Current Issue

• 2020, No. 07　Published:25 July 2020 Last issue    Next issue
  Columns in this issue:

• Analysis and optimization of power consumption  characteristics of Network-on-Chip

  SUN Xiao-le, QIAN Ya-long, QI Xin-xin, ZHANG Yun-fang, CHEN Juan, YUAN Yuan, DONG Yong

  2020, 42(07): 1141-1150. doi:

  Abstract ( 187 )   PDF (934KB) ( 171 )   　　

  With the increase of the number of processor cores, the structure of Network-on-Chip (NoC) is becoming more and more complex, which leads to an increase in the power consumption of NoCs. The difficulty of analyzing the power consumption of NoCs is also increasing. Task mapping of NoCs not only ensure the high performance of the communication between multi-processor cores, but also ensure that the power consumption and area are as small as possible. That is to say, high performance is achieved under limited power consumption and area overhead. During task mapping, the communication distance between cores is the key to reduce the power consumption of task communication. Continuous and near-convex areas help reduce the communication distance of the task. This paper analyzes a power-optimized NoC heuristic mapping algorithm (INC), which is composed of region selection algorithm and node mapping algorithm. The two factors of the region selection algorithm are improved to minimize the total communication energy consumption in the system and ensure that the subsequent applications do the region selection at the minimum communication cost. Therefore, a new task mapping algorithm based on region selection is proposed. They are all used in dynamic application mapping. The experimental results show that the new region selection algorithm and the node mapping algorithm can reduce the power consumption by 12.10%, and optimize the communication latency by 11.23%, in comparison to INC.
  
  

  Microservice performance simulation test based on Kubemark

  LEI Qing

  2020, 42(07): 1151-1157. doi:

  Abstract ( 261 )   PDF (800KB) ( 156 )   　　

  Virtualization technology has greatly accelerated the expansion of applications on the microservice architecture. As the complexity of these applications continues to increase, microservice performance may be significantly different from the expected. Therefore, the methods and evaluation criteria of microservice performance testing have become a topic that scholars have begun to explore. This paper draws on the testing methods and evaluation standards of Web service quality, and uses the simulation testing methods in the experiments. The Kubemark tool is used to test the performance of the microservice system on the Kubernetes platform, and the p-percentile indicator in RFC 2679 standard is used to analyze the results. Experimental results show that the performance of microservices is significantly affected by the type of load microservices, and simulation testing is an effective research method for microservices performance testing.

  Application of time-constrained Louvain algorithm in modularization of dynamic brain function network

  DAN Yang-chao, WANG Bin, XUE Jie, SHENG Jing-ye, LIU Chang, ZHANG Wei-wei

  2020, 42(07): 1158-1167. doi:

  Abstract ( 183 )   PDF (1464KB) ( 77 )   　　

  In the study of modularization attributes of dynamic brain function network, Louvain algorithm maximizes the modularity value, which leads to low recognition of dynamic brain function network modules. For solving this problem, a time-constrained Louvain algorithm is proposed. In this algorithm, the iterative end condition for time constraint model is constructed based on the distribution of modularity values on the entire data acquisition time period. With this time constraint condition, the ba- lance of both scale and quantity of modules can be achieved, which will be helpful to ensure the reasonability of modules. When this method is used in the module partitioning experiment of resting-state brain function network, the experiment results show that, compared with Louvain algorithm, the time- constrained Louvain algorithm can provide a more reasonable modularity recognition result. Furthermore, the dynamic brain function network modularity comparison experiments with this method between healthy people and autistic patients are carried out, and the results prove that there are significant differences in the modularization features between these two groups, which can verify the effectiveness of this method. 
  

