Loading...
  • 中国计算机学会会刊
  • 中国科技核心期刊
  • 中文核心期刊

Current Issue

    • 论文
      A highthroughput computational approach
      for SQS binary alloy design 
      WANG Juan1,2,YANG Xiaoyu1,WANG Zongguo1,WANG Xingyang1,2,ZHANG Xiaoli3,ZENG
      2016, 38(03): 401-410. doi:
      Abstract ( 178 )   PDF (974KB) ( 610 )     

      Since the United States announced the "Materials Genome Initiative" in June of 2011, developing integrated highthroughput computational material platform and computational material database to accelerate the development of new materials has become increasingly important. MatCloud is a highthroughput computational material platform and a software framework for highthroughput material automatic computation, which aims to propose universal methods and techniques for new material designing and computing. Creating a batch of structures, screening structures from vast amount of unknown structures and property calculation are the key problems of designing binary alloy. In order to support the widely used special quasirandom structure (SQS) method for alloy design, we propose a highthroughput automatic calculation algorithm, and develop a "plugin" following the specifications and interfaces provided by MatCloud for highthroughput SQS binary alloy design and automatic computation. The elastic modulus of ZrxTi(1-x)  alloy are then calculated using the proposed method. As a result, all the calculation jobs are computed successfully and automatically, the results are processed without human operations, the preset effects are achieved and the calculation performance is improved. The proposed method and the “plugin” design for constructing binary alloy are proved appropriate and effective, and they can help accelerate the development of binary alloy.

      A reconfigurable hardware architecture
      design for multiple Hash algorithms 
      LIU Heng1,HUANG Kai1,XIU Siwen2,LI Yijun3,YAN Xiaolang1
      2016, 38(03): 411-417. doi:
      Abstract ( 170 )   PDF (750KB) ( 296 )     

      Since the existing hardware architecture for Hash algorithms can only implement a few algorithms, we design a reconfigurable IP, which can implement seven Hash algorithms including SM3, MD5, SHA1 and SHA2 family, and it can meet the demand of a system for algorithm diversity. By analyzing all these Hash algorithms and estimating their similarity, the design reuses adders and registers to the maximum extent and therefore greatly reduces the total area. Besides, the design is flexibly configurable and can access the memory directly. The implementation results based on the FPGA of Stratix II of Altera Corporation show that, in comparison with the existing designs, the maximum frequency can achieve 100MHz, the whole area is decreased by more than 26.7% and the throughputperarea for each of the seven algorithms is increased.

      A construction algorithm of the stable shortest path tree  
      YANG Xiaohua1,2,WU Jigang1,2,SHI Wenjun1,2,ZHAO Guodong1,2
      2016, 38(03): 418-424. doi:
      Abstract ( 134 )   PDF (813KB) ( 424 )     

      Constructing the shortest path tree is an important problem in dynamic networks, in which the changed edges will lead to dynamic reconstruction of the shortest path tree. Repeated calculation not only consumes a lot of time, but also results in frequent changes of the shortest path tree. We propose an algorithm for constructing the stable shortest path tree, which takes fewer update operations on nodes. The algorithm updates the shortest path tree by excluding the unstable edges so as to effectively reduce the number of the update operations. Experimental results show that compared with traditional dynamic network shortest path tree algorithms, the proposed algorithm can generate relatively more stable shortest path tree. The update time is improved by 57-24%, and the number of update operations on nodes is reduced by 43-6%.

      A new session identification
      method based on MapReduce 
      HUANG Weijian,SONG Yuanyuan
      2016, 38(03): 425-430. doi:
      Abstract ( 108 )   PDF (627KB) ( 244 )     

      Web log preprocessing attracts more and more attention due to the importance of its output result. Meanwhile distributed processing of massive data based on Hadoop is being widely studied and applied, so Web log preprocessing with MapReduce becomes an inevitable development trend. In order to improve the accuracy of session identification results, we propose a new method to identify user session based on network topology and dynamic threshold. The current research state is analyzed and the advantages of this method are also discussed. Then, the MapReduce model is used to implement the distributed processing of the new method. Experimental results demonstrate high efficiency and high accuracy of the proposed method.

