Current Issue

• 2013, No. 8　Published:25 August 2013 Last issue    Next issue
  Columns in this issue: 论文

• 论文

  Data placement and task scheduling based on
  associated amount in cloud computing

  GUO Lizheng1,2,ZHAO Shuguang1,JIANG Changyuan1

  2013, 35(8): 1-7. doi:

  Abstract ( 114 )   PDF (952KB) ( 258 )   　　

  In scientific workflows, cluster or grid platform is used to deal with complex problems, and the emergence of cloud computing offers an alternative. The transferred data are huge in the dataintensive scientific workflows in cloud computing. Reducing the count and amount of the data transfers between different clusters in a datacenter is a challenging problem. There exists dependency in data of the scientific workflow. Firstly, a relational matrix is built based on the associate amount. Secondly, the relational matrix is clustered using bond energy algorithm. Thirdly, the clustered matrix is partitioned to k parts and each part is deployed to corresponding clusters in the datacenter. The simulation results show that the proposed method reduces the count and amount of the data transfers between clusters in a datacenter.

  Design and implementation of saturation
  arithmetic instructions mapping in Matrix complier       

  DAN Xiaoqiang,CHEN Yueyue,SUN Haiyan,YANG Liu,LUO Jie,XIN Naijun,WANG Ji

  2013, 35(8): 8-14. doi:

  Abstract ( 118 )   PDF (1278KB) ( 189 )   　　

  Matrix is a highperformance DSP processor for soft base stations. The saturation arithmetic instruction is of importance in Matrix, which enables the DSP algorithms more secure, accurate and efficient. As a result, supporting saturation arithmetic instructions mapping is critical for Matrix compiler, which is developed based on GCC. However, the current versions of GCC only support saturation arithmetic instruction with fixed-point types rather than integer or float types. Based on the internal mechanism of instructions mapping in GCC, an implementation scheme of C extensions based saturation arithmetic instructions is proposed. Further more, the proposed scheme is validated by implementing the saturation integer addition instruction mapping, and the results show that this scheme can support Matrix's saturation arithmetic instruction mapping efficiently and accurately.

  A novel data exchange architecture
  based on cloud computing      

  ZHAO Wei,ZHUO Wei,LI Zhanbo,WANG Zhihua

  2013, 35(8): 15-19. doi:

  Abstract ( 117 )   PDF (895KB) ( 299 )   　　

  According to the present situation that in enterprises there are many business systems and they are lack of data sharing, a novel MapReduce based data exchange scheme is proposed by using the advantage of clouding computing in massive data storage and analysis. The scheme first describes the MapReduce programming model. Then, a novel dataexchange architecture based on the cloud computing platform is designed. Based on the architecture, business flow chart, sequence diagram, