Table of Content

30 November 2020, Volume 38 Issue 6

Previous Issue    Next Issue

Communication Engineering

Three-Dimensional Tension Sensor of Fiber Bragg Grating

XIE Kai, TAN Tao, SI Xuezhen, Lü Zhongbin, REN Pengliang, QIU Chengjun, DUAN Chao, TIAN Ye, CHAI Quan, GAO Fei, ZHANG Jianzhong

2020, 38(6):  843-852.  doi:10.3969/j.issn.0255-8297.2020.06.001

Asbtract ( 448 )   PDF (16875KB) ( 80 )  

References | Related Articles | Metrics

A three-dimensional tension sensor based on fiber Bragg grating sensing technology is designed to monitor the stress state of substation bushing. In combination with finite element simulation and experiment, the three-dimensional tension sensor is implemented and signal demodulation is realized. It is shown that the tension sensitivities in three directions of the sensor are all between 0.01με·N^-1 and 0.02 με·N^-1, loading test results are basically consistent with the values of added load, and the average error of the sensor is 5.13% for the load of 0～600N. Therfore, the proposed tension sensor based on fiber Bragg grating is applicible in the measurement of three-dimensional tension with high accuracy.

Application of Optical Fiber Grating Technology in Edge Detection of Glass Curtain Wall

WANG Yongxiang, HE Haitao, WU Jun, YANG Qijiang, XU Donghua

2020, 38(6):  853-863.  doi:10.3969/j.issn.0255-8297.2020.06.002

Asbtract ( 496 )   PDF (17438KB) ( 246 )  

References | Related Articles | Metrics

Address to safety accidents problems of glass curtain wall such as glass drop induced by structural adhesive failure and aging or high wind pressure load, we put forward a kind of edge dynamic stress detection method based on fiber Bragg grating sensing technology. We construct a multimode coupling model between glass panel edge strain and structural glue for predicting the structure safety of glass curtain wall. By comparing and analyzing the simulation and experimental data of multimodal strain, it can be obtained that the quasi-distributed fiber Bragg grating can indicate the failure position of the structural glue, and evaluate and feedback the safety performance of the structural glue of glass curtain wall in advance. It is of practical value to timely replace insecure glass, reduce the falling accidence of glass curtain wall, reduce economic loss and improve the safety of glass curtain wall.

Distributed Vibration Sensing System Based on Optical Frequency Domain Reflectometry and Cross-Correlation Algorithm

LIU Xiao, CHE Qian, LI Xinyu, WEN Hongqiao

2020, 38(6):  864-870.  doi:10.3969/j.issn.0255-8297.2020.06.003

Asbtract ( 527 )   PDF (8422KB) ( 194 )  

References | Related Articles | Metrics

Based on characteristics of Rayleigh scattering light, one can achieve distributed measurements based on the correlation analysis of signals in the optical frequency domain reflectometry (OFDR) system. In this paper, we propose to apply overlapped sliding window in the subsections of signals and choose appropriate weighted function of generalized cross-correlation algorithm through experimental comparison in an OFDR distributed fiber-optic vibration sensing system based on cross-correlation algorithm. Experimental results show that by using the proposed technique, the OFDR performs with an improved positioning accuracy and a reduced false alarm rate. Furthermore, a distributed vibration sensor with the positioning accuracy of 0.247 m and the measurable vibration frequency ranging from 5 kHz to 50 kHz is demonstrated.

D2D Communication Relay Selection Algorithm Based on Game Theory

PENG Yi, ZHANG Shen, ZHU Hao, LI Qiqian

2020, 38(6):  871-881.  doi:10.3969/j.issn.0255-8297.2020.06.004

Asbtract ( 511 )   PDF (6626KB) ( 84 )  

References | Related Articles | Metrics

To solve the problem of poor communication quality and high probability of communication interruption when the distance between the source node and the destination node is too large in D2D communication, this paper presents an algorithm of D2D communication relay selection based on game theory. Based on distance loss, the algorithm firstly defines the location region of the relay, and then analyzes the forwarding ability of the candidate nodes considering the communication interruption caused by too low energy. The candidate relay sets are obtained through filtering, consider the relationships between the nodes at the same time, the interruption of communication is analyzed, and finally the optimal relay nodes are selected by game theory source nodes for data transmission. Simulation results show that compared with the random selection algorithm and the optimal relay selection algorithm based on channel state information, the proposed algorithm can effectively improve the coverage of D2D communication, reduce the interrupt probability of communication links, improve the overall throughput of the system and improve the stability of the relay system.