      Application of the mesh fusion method in numerical
      simulation of a highspeed train passing through a tunnel  
      CHEN Xiaoli1,ZHANG Chengyu1,XU Jianlin1,JIN Yongrong2 ,MEI Yuangui1
      2016, 38(03): 431-436. doi:
      Abstract ( 126 )   PDF (786KB) ( 290 )     

      In numerical simulation of turbulent flow around a highspeed train passing through a tunnel, in order to get more physically accurate results, detailed local compositions such as bogies, pantograph cover and windshield have to be considered, thus the model becomes increasingly complicated. The multiblock mesh generation and mesh fusion method are adopted to obtain a computational grid with high quality and applicability. Unstructured tetrahedral meshes are generated around the highspeed train, and structured hexahedral meshes are generated in the moving zone and the rest of solution domain. On the interfaces, the triangular and quadrangular surface meshes of two blocks are fused by controlling the change of node positions, which makes the topology consistent and the grids connected. The subsequent simulation results of the computational grids are compared with those of one dimensional characteristics method, which shows that the presented method can provide less grids points and higher generation rate under the same accuracy requirement. The method can be further used in similar studies.

      Design and implementation of a lightweight
      object communication mechanism  
      ZENG Yingchao,WANG Maozhi,FENG Weihuan
      2016, 38(03): 437-442. doi:
      Abstract ( 183 )   PDF (786KB) ( 238 )     

      When objects communicate with others across different DLLs or threads, the system coupling degree can be easily increased. In order to solve this problem, using the idea of Event model in Action Script 3.0 and the computer bus communication for reference, we design a lightweight object communication mechanism based on C++11. This lightweight object communication mechanism uses events as the message carrier, which can not only be transmitted and forms an event stream on the bus, but also can be caught by the objects with corresponding listener types. Objects or DLLs essentially interact with the bus, and the interface between the two is predefined events, thus the coupling between objects can be effectively reduced. This mechanism builds a twolayer architecture: the transport layer and the adaptation layer. The DLL belonging to the transport layer will load the DLL provided by the adaptation layer through implicit linking. The DLL provided by the adaptation layer contains a static object queue, a static event queue and a bus, thus all the individual objects in the adaptation layer can easily and safely complete the interaction between objects by calling its own member functions. 

      A study on the smart city server system
      based on contextaware technology  
      LI Yingzhi1,XIAO Laiyuan2
      2016, 38(03): 443-448. doi:
      Abstract ( 135 )   PDF (803KB) ( 287 )     

      As the city scale is becoming bigger and bigger, the governments at all levels are working on city Internet of things (IOT) and information system construction. Although the Internet backbone networks are set up and a large number of sensors are deployed which can collect numerous data on city behaviors, the poor information management mode hinders the realization of the value of information which should be used to promote service, and results in enormous waste. So we introduce the situational awareness technology to smart city services, and build a city service system based on situational awareness. Key techniques such as information collection, context information inference and service configuration model are employed to achieve realtime situation perception of the cities and provide intelligent business service portfolio. Experiments on a community smart management system verify the effect of the system designed in this paper.

      An efficient completely dynamic SPT algorithm  
      WANG Xiaojie,GUO Wenqiang,WANG Sixiu,CAI Yongmei
      2016, 38(03): 449-453. doi:
      Abstract ( 135 )   PDF (610KB) ( 621 )     

      Computation of the shortest path between two nodes in communication networks is the basis of the link state routing protocol. Based on the study on the present dynamic SPT algorithms, we propose a new completely dynamic SPT algorithm (DSPTID) for handling network topology changing. This algorithm establishes an SPT update queue and pays attention only to the changes of nodes and edges by making use of information provided by the existing SPT, which can achieve incremental update. When the topology of the network changes, the increase or decrease of edge weights can be processed by the DSPTID algorithm. The analysis of algorithm complexity and simulation results show that the DSPTID requires less update of nodes and has higher efficiency. 