  
  

  Cloud computing resource load prediction  based on combined prediction model

  LIN Tao, FENG Jing-kai, HAO Zhang-xiao, HUANG Shao-qun

  2020, 42(07): 1168-1173. doi:

  Abstract ( 243 )   PDF (697KB) ( 159 )   　　

  With the continuous development of cloud computing technology, cloud computing resource load changes exhibits more and more complex features. For the workload prediction problem of cloud computing resources, the linear and nonlinear characteristics of resource workload time series in cloud computing environment are considered comprehensively. This paper proposes a combined prediction model based on auto-regressive integrated moving average (ARIMA) and long short-term memory (LSTM). The experiments are carried out to compare the proposal and the traditional load prediction algorithm on the public dataset. The experimental results show that the cloud computing resource combination prediction model has significantly higher prediction accuracy than other prediction models, which significantly reduces the real-time prediction error of resource workload in the cloud environment.
  

  
  

  Conditional decoupling design framework and performance analysis of dual-connectivity NOMA heterogeneous networks

  JI Peng-shan, JIA Xiang-dong, LU Yi, XU Wen-juan

  2020, 42(07): 1174-1183. doi: 10.3969/j.issn.1007-130X.2020.07.005

  Abstract ( 151 )   PDF (1051KB) ( 87 )   　　

  In order to improve the uplink performance of heterogeneous networks and coverage probability, this paper proposes a non-orthogonal multiple access (NOMA) heterogeneous network model by integrating decoupled uplink/downlink association (DUDA) and dual-connectivity schemes, and an analysis framework of conditional decoupled uplink and downlink association is designed. Meanwhile, three possible uplink association scenarios are considered, the design and analysis of the decoupled uplink associations are performed under the given conditional downlink association, and the probability expressions of the coupled uplink associations are derived. We get the statistical description of the primary and secondary access distances. Firstly, the numerical and simulation analysis shows the effect of system parameters on the association probability. Secondly, we study the conditional coverage probability and spectrum efficiency of the NOMA heterogeneous network under coupled uplink and downlink association and coupled uplink and downlink association schemes, respectively. The experimental results show that, the coverage probability performance of the proposed DUDA scheme is better than the traditional coupled uplink and downlink association (CUDA) scheme. In addition, the spectrum efficiency performance of the two schemes is related to system parameters.

  A case study on detection and analysis of  large-scale prefix hijacking incidents on the Internet

  SUI Dong-fang, TANG Yong, LIU Yu-jing, WANG En-ze

  2020, 42(07): 1184-1190. doi: 10.3969/j.issn.1007-130X.2020.07.006

  Abstract ( 323 )   PDF (696KB) ( 146 )   　　

  Due to the vulnerability of BGP protocol, BGP prefix hijacking has long been a serious security threat to the Internet. Detection and analysis of large-scale prefix hijacking incidents is a very ne- cessary but challenging task. This paper takes the large-scale European route leakage incident leading to route hijacking in 2019 as a case, and develops an effective detection and analysis method based on public BGP data. The analysis results include the following: firstly, the "attacker" of this hijacking is AS21217, and AS4134 is the key point in the process of hijacking route transmission; secondly, the hijacking caused serious multi-source AS conflict and as-path PATH expansion; thirdly, the hijacking types of this event include hijacking prefix and tampering with AS path and hijacking subprefix and tampering with AS path; fourthly, 311 AS were detected to be infected, with the largest number of infected chains of length 4, and 28 118 prefix IP segments belonging to 3 895 AS became victims. At the same time, a visual system is implemented to show the global network situation when the hijacking occurred. On the one hand, these results are consistent with the results published by Oracle and other companies; on the other hand, more detailed experiments and supplements have been carried out in multiple directions.
  

  
  

  Design and implementation of a high  level code clone detection method

  ZOU Yue, WU Ming, XU Yun

  2020, 42(07): 1191-1196. doi:

  Abstract ( 146 )   PDF (550KB) ( 108 )   　　

  Code clone detection is a basic research in software engineering, and it is widely used in software analysis and maintenance. At present, for detecting high-level clone with text difference, namely type-3/type-4 clone defined in the academic field, the existing methods have the problem of low detection rate (recall rate). The PDG (Program Dependency Graph) based detection methods are very important in high-level clone detection area, but these methods mostly rely on the accurate graph matching algorithms such as subgraph isomorphism, which have high time complexity and low recovery. Therefore, we propose a novel high-level code clone detection method, which uses the approximate graph matching algorithm based on Weisfeiler -Lehman graph kernel to detect clones. The experimental results show that our method can detect more high-level clones and run faster than the existing methods.
  

  
  

  Verification of pattern driven system security design

  ZHENG Xiao-yu, LIU Dong-mei, DU Yi-ning, ZHOU Zi-jian, QIU Mei-mei, ZHU Hong

  2020, 42(07): 1197-1207. doi: 10.3969/j.issn.1007-130X.2020.07.008

  Abstract ( 192 )   PDF (731KB) ( 102 )   　　

  With the wide use of the World Wide Web and mobile computing technology, system security has received more and more attentions. Using security models to design and verify system security solutions is an effective way to improve the system security. The existing methods select the applicable security patterns according to the security requirements of the system and then combine the patterns into a system security solution. The security is verified by the model detection method. However, these methods often regard the scheme as a whole for verification, ignoring the combination details of internal security patterns, and it is difficult to locate defects in a complex system containing a large number of patterns. A verification method for pattern-driven system security design is proposed. Firstly, the algebraic protocol language SOFIA is used to describe the security model and its combination to build a formal model of the system security solution. Secondly, the SOFIA protocol is converted to the Alloy protocol, and the model checking tool is used to verify the correctness of the pattern combination and the security of the system. A case study demonstrates that this method can effectively verify the correctness of system security solution.
  

  
  

  An adaptive graph planning protocol generation algorithm based on agent capability commitment collaboration

  GUO Jing-zhi , LIU Wei, XU Long-long, CHEN Deng

  2020, 42(07): 1208-1214. doi: 10.3969/j.issn.1007-130X.2020.07.009

  Abstract ( 99 )   PDF (712KB) ( 72 )   　　

  The adaptability of Open Multi-Agent System (OMAS) is embodied in the adaptation of agents to the changes of environment at runtime by co-adjusting the System behavior．Aiming at the problem that the predefined collaboration mechanism of existing multi-agent design cannot satisfy the runtime adaptive collaboration, this paper proposes an adaptive multi-agent Capability Commitment collaboration Graph-planning Protocol (CCGP) algorithm based on the Goal-Capability-Commitment (GCC) metamodel. Firstly, the GCC model supporting the semantic similarity calculation of context environment is proposed. Secondly, the semantic matching degree between goals and capabilities (or commitments) is introduced into the graph-planning protocol based on capability collaboration to optimize the algorithm. Finally, the medical waste Automated Guided Vehicle transportation system is taken as the experimental scenario to conduct two sets of comparative experiments. The experimental results show that the execution time of CCGP algorithm for graph planning protocol generation algorithm is significantly improved.
  

  
  

  An improved TLD real-time target tracking  algorithm based on CN algorithm

  ZHANG Jing, HUANG Hao-miao, WANG Jian-min, BAO Jun-rong

  2020, 42(07): 1215-1225. doi: 10.3969/j.issn.1007-130X.2020.07.010

  Abstract ( 248 )   PDF (1382KB) ( 113 )   　　

  Aiming at the problem that the tracking frame of the tracking-learning-detection (TLD) algorithm can cause tracking drift when the target is not rigidly deformed and rotated and the background is cluttered, an improved TLD real-time visual tracking algorithm incorporating the CN algorithm (TLD-CN) is proposed. Firstly, TLD-CN calculates the image saliency of the area within the tracking frame to obtain the threshold of the feature points sampled by the BRISK algorithm.  The appropriate feature points are obtained to establish the rotation and scale normalized descriptors. Secondly, the color features and texture features are fused to track the descriptors in the frame before and after the frame. The optimal similarity matching is performed to obtain a matched feature point set. After the discriminative encoding of the discriminative dictionary for the feature points in the set, the similarity measures are performed with the central pixel points of the CN tracking box and the TLD tracking box to obtain the weighting factors after output box adjustment. The experimental results show that the TLD-CN algorithm measures the weight value of the output box adjustment after fusing the two algorithms through the feature points, and has high accuracy and success in complex tracking scenarios such as target deformation, rotation, background clutter, and fast motion rate. The adaptive updating of the weight coefficients avoids model over-fitting and achieves real-time tracking effect.
  