Cognitive Radio Spectrum Allocation Based on Crazy Adaptive Fish Swarm Algorithm

SU Huihui, PENG Yi, QU Wenbo

2020, 38(6):  882-889.  doi:10.3969/j.issn.0255-8297.2020.06.005

Asbtract ( 391 )   PDF (3730KB) ( 46 )  

References | Related Articles | Metrics

Facing the facts in cognitive radio networks that spectrum utilization cannot meet the demand of exponentially increasing communication, and artificial fish swarm algorithm maintains poor population diversity and weak global search ability, this paper propose an improved artificial fish swarm algorithm under the graph theory spectrum allocation model to obtain the optimal spectrum allocation under the network benefit function. Firstly, the field of visual and step size are adjusted adaptively to ensure the algorithm having strong global search ability of the algorithm in earlier stages and high convergence accuracy in later stages. Secondly, crazy operators are introduced into the food source to generate disturbances with increased population diversity. By comparing the total benefits of four algorithms in the same spectrum environment, and setting up control variable method separately for the available spectrum and cognitive users, we test the performance of the algorithm. Simulation experiment indicates that the improved artificial fish swarm algorithm has better global search capability and robustness compared with other three algorithms.

Signal and Information Processing

Hyperspectral Sparse Unmixing Based on Local Weighted Low-Rank Prior

HUANG Wei, WU Feiyang, SUN Le

2020, 38(6):  890-905.  doi:10.3969/j.issn.0255-8297.2020.06.006

Asbtract ( 370 )   PDF (18971KB) ( 72 )  

References | Related Articles | Metrics

In order to fully exploit the intrinsic properties of abundance maps and improve the sparse unmixing accuracy of hyperspectral image, a sparse unmixing method based on local weighted low rank prior is proposed. The low rank prior is mainly based on the fact that the local cubes in hyperspectral images have higher spatial correlation and spectral correlation. The weighted low rank prior can explore the low-dimensional structural features inherent in the local block, effectively suppress the noise and maintain the detailed structure of the data. In combination with the existing total variation regularization and collaborative sparse regularization, the proposed method shows improved ability to exploit the detailed structure of the abundance coefficients, local smoothness and row sparsity. Experimental results on the simulation data and real hyperspectral data show that the proposed method can better maintain the fine details of the data and improve the unmixing accuracy compared with other state-of-the-art methods.

Circular Marker Detection of Under-Exposed Images of Helicopter Blades Based on YOLOv3 and Watershed

ZHANG Yubin, XIONG Bangshu, OU Qiaofeng, HUANG Jianping, CHEN Yaofeng

2020, 38(6):  906-915.  doi:10.3969/j.issn.0255-8297.2020.06.007

Asbtract ( 619 )   PDF (16158KB) ( 237 )  

References | Related Articles | Metrics

A method of circular marker detection in under-exposed helicopter blade images based on YOLOv3 (you only look once) and watershed algorithm is proposed, aiming to improve detection adaptability, speed up the detection and obtain accurate position of circular markers. Firstly, the real under-exposed blade images are labeled for dataset making, on which the YOLOv3 network is trained. Secondly, the circular marker regions in blade images from testing dataset are detected by the trained YOLOv3 network. Thirdly, the traditional watershed marker detection method is improved, and the circular marker regions are used for the watershed transformation by the multi-threading technology in parallel, and the edge detection results of the circular marker are obtained. Finally, the circular markers are accurately located by the least square circle fitting and the method of removing the singular points. The proposed method is proved to be adaptable, fast and accurate by a number of experiments in many under-exposed helicopter blade images, and has been applied to circular marker detection in under-exposed helicopter blade images.

Research on Influence Mechanism of Joint Uncertainty of Bio-images on Change Detection Accuracy

CHAO Jian, ZHANG Huifang, XU Changjun, ZHANG Penglin

2020, 38(6):  916-923.  doi:10.3969/j.issn.0255-8297.2020.06.008

Asbtract ( 371 )   PDF (18872KB) ( 43 )  

References | Related Articles | Metrics

This paper aims at studying the influence of image uncertainty on change detection accuracy, and revealing a theoretical basis for improving change detection accuracy by suppressing uncertainty. Firstly, joint entropy is used to evaluate the joint uncertainty of a two-phase image. Then, based on spatial statistical correlation method, the relationship between the joint uncertainty and the indexes characterizing the change detection results' accuracy of the two-phase image is studied. Finally, according to the relationship, an effect model about the joint uncertainty and the accuracy of change detection of the two-phase image is established. Experimental results show that the joint uncertainty of bio-images performs a strong negative-correlation with the accuracy of change detection results in an influence mode of linear feature.

