72) PDF(21)
With the rapid development of wireless communications, cellular communication and distributed wireless network is fragile to eavesdropping due to distributed users and transparent communication. However, to adopt bigger transmit power at a given area to interfere potential eavesdroppers not only incurs huge energy waste but also may suppresses regular communication in this area. To this end, we focus on secure communication in multi-hop wireless communication network, and propose two communicating while jamming schemes for secure communication in presence of potential eavesdroppers for the narrow band and broad band point-to-point (P2P) systems respectively with the aid of artificial noise transmitted by a chosen cooperative interferer. Furthermore, to achieve the end-to-end (E2E) multi-hop secure communication, we devise the secure network topology discovering scheme via constructing a proper network topology with at least one proper node as the cooperative interferer in each hop, and then propose the secure transmission path planning scheme to find an E2E secure transmission route from source to destination, respectively. Experiments on the wireless open-access research platform demonstrate the feasibility of the proposed schemes. Besides, simulations results validate that the proposed schemes can achieve better performance compared with existing methods in both the P2P communication case and E2E multi-hop communication network scenario.
Survey on Fake Information Generation, Dissemination and Detection
CUI Wanqiu, WANG Dawei
, Available online  , doi: 10.23919/cje.2022.00.362
Abstract(17) HTML (9) PDF(2)
The current booming development of the Internet has put the public in an era of information overload, in which false information is mixed and spread unscrupulously. This phenomenon has seriously disturbed the social network order. Thus, a substantial of research is beginning to be devoted to the effective management of fake information. We analyze the abnormal characteristics of fake information from its mechanism of generation and dissemination. In view of different exceptional features, we systematically sort out and evaluate the existing studies on false content detection. The commonly used public datasets, metrics, and performance are categorized and compared, hoping to provide a basis and guidance for related research. The study found that the current active social platforms show different novelty. The future direction should point to mining platform features of multi-domain sources, multi-data forms, and multi-language heterogeneity to provide more valuable clues for fake information.
Accepted Research Article for Publication
Enhanced Acceleration for Generalized Nonconvex Low-Rank Matrix Learning
ZHANG Hengmin, YANG Jian, DU Wenli, ZHANG Bob, ZHA Zhiyuan, WEN Bihan
, Available online  , doi: 10.23919/cje.2023.00.340
Abstract(72) HTML (36) PDF(10)
Matrix minimization techniques that employ the nuclear norm have gained recognition for their applicability in tasks like image inpainting, clustering, classification, and reconstruction. However, they come with inherent biases and computational burdens, especially when used to relax the rank function, making them less effective and efficient in real-world scenarios. To address these challenges, our research focuses on generalized nonconvex rank regularization problems in robust matrix completion (RMC), low-rank representation (LRR), and robust matrix regression (RMR). We introduce innovative approaches for effective and efficient low-rank matrix learning, grounded in generalized nonconvex rank relaxations inspired by various substitutes for the $\ell_0$-norm relaxed functions. These relaxations allow us to more accurately capture low-rank structures. Our optimization strategy employs a nonconvex and multi-variable alternating direction method of multipliers (ADMM), backed by rigorous theoretical analysis for complexity and convergence. This algorithm iteratively updates blocks of variables, ensuring efficient convergence. Additionally, we incorporate the randomized singular value decomposition (RSVD) technique and/or other acceleration strategies to enhance the computational efficiency of our approach, particularly for large-scale constrained minimization problems. In conclusion, our experimental results across a variety of image vision-related application tasks unequivocally demonstrate the superiority of our proposed methodologies in terms of both efficacy and efficiency when compared to most other related learning methods.