      An indoor RFID tag
      localization algorithm in NLOS environment      
      SHEN Guohao,MA Yongtao,LIU Kaihua,ZHAO Yu
      2016, 38(03): 454-459. doi:
      Abstract ( 184 )   PDF (761KB) ( 272 )     

      In indoor wireless localization environment, signal propagation is complicated due to the nonline of sight (NLOS). Traditional algorithms such as LS and Chan cannot achieve satisfactory accuracy. In this Paper, using the passive UHF RFID technology, we analyze and build a channel model, and then discuss localization errors. Based on the distance information obtained from the phase of arrival (POA), we propose a twostep tag localization algorithm. First, we use the interior point (IP) algorithm to solve the NLOS problem. Through multiple iterations, the NLOS bias can be removed from phasebased range measurement. Based on the revised range results, we utilize the residual weighted algorithm to get the accurate RFID tag position. Comparison of the simulation results demonstrates that the proposed method outperforms the traditional methods.

      A method for protecting user data
      files based on private directory container        
      WEI Lifeng,DING Yan,CHEN Songzheng,HUANG Chenlin
      2016, 38(03): 460-464. doi:
      Abstract ( 113 )   PDF (789KB) ( 198 )     

      Because of the existence of privilege users such as the administrator, user  data files are difficult to protect. In order to make sure that only the file owner can access these data files, we create a private directory container (PDC), which includes token attributes of user. The PDC access control mechanism can be realized under the OS kernel mandatory access control (MAC) framework, thus protecting user private data files. Only the owner of the PDC can access the files in the PDC, no exception, including administrators. User’s token can only be got by identity authentication, so the owner’s PDC identity cannot be got by others, and private files can be protected effectively.

      An improved remote user anonymous
      authentication scheme using smart cards  
      LIU Runjie,LIU Hengchao,SHEN Jinyuan
      2016, 38(03): 465-470. doi:
      Abstract ( 113 )   PDF (445KB) ( 196 )     

      We find some security flaws in Sonwanshi’s remote user authentication scheme, such as poor session key security and incapability to resist impersonation attacks and offline password guessing attacks. We propose an improvement scheme, which mainly enhances the security of Sonwanshi’s scheme in the registration and login phase. In the registration phase, users’ passwords are directly stored in the local smart cards rather than be submitted to the server, which not only reduces the costs of servers for password storage and maintenance, but also improves the security performance. In the login phase, the original time stamp mode is replaced by a random number challenge response mode to avoid authentication failure caused by clock asynchronization. The analysis on security performance and efficiency shows that the proposed scheme not only eliminates the defects of Sonwanshi’s scheme, but also reduces the time complexity in comparison with similar schemes. It, therefore, is suitable for those devices with low processing power and high security requirements.

      An autonomic rekeying model for group key management  
      SUN Liyan1,ZHOU Senxin1,ZHOU Jian1,2
      2016, 38(03): 471-477. doi:
      Abstract ( 101 )   PDF (735KB) ( 211 )     

      In group key management of distributed wireless networks, nonupdate members take part in shared key calculation, which incurs much interactive time delay. To deal with the challenge, we present a novel autonomic rekeying model, and then an autonomic group key management is designed on basis of DH protocol and polynomial formula without support of online KMC, which meets the requirement of oneencryptionkey multidecryptionkey. All nonupdate members do not participate in rekeying for key joining/leaving operations, and the fresh public encryption keys cannot break the validity of secret decryption keys belonging to nonupdate members. Therefore, the rekeying delay time of the proposed scheme is reduced efficiently, and it also has the properties of  the autonomic rekeying model logically. In addition, the scheme is suitable for wireless networks whose time delay is limited strictly.