Vehicle Point Cloud Clustering Based on Contextual Feature and Graph Cut

LIU Yawen, ZHANG Ying

2020, 38(6):  924-935.  doi:10.3969/j.issn.0255-8297.2020.06.009

Asbtract ( 382 )   PDF (15147KB) ( 109 )  

References | Related Articles | Metrics

Reliable and accurate clustering of point cloud is the basis for subsequent high-precision analysis and interpretation of scene target. This paper presents a method for vehicle point cloud segment using supervoxel and graph-cut with contextual feature. Firstly, density-based spatial clustering of applications with noise (DBSCAN) is used to segment the point cloud data, and density-reachable supervoxels can be obtained. Secondly, spatial and attribute contextual features are introduced to describe the correlation between supervoxels and to define the weights of the edges of the graph model constructed by supervoxels. Finally, the optimal supervoxel clustering is obtained based on multi-label graph-cut optimization algorithm. Experimental results show that the proposed method has improved accuracy and performance on over segmentation in clustering.

Short-Term Prediction Model of Subway Passenger Flow Based on EMD Optimized NAR Dynamic Neural Network

MA Feihu, JIN Yichen, SUN Cuiyu

2020, 38(6):  936-943.  doi:10.3969/j.issn.0255-8297.2020.06.010

Asbtract ( 656 )   PDF (9776KB) ( 262 )  

References | Related Articles | Metrics

In order to forecast the subway passenger flow more accurately, a short-term prediction model of subway passenger flow based on empirical mode decomposition (EMD) optimization of nonlinear auto regressive (NAR) dynamic neural network is proposed. First, by analyzing subway passenger flow data, we find that the daily passenger flow performs in a certain change rule. Second, we select a time-based NAR dynamic neural network, which has excellent nonlinear dynamic fitting ability and feedback memory function. Furthermore, in order to reduce prediction errors and improve prediction accuracy, we use EMD empirical mode decomposition algorithm to optimize the NAR dynamic neural network prediction model. Simulation results show that the EMD-NAR neural network combined prediction model is well applicable for short-term prediction of subway passenger flow with high prediction result accuracy about 93%.

Customer Churn Prediction Method Based on Stacking Ensemble Learning

ZHENG Hong, YE Cheng, JIN Yonghong, CHENG Yunhui

2020, 38(6):  944-954.  doi:10.3969/j.issn.0255-8297.2020.06.011

Asbtract ( 429 )   PDF (4630KB) ( 164 )  

References | Related Articles | Metrics

The machine learning algorithm is used to predict the customer loss problem in business activities. Inspired by the idea of Bagging ensemble method, we proposed a Stacking ensemble learning based on bootstrap sampling. By multiple bootstrap sampling of the data set and adding attribute disturbance, multiple copies of the base classifier are trained with the data subset, and the decision result of the base classifier is determined by the vote of the corresponding copy of the base classifier. Experimental results show that the method we proposed in this paper has better performance than all base classifiers and the classical Stacking ensemble method of the same structure in terms of accuracy, precision rate and F1-score.

Index of Binary Heap for Transportation Network and Optimization of Shortest Path Algorithms

WANG Ya, REN Yan, XIA Linyuan

2020, 38(6):  955-965.  doi:10.3969/j.issn.0255-8297.2020.06.012

Asbtract ( 406 )   PDF (11964KB) ( 132 )  

References | Related Articles | Metrics

Shortest path finding of transportation network is one of the commonest application scenes for geographic information system network analysis. This paper improves Dijkstra and A^* algorithms based on binary heap index. It presents several optimal strategies to improve the efficiency of both algorithms. Firstly, it uses binary heap index to improve the access efficiency of storage structure in GIS networks. Then it simplifies the data types and operation types as keeping the precision of data calculation of algorithms. Additionally, it decreases searching space effectively by optimizing the estimation function of A^* algorithm. Experimental results show that the improvement makes Dijkstra algorithm up to 7 times and A^* algorithm up to 200 times faster than before.

Robust Watermarking Algorithm Based on Two-Level Singular Value Decomposition

CHEN Qing, XIA Lanting, BU Ying

2020, 38(6):  966-975.  doi:10.3969/j.issn.0255-8297.2020.06.013

Asbtract ( 526 )   PDF (16577KB) ( 56 )  

References | Related Articles | Metrics

The paper aims to propose a watermarking algorithm based on rotation stable region and two-level SVD decomposition to improve the imperceptibility and the robustness of watermarking systems. First, the stable region where pixels are approximately lossless is selected. Then, the region is transformed by using redundant discrete wavelet transform, and the low-frequency sub-block is divided into non-overlapping blocks. Next, each subblock is decomposed by using SVD, and the first row and the first column of singular values are extracted to construct a characteristic matrix, and the characteristic matrix is decomposed by using SVD for the second time. Finally, an appropriate scaling factor is selected to embed the watermarking into the singular values of the characteristic matrix. When extracting watermarking, Radon transform is used to correct the host image containing the watermarking information, and the watermarking information is extracted. Experiments with the proposed algorithm show high PSNR values of above 45 dB and high normalized correlation (NC) values of greater than 0.90. It is verified that the watermarking algorithm performs good imperceptibility and ability of anti-geometric attack and highly guaranteed embedding capacity of the watermark. The algorithm has high practical application value in digital copyright protection.

Text Steganography Based on Neural Machine Translation

YU Shuangsheng, YANG Zhongliang, JIANG Minyu, HUANG Yongfeng

2020, 38(6):  976-985.  doi:10.3969/j.issn.0255-8297.2020.06.014

Asbtract ( 574 )   PDF (6815KB) ( 242 )  

References | Related Articles | Metrics

Deep learning has promoted the development of natural language processing technology, and the information hiding methods based on text generation show great potential in this area. This paper proposes a text information hiding method based on neural machine translation, which embeds information in the process of generating translated text. The neural machine translation model uses a Beam Search decoder, which is used to obtain the candidate words in the sentence sequence in the translation process, and to encode the candidate words according to the probability ranking. Then, in the process of decoding and outputting target language texts, corresponding encoded candidate words are selected according to the binary bitstream of secret information, so as to realize information embedding at word level. Experimental results show that compared with the existing text information hiding methods based on machine translation, this method significantly improves the rate of information embedding, and shows good capability and security in anti-steganography performance.

Improved Method to Craft Universal Perturbations Based on Fast Feature Fool

WEI Jianjie, Lü Donghui, LU Xiaofeng, SUN Guangling

2020, 38(6):  986-994.  doi:10.3969/j.issn.0255-8297.2020.06.015

Asbtract ( 559 )   PDF (12841KB) ( 129 )  

References | Related Articles | Metrics

Although deep neural networks have been widely applied in recent years, they are readily fooled by adversarial input perturbations which are imperceptible to humans. Such vulnerability to adversarial attacks has imposed threats for system deployment in security-crucial setting, thus it is necessary to study the risky generation method of perturbations to boost the anti-risk capability. As a universal perturbation, fast feature fool (FFF) is an effective attacking method for visual tasks. Beyond solely mixing the convolutional layer's output irrespective of the input activation status, this paper improves the FFF method by maximizing the feature difference between the input image and corresponding adversarial image during which the contributions of multiple convolutional layers are weighted differently. Experimental results demonstrate that the improved FFF actually has obtained higher success attacking rate and stronger cross-model transfer ability than the original one.

Research on Cooperative Strategy of Automobile Sharing Service Alliance Based on Game Theory and Multi-agent

WEI Xiaochao, FAN Yuyao

2020, 38(6):  995-1005.  doi:10.3969/j.issn.0255-8297.2020.06.016

Asbtract ( 379 )   PDF (10678KB) ( 89 )  

References | Related Articles | Metrics

Based on evolutionary game and multi-agent simulation, the cooperative strategy of automobile sharing service alliance studied to strengthen alliance management and enhance the cooperation enthusiasm of alliance service providers. From the perspective of evolutionary game, a multi-agent simulation model of automobile sharing service alliance is designed, which considers the learning rules of history information of the service provider and other members in the alliance. The multi-agent simulation model of automobile sharing service alliance is constructed. In addition, the influence of different parameter settings, such as revenue, cost and penalty on the strategy of service provider alliance is studied. Based on simulation experiment, we propose strategies of strengthening alliance cooperation under different parameters, accordingly providing decision support for the cooperation strategy of automobile sharing service alliance.

Construction of Splitting Authentication Codes Using Strongly Partially Balanced t-Design

WANG Xiuli, CAO Miao, WANG Lina

2020, 38(6):  1006-1016.  doi:10.3969/j.issn.0255-8297.2020.06.017

Asbtract ( 335 )   PDF (2715KB) ( 140 )  

References | Related Articles | Metrics

In this paper, we propose a restricted strongly partially-balanced t-design in a finite two-dimension vector space, and construct a perfect splitting authentication code on the basis of this design. First, based on the theory of equations, a strongly partially-balanced t-design is constructed, and directly one type of splitting authentication codes is obtained. Second, one more type of authentication codes is generated by adding constraints to the above construction process. The two types of codes are proved perfect on splitting authentication by calculating their r-order probabilities of successful spoofing attacks respectively. Finally, the performances of two types of codes are analyzed. By simulations with a specific example, the rationality and validity of the construction approaches are verified. Compared with previous works, it is concluded that in this method, the successful probability of each order deception attack could reach the minimum by using large number of sources, and that the coding algorithm and simulation are easy to implement due to its simple theoretical basis. Therefore, the code constructed in this paper is competitive in terms of amount of transmitting information, security and practicability.