  
  

  Fusion algorithm of infrared and  visible images in DTCWT domain

  ZHANG Gui-cang, SU Jin-feng, TUO Ming-xiu

  2020, 42(07): 1226-1233. doi: 10.3969/j.issn.1007-130X.2020.07.011

  Abstract ( 183 )   PDF (836KB) ( 105 )   　　

  Aiming at the problem that fusing infrared and visible images has low construct and clarity and high noise interference, a fusion algorithm of infrared and visible images in DTCWT domain is proposed. Firstly, the source image is pre-enhanced, then the low-frequency and high-frequency subband images are obtained by DTCWT positive transformation. Secondly, the low-frequency fusion rules based on intuitionistic fuzzy sets and the high-frequency fusion rules based on differential contrast of information are fused respectively. Finally, the fused image is obtained by DTCWT inverse transform of the low-frequency and high-frequency subbands. The experimental results show that the proposed algorithm can effectively improve the contrast and clarity of the fused image and reduce the noise interference, it has better objective evaluation index than the existing algorithms, and the operation efficiency is also improved.
  

  
  

  Vehicle distance measurement with vehicle lower edge estimation  and inverse perspective mapping based on monocular vision

  WANG Yong-sen, LIU Hong-zhe

  2020, 42(07): 1234-1243. doi: 10.3969/j.issn.1007-130X.2020.07.012

  Abstract ( 273 )   PDF (1006KB) ( 164 )   　　

  Front vehicle ranging plays a vital role in the field of autonomous vehicle technology. Aiming at the problem that the current vehicle ranging technology based on monocular vision neglects the lower edge of the vehicle connected to the ground, this paper proposes a monocular vision ranging model based on vehicle lower edge estimation and inverse perspective transformation, which realizes high- precision distance measurement of lateral and longitudinal directions of vehicles in front. Firstly, the vehicle's key point estimation and geometric relationship model are used to estimate the lower edge of the vehicle, then the distance measurement key points are calculated, and the point-based inverse perspective transformation distance measurement model is used to calculate the distance. The experimental results show that, compared with other monocular vision vehicle ranging algorithms, this method improves the accuracy and stability of ranging.
  
  
  

  A Retinex image enhancement algorithm based on  L0-norm

  CHEN Ru-xia, QIANG Zhen-ping, SHAO Xiao-feng, HE Li-bo

  2020, 42(07): 1244-1252. doi: 10.3969/j.issn.1007-130X.2020.07.013

  Abstract ( 141 )   PDF (1457KB) ( 117 )   　　

  In the image acquisition process, some issues such as uneven exposure caused by mechanical equipment, weather conditions and other reasons will make the quality of the taken picture poor and unable to meet the needs of practical applications. When the traditional Retinex algorithm is applied to enhance the image, it will cause the problems such as blurred image edges and graying. Therefore, aiming at the existing problems of the traditional Retinex algorithm, this paper proposes a novel image enhancement algorithm: Retinex image enhancement algorithm based on  L0-norm. Firstly, the contour components of the image are extracted by the global   L0 gradient minimization method and processed by the Retinex algorithm, and then the extracted contour components are fused to the original image to realize the enhancement of the original image. In the implementation of the algorithm, different global  L0  gradient minimization factors are used to ensure the uniform enhancement of contour components at different scales. Experimental results show that the proposed algorithm can better preserve the edge information of the image while enhancing the contrast of the image.
  
  
  
  

  
  

  Forest fire monitoring based on machine vision in foggy weather

  LIU Shu-dong, YAO Wen-bo, ZHANG Yan

  2020, 42(07): 1253-1261. doi: 10.3969/j.issn.1007-130X.2020.07.014

  Abstract ( 200 )   PDF (1011KB) ( 129 )   　　

  Forest fire monitoring based on machine vision has gradually become an important development direction in the field of forest fire monitoring. Smoke is an important indicator of forest fire monitoring. However, many disturbances such as clouds and similar smoke greatly reduce the accuracy of fire recognition. In response to this problem, this paper proposes a machine vision based forest fire monitoring method that combines dehazed and smoke detection. Firstly, several frame images in the appro- priate sample video are extracted as sample images, and the sample images are dehazed by the Haze-Line based dehazing algorithm. Secondly, the smoke detection method based on the Horn-Schunck optical flow method is used to detect the smoke. The maximum inter-class variance method is used to remove the influence of the pixel quality difference between two adjacent frames on the smoke detection. Finally, diffusion analysis is used to do fire monitoring. Results of simulation experiments and comparative analysis show that the proposed method can detect the trend that smoke area gradually increases with time, so as to monitor forest fire under foggy conditions effectively with higher accuracy and robustness. 
  
  
  
  

  Fast detection of target face based on the improved MTCNN network

  JIA Xiao-shuo, ZENG Shang-you, PAN Bing, ZHOU Yue

  2020, 42(07): 1262-1266. doi: 10.3969/j.issn.1007-130X.2020.07.015

  Abstract ( 332 )   PDF (696KB) ( 149 )   　　

  Traditional detection networks have always had problems of low detection efficiency and low accuracy in complex backgrounds. Aiming at the above problems, this paper further designs the MT-Siam network based on the MTCNN algorithm, which mainly provides accurate position positioning for the independent segmentation of single target, image processing of single target and other operations in the future, so as to quickly obtain the target position and achieve the purpose of improving the detection efficiency. In the experiments, a comprehensive comparison is made on the basic model of YOLOv3, SSD300 and MTCNN to verify the superiority of MTCNN network. The comparative experiments show that, compared with MTCNN, the proposed MT-Siam algorithm can improve the detection speed by 70% to 85% while maintaining high precision.
  

  
  

   A survey of order dispatch policy based on online ride-hailing services

  ZHENG Xiao-hong, LONG Jun, CAI Zhi-ping

  2020, 42(07): 1267-1275. doi: 10.3969/j.issn.1007-130X.2020.07.016

  Abstract ( 291 )   PDF (547KB) ( 190 )   　　

  Online ride-hailing services play an extremely important role in our daily travel life. With the development of the times, more and more people are accustomed to using mobile phones to take taxis through network platforms, but sometimes passenger requests are not met for a long time, and drivers are unloaded for a long distance, which not only seriously affects the experience of passengers and dri- vers, but also reduces people's travel efficiency. How to better match passenger requests and the needs of no-load drivers has always been a key research focus of the on-demand ride-hailing platforms. The study of online ride-hailing order dispatch strategy will help reduce passenger waiting time, increase driver revenue, reduce the no-load distance of drivers, and improve resource utilization. This survey firstly describes the entire process from the passenger’s initiating a taxi request to time the order is responded to. Secondly, it introduces the order allocation strategy under different dispatch modes in detail. Finally, it comprehensively lists the evaluation metrics for order dispatch strategy.
  
  
  

  Application of Galois connection in constructing matroids

  WANG Gang, MAO Hua, WU Zhen-yu

  2020, 42(07): 1276-1286. doi: 10.3969/j.issn.1007-130X.2020.07.017

  Abstract ( 213 )   PDF (647KB) ( 118 )   　　

  How to apply the effective methods of processing data to the research of constructing the matroid structure, so that the results of matroid theory can be applied to the research of data processing in more ways, is one of the problems that need to be solved in the research of matroid theory. Therefore, Galois connection, which is the basic operation in the theory of formal concept analysis that is an effective way in dealing data, is used as a bridge to give the definition of Galois context and obtain a method to establish a Galois context with a matroid. Furthermore, it finds that, under the matroid isomorphism and Galois isomorphism, there is a bijection between the family of matroids and the family of Galois contexts, and a bijection between the family of matroids and the family of Galois connections. Moreover, the relationship between matroids and Galois connections is found. This relationship is used to obtain the two methods for constructing matroids with Galois connections. Comparison with some relative known methods shows the advantages of the two methods. In addition, the biology examples show the role of the two methods in dealing the data of biological information. 
  
  

  Research on intelligent vehicle path planning based  on  improved bidirectional random tree

  SHI Yang-yang, YANG Jia-fu, MEI Miao, ZHU Lin-feng,

  2020, 42(07): 1287-1293. doi: 10.3969/j.issn.1007-130X.2020.07.018

  Abstract ( 128 )   PDF (662KB) ( 113 )   　　

  Aiming at the problems of large randomness, slow convergence speed and deviation of rapidly-exploring random tree algorithm, the basic fast random tree algorithm is improved by using cyclic alternating iteration search method to generate a new node and adopting a bidirectional random tree to do search simultaneously. A vehicle steering model is established to determine the constraint range of the vehicle steering angle, and the vehicle turning angle constraints are increased in the algorithm to reduce the deviation of the generated path, and improve the quality of the generated path. Node optimization is performed on the generated path to remove redundant nodes, shorten the path length, and improve the feasibility of the path. The B-spline curve is used to smooth the path by inserting the local end points, thus making the generated path more in line with the driving conditions of the vehicle. Simulation in MATLAB verifies the correctness of the algorithm.
  
  

  Robot path planning using a hybrid grey wolf optimization algorithm

  WANG Yong-qi, JIANG Xiao-xiao

  2020, 42(07): 1294-1301. doi: 10.3969/j.issn.1007-130X.2020.07.019

  Abstract ( 250 )   PDF (861KB) ( 153 )   　　

  To overcome drawbacks of grey wolf optimization algorithm (GWO), such as low convergence accuracy and easily trapping in local optimum, this paper proposes a hybrid grey wolf optimization (HGWO) algorithm and applies it to the robot path planning (RPP) problem. Firstly, HGWO uses the opposition-based learning method to generate initial population with high qualities. Secondly, the algorithm incorporates its historical information into the individual update method, so as to guide the population evolution. Meanwhile, an elite opposition strategy is applied to explore the space of elite solutions in current population, in order to strengthen the algorithm's exploitation ability. In addition, HGWO adopts the spline interpolation technique to guarantee convergence accuracy and to reduce the optimization difficulty of RPP. Finally, a comparative experiment of function optimization and path planning is carried out. The experimental results show that HGWO has good solution accuracy and robustness.
  
  
  
  
  

  Automatic construction of the garbage barrage  shielding dictionary based on seed words and dataset

  WANG Ge, WU Fang-jun,

  2020, 42(07): 1302-1308. doi: 10.3969/j.issn.1007-130X.2020.07.020

  Abstract ( 231 )   PDF (751KB) ( 83 )   　　

  With the popularity of barrage video, barrage has become a form of interactive communication among young people in the Internet age, but with the increase in the number of barrage, how to block junk barrage has become a problem. On the basis of keyword masking method proposed by various video websites, this paper proposes two automatic shielding dictionary construction methods based on seed words and data sets respectively. The first method mainly uses Google’s natural language proces- sing tool (word2vec) and point mutual information (PMI). These words with greater similarity to seed words or more co-occurrences are added into the shielding dictionary. The second method mainly adopts TF-IDF (Term Frequency Inverse Document Frequency), LDA (Latent Dirichlet Allocation) topic model and IG (Information Gain), and extracts the keywords from the garbage barrage dataset to construct the shielding dictionary. Finally, the constructed shielding dictionaries are evaluated. The experimental results show that the barrage shielding effect is best when the dictionary scale is 400~ 500. Besides, the influence of LDA topic number and dataset size on the shielding effect of the barrage is also investigated.