Research on Semantic Communication Based on Joint Control Mechanism of Shallow and Deep Neural Network
XIE Wenwu, XIONG Ming, REN Ziqing, WANG Ji, YANG Zhihe
, Available online  , doi: 10.23919/cje.2023.00.278
Abstract(82) HTML (44) PDF(7)
With the rapid development of deep learning, various semantic communication models are emerging, but the current semantic communication models still have much room for improvement in the coding layer. For this reason, a joint-residual neural networks (Joint-ResNets) framework based on the joint control of shallow neural networks (SNNs) and deep neural networks (DNNs) is proposed to cope with the problems in semantic communication coding. The framework synergizes SNNs and DNNs based on their shared utility, and uses variable weight $\alpha$ term to control the ratio of SNNs and DNNs to fully utilize the simplicity of SNNs and the richness of DNNs. The article details the construction of the Joint-ResNets framework and its canonical use in classical semantic communication models, and illustrates the control mechanism of the variable weight $\alpha$ term in the Joint-ResNets framework and its importance in balancing the model complexity between SNNs and DNNs. The article takes the task-oriented communication model in the device edge collaborative reasoning system as an example for experimentation and analysis. The experimental validation shows that DNNs and SNNs can be combined in a more effective way to standardize semantic coding, which improves the overall predictive performance, interpretability, and robustness of semantic communication models, and this framework is expected to bring new breakthroughs in the field of semantic communication.
Cooperative Self-Learning: A Framework for Few-Shot Jamming Identification
SHI Yuxin, LU Xinjin, SUN Yifu, AN Kang, LI Yusheng
, Available online  , doi: 10.23919/cje.2023.00.229
Abstract(34) HTML (18) PDF(2)
Jamming identification is the key objective behind effective anti-jamming methods. Due to the requirement of low-complexity and the condition of few labeled shots for a real jamming identification, it is very challenging to identify jamming patterns with high accuracy. To this end, we first propose a general framework of cooperative jamming identification among multiple nodes. Moreover, we further propose a novel fusion center (FC) aided self-learning scheme, which uses the guidance of the FC to improve the effectiveness of the identification. Simulations show that the proposed framework of the cooperative jamming identification can significantly enhance the average accuracy with low-complexity. It is also demonstrated that the proposed FC aided self-learning scheme has the superior average accuracy compared with other identification schemes, which is very effective especially in the few labeled jamming shots scenarios.
Distributed Cell-Free Massive MIMO versus Cellular Massive MIMO under UE Hardware Impairments
LI Ning, FAN Pingzhi
, Available online  , doi: 10.23919/cje.2023.00.045
Abstract(118) HTML (57) PDF(59)
This paper first investigates and compares the uplink spectral efficiency (SE) of distributed cell-free (CF) massive multiple-input multiple-output (mMIMO) and cellular mMIMO networks, both with user equipment (UE) hardware impairments. We derive a lower bound on the uplink ergodic channel capacity of the cellular mMIMO with UE hardware impairments, based on which we determine the optimal receive combining that maximizes the instantaneous effective signal-to-interference-and-noise ratio. Then, a lower bound on the uplink capacity of a distributed CF mMIMO with UE hardware impairments is derived using the use-and-then-forget technique. On this basis, the optimum large-scale fading decoding vector is found using generalized Rayleigh entropy. By using three combining schemes of minimum mean-square error (MMSE), regularized zero-forcing (RZF), and maximum ratio, the uplink SEs of distributed CF mMIMO and cellular mMIMO networks are analyzed and compared. The results show that the two-layer decoding distributed CF mMIMO network with MMSE combining outperforms the cellular mMIMO network, and the advantage is more evident as the hardware impairment factor increases. Finally, the uplink energy efficiency (EE) of the distributed CF mMIMO networks is analyzed and evaluated through the established realistic power consumption model with hardware impairments. Simulation results show that two-layer decoding provides higher SE and EE than single-layer decoding. In addition, RZF achieves almost the same SE and EE as MMSE in a two-layer decoding architecture.
Sharper Hardy Uncertainty Relations on Signal Concentration in terms of Linear Canonical Transform
XU Xiaogang, XU Guanlei, WANG Xiaotong
, Available online  , doi: 10.23919/cje.2023.00.096
Abstract(150) HTML (78) PDF(13)
Linear canonical transform is of much significance to optics and information science. Hardy uncertainty principle, like Heisenberg uncertainty principle, plays an important role in various fields. In this paper, four new sharper Hardy uncertainty relations on linear canonical transform are derived. These new derived uncertainty relations are connected with the linear canonical transform parameters and indicate new insights for signal energy concentration. Especially, for certain transform parameters, e.g. b=0, these new proposed uncertainty relations break the traditional counterparts in signal energy concentration, as will result in new physical interpretation in terms of uncertainty principle. Theoretical analysis and numerical examples are given to show the efficiency of these new relations.
A Fast Startup Crystal Oscillator with Digital SAR-AFC based Two-Step Injection
ZHOU Bo, LI Yifan, WANG Zuhang
, Available online  , doi: 10.23919/cje.2023.00.043
Abstract(93) HTML (46) PDF(28)
Crystal oscillators (XOs) provide a high-precision reference frequency but have a long startup time, which severely increases the average power consumption in duty-cycled systems. This paper proposes a fully-digital low-cost two-step injection (TSI) technique, by using a successive approximation register (SAR) based auto frequency control (AFC) loop, to speed up the startup behavior of XOs. A theoretical analysis is carried out to determine the optimum injection time and design low-power XOs. Fabricated in a 65-nm CMOS process, the proposed 12-MHz fast startup XO occupies an active area of 0.02 mm$ ^{2} $ and achieves a startup time less than 35 µs. The XO power consumption in the steady state is 40 µW from a 1.0-V supply, with a startup energy of 17.2 nJ.
Method of Single Event Effects Radiation Hardened Design for DC-DC Converter Based Load Transient Detection
GUO Zhongjie, LIU Nan, LU Hu, LI Mengli, QIU Ziyi
, Available online  , doi: 10.23919/cje.2022.00.442
Abstract(107) HTML (54) PDF(12)
Aiming at the impact of load current change on single-event transient, the essential difference between single-event transient and load transient of DC-DC converter is deeply studied. A hardened circuit based on load transient detection is proposed. The circuit detects the load transient information in time and outputs a control signal to control the single event hardened circuit, thereby realizing the improvement of the transient characteristics of the system under dynamic conditions. Based on the 180nm BCD process, the design and physical verification of a Boost converter are completed. The experimental results show that the input voltage range is 2.9~4.5V, the output voltage range is 5.8~7.9V, and the load current is 0~55 mA. During load transients, the load detection circuit turns off the hardened circuit in time, avoiding system oscillation and widening the dynamic range of the hardening circuit. Under the single event transient, the output voltage fluctuation of the system does not exceed the maximum ripple voltage, and the single event transient suppression ability reaches more than 86%, the system can work well with linear energy transient of about 100 MeV·cm2/mg.
SiC Double Trench MOSFET with Split Gate and Integrated Schottky Barrier Diode for Ultra-low Power Loss and Improved Short-circuit Capability
ZHANG Jinping, WU Qinglin, CHEN Zixun, ZOU Hua, ZHANG Bo
, Available online  , doi: 10.23919/cje.2022.00.394
Abstract(259) HTML (124) PDF(37)
A silicon carbide (SiC) double trench metal-oxide-semiconductor field effect transistor (DTMOS) with split gate (SG) and integrated Schottky barrier diode (SBD) is proposed for the first time. The proposed device features two enhanced deep trenches in the surface, in which a source-connected SG with a thicker dielectric layer is located at the bottom of the deep gate trench and an integrated SBD is located at the sidewall of the deep source trench (DST). Combined with shielding effect provided by the p+ shield layer under the DST and integrated SBD, the proposed structure not only reduces the reverse transfer capacitance ($ C $$ _{\rm rss} $) and gate-drain charge ($ {Q} $$ _{\rm gd} $) but also restrains the saturation drain current ($ {I} $$ _{\rm d, sat} $) and improves the diode performance of the device. Numerical analysis results show that compared with the Con-DTMOS and Con-DTMOS with external SBD diode, the turn-on loss ($ {E} $$ _{\rm on} $) and turn-off loss ($ {E} $$ _{\rm off} $) for the proposed device are reduced by 56.4%/70.4% and 56.6%/69.9%, respectively. Moreover, the $ {I} $$ _{\rm d, sat} $ at the $ {V} $$ _{\rm gs} $ of 18V for the proposed device is reduced by 74.4% and the short-circuit withstand time ($ {t} $$ _{\rm SC} $) is improved by about 7.5 times. As a result, an ultra-low power loss and improved short-circuit capability is obtained for the proposed device.
Detecting Double Mixed Compressed Images Based on Quaternion Convolutional Neural Network
WANG Hao, WANG Jinwei, HU Xuelong, HU Bingtao, YIN Qilin, LUO Xiangyang, MA Bin, SUN Jinsheng
, Available online  , doi: 10.23919/cje.2022.00.179
Abstract(221) HTML (111) PDF(30)
Detection of color images that have undergone double compression is a critical aspect of digital image forensics. Despite the existence of various methods capable of detecting double joint photographic experts group (JPEG) compression, they are unable to address the issue of mixed double compression resulting from the use of different compression standards. In particular, the implementation of joint photographic experts group 2000 (JPEG2000) as the secondary compression standard can result in a decline or complete loss of performance in existing methods. To tackle this challenge of JPEG+ JPEG2000 compression, a detection method based on quaternion convolutional neural networks (QCNN) is proposed. The QCNN processes the data as a quaternion, transforming the components of a traditional convolutional neural network (CNN) into a quaternion representation. The relationships between the color channels of the image are preserved, and the utilization of color information is optimized. Additionally, the method includes a feature conversion module that converts the extracted features into quaternion statistical features, thereby amplifying the evidence of double compression. Experimental results indicate that the proposed QCNN-based method improves, on average, by 27% compared to existing methods in the detection of JPEG+JPEG2000 compression.
A Recursive DRL-based Resource Allocation Method for Multibeam Satellite Communication Systems
MENG Haowei, XIN Ning, QIN Hao, ZHAO Di
, Available online  , doi: 10.23919/cje.2022.00.135
Abstract(290) HTML (143) PDF(33)
Optimization-based radio resource management (RRM) has shown significant performance gains on high-throughput satellites (HTSs). However, as the number of allocable on-board resources increases, traditional RRM are difficult to apply in real satellite systems due to its intense computational complexity. DRL is a promising solution for the resource allocation problem due to its model-free advantages. Nevertheless, the action space faced by DRL increases exponentially with the increase of communication scale, which leads to an excessive exploration cost of the algorithm. In this paper, we propose a recursive frequency resource allocation algorithm based on long-short term memory (LSTM) and proximal policy optimization (PPO), called PPO-RA-LOOP, where RA means resource allocation and LOOP means the algorithm outputs actions in a recursive manner. Specifically, the PPO algorithm uses LSTM network to recursively generate sub-actions about frequency resource allocation for each beam, which significantly cut down the action space. In addition, the LSTM-based recursive architecture allows PPO to better allocate the next frequency resource by using the generated sub-actions information as a prior knowledge, which reduces the complexity of the neural network. The simulation results show that PPO-RA-LOOP achieved higher spectral efficiency and system satisfaction compared with other frequency allocation algorithms.
MIMO Radar Transmit-Receive Design for Extended Target Detection against Signal-Dependent Interference
YAO Yu, LI Yanjie, LI Zeqing, WU Lenan, LIU Haitao
, Available online  , doi: 10.23919/cje.2021.00.140
Abstract(623) HTML (302) PDF(50)
Considering uncertain knowledge of target aspect angle (TAA), this paper copes with the joint optimization of transmit sequences and receive filter array for the detection of extended target in the presence of clutter disturbance. We consider joint transmit-receive design in multiple-input multiple-output (MIMO) structure to optimize the worst Signal to interference plus noise ratio (SINR) at the output of the receive filter array. Through a suitable reformulation, we propose a sequential optimization algorithm which monotonically enhances the worst SINR value. Each iteration of the process, includes a convex and a worst-case optimization problem which can be handled by the generalized Dinkelbachs method with a lower computational burden. In addition, resorting to several mathematical manipulations, the original problem is transformed into an equivalent convex problem, which can also be solved via interior-point techniques. Finally, the usefulness of two optimization techniques is confirmed through experimental simulation, emphasizing the detection capability improvement generated by the proposed approaches.
Gmean Maximum FSVMI Model and Its Application for Carotid Artery Stenosis Risk Prediction
ZHANG Xueying, GUO Yuling, LI Fenglian, WEI Xin, HU Fengyun, HUI Haisheng, JIA Wenhui
, Available online  , doi: 10.23919/cje.2020.00.185
Abstract(497) HTML (233) PDF(18)

Carotid artery stenosis is a serious medical condition that can lead to stroke. Using machine learning method to construct classifier model, carotid artery stenosis can be diagnosed with transcranial doppler data. We propose an improved fuzzy support vector machine (FSVMI) model to predict carotid artery stenosis, with the maximum geometric mean (Gmean) as the optimization target. The fuzzy membership function is obtained by combining information entropy with the normalized class-center distance. Experimental results showed that the proposed model was superior to the benchmark models in sensitivity and geometric mean criteria.