      Formal description and compatibility
      analysis of OWLS process model 
      TANG Zhengyi1,3,WANG Jinshui1,WEI Li2,3,XUE Xingsi1,HU Wenyu1
      2016, 38(03): 478-485. doi:
      Abstract ( 105 )   PDF (1210KB) ( 170 )     

      Building the formal model of composite services is the basis of verification. But in the absence of a unified modeling framework, automatic operations of the modeling process are difficult to realize. We build a services process automata by expanding the deterministic finite automata. It is a finite migratory system which is suitable for describing the process model of OWLS. This work provides a unified framework and a foundation of automatic operations for the formal modeling of composite services. In addition, we divide the compatibility into three grades by analyzing the acceptable locations of the services process automata and the constraints possibly produced during the composition process. So the proposal can be adaptive to verification of composite services with different reliability requirements.

      An evaluation model of scientific research ability
      of universities based on cooperative IWD and RBNN          
      LIU Chunxia1,2,TIAN Yun2
      2016, 38(03): 486-493. doi:
      Abstract ( 113 )   PDF (1182KB) ( 187 )     

      Aiming at the problems in the evaluation process of the scientific research abilities due to its multiple factors and high non linear characteristics, as well as the subjective assessment in classic models, which leads to low accuracy, we propose an evaluation model of the collaborative ability of scientific research based on the intelligent water drops (IWD) algorithm and the rough set block neural network (RBNN). We first introduce the IWD algorithm, and design a local spatial auto scaling algorithm (LSSA) to solve the problem of fixed beside domain search range of the traditional IWD algorithm that is not conducive to improve search efficiency. The LSSA can automatically adjust the next search space size according to the best individual of current population, thus improving the evolutionary efficiency of algorithms. Based on the rough set theory, the data of scientific research ability of universities is preprocessed, which can simplify data calculation. Finally, the parameters of the block neural network and the rough set are encoded, and the model of scientific research abilities is evaluated. Simulation results show that the model has high accuracy and fast computational efficiency.  

      Robotassisted ultrasound scanning system based on Kinect       
      MENG Bo1,CAO Lei2
      2016, 38(03): 494-500. doi:
      Abstract ( 213 )   PDF (2015KB) ( 328 )     

      We propose a novel Kinectbased robotassisted ultrasound scanning system. The system is composed of a robotic arm, a linear ultrasound probe with marker, and a Kinect as a vision servo. The robot follows the navigation of the Kinect sensors and scans the legs automatically. A Kinect is chosen here to obtain the 3D images of legs and the probe. The probe and legs are segmented in RGB images and depth images in real time. The probe’s position and orientation are registered according to the registration. Then the pose of the robot can be calculated according to the calibration result of Kinect coordinate system and robot coordinate system. The robot holds the probe to scan the legs according to the planned path and current pose. Both phantom and human leg platforms with robot and ultrasound probe and Kinect are built up and scan experiments are carried out. The transformation matrixes between two legs, depth imaging, Kinect and ultrasound probe coordinate systems, are obtained. The distance between the probe and the scanned leg is calculated frame by frame to guarantee safe scanning without force sensors. Results from the initial experiments indicate that the idea is feasible and promising for its improved inspection efficiency. Clinically, the method can be implemented for preoperative procedures and this may improve the ultrasound scanning effiency.

      A mean particle swarm optimization algorithm
      based on adaptive inertia weight  
      ZHAO Zhigang,LIN Yujiao,YIN Zhaoyuan
      2016, 38(03): 501-506. doi:
      Abstract ( 199 )   PDF (1912KB) ( 377 )     

      In order to tackle the problems of slow convergence, low accuracy and parameterdependence of the standard particle swarm optimization (PSO) algorithm, we propose a mean PSO algorithm based on adaptive inertia weight. The new algorithm improves its performance by adjusting inertia weight dynamically and adaptively. It divides particles into three groups according to each fitness value and applies different inertia weight strategies for  particles of different groups, making the particles choose appropriate inertia weight values according to their own position, and converges to the global optimal position faster. Furthermore, the individual and global optima are replaced by their linear combination during the algorithm iteration phase to enhance the calculation performance of the PSO. Experimental results show that the proposed algorithm outperforms the standard PSO and some other improved PSO algorithms with less iteration for finding the optimal solution, and has faster convergence and higher convergence precision.

      Train operation adjustment using
      an improved genetic algorithm  
      CHEN Dong1,PENG Qiyuan2,ZHANG Yan1,LI Yonghui1
      2016, 38(03): 507-513. doi:
      Abstract ( 95 )   PDF (779KB) ( 218 )     

      We study the collaborative optimization of wagon flow organization in technical stations and train operation adjustment in sections, realize “flow and line integration” of transportation regions by adjusting the trains on the way, thus obtaining better transportation production index. We take the wagon flow requirements of the forward technical station as one of the objectives of train operation adjustment and refer it as matching flow adjustment. Analyzing the matching flow adjustment strategies and their application scenarios, we build a matching flow model and design the solving algorithm based on the improved genetic algorithm. The algorithm design fully integrates the features of train operation adjustment and effectively avoids problems of premature and slow convergence. It has good applicability with acceptable computation time. The feasibility of the adjustment targets can be quickly verified, and specific adjustment measures are then proposed. Our findings lay a foundation for further study on the collaborative optimization of wagon flow organization and train operation adjustment in technical stations.

      A network pornographic image
      filtering algorithm based on ratio features   
      WANG Jingzhong,ZHOU Jing
      2016, 38(03): 514-519. doi:
      Abstract ( 113 )   PDF (923KB) ( 226 )     

      Currently the exposed skin ratio is still frequently used by most network pornographic image filtering methods. However, it has some defects such as the interference of analogous skin color, the interference of nonkeyarea skin and sometimes the irrelevance of the whole exposed skin ratio and the exposed extent. We propose a network pornographic image filtering algorithm based on the consistence of skin color and human body structure, which calculates the face size ratio, the whole exposed skin ratio after the skin color is improved by face skin color data in the skin color model and the keyarea exposed skin ratio. The problems mentioned above have been effectively solved by using these three ratio features. Besides, we add a rotation process to the face detection module, so it can detect severely inclined face effectively. Experimental results show that the recognition rate of the proposed algorithm can achieve 91.5%, which meets the practical demand.

      Sketch-based image retrieval based on multi-scale HOG          
      LI Sisi,CHEN Xi,XIAO Jian
      2016, 38(03): 520-527. doi:
      Abstract ( 117 )   PDF (1224KB) ( 250 )     

      Sketch-based image retrieval is an important research topic in the field of image processing. We present an improved feature extraction method for sketchbased image retrieval, which combines the Gaussian pyramid and local histogram of gradient (HOG) features. By using Gaussian pyramid, the image is decomposed into multiple resolution images on which the points of interest are extracted, and the points based multiscale HOG features are obtained. The visual word dictionary is generated from the multiscale HOG features, and finally feature vectors relative to visual words are formed. The images are ranked according to decreasing similarity which is calculated by the distance between the feature vectors of sketches and the images. Compared with the HOG algorithm at single scale and several other algorithms,  experimental results demonstrate the feasibility and effectiveness of the proposed method.

      A simple and efficient reversible data hiding
      scheme based on histogram of difference image 
      ZHANG Kai
      2016, 38(03): 528-533. doi:
      Abstract ( 96 )   PDF (809KB) ( 365 )     

      We propose a reversible data hiding scheme based on histogram of difference image, which improves both embedding capacity and image quality after data embedding. The original image is firstly divided into blocks and scanned, and then differences between neighbor pixels are calculated. After that, the two peak points of the histogram of difference image are shifted bidirectionally to get the gap for data embedding. The data is preprocessed before it is embedded. The scheme makes full use of the similarity between neighbor pixels of natural images to increase both the number and the height of peak points of the histogram, thus improving the embedding capacity obviously. Besides, the preprocess of the information to be embedded makes the impact on data embedding smaller, and achieves embedded images of better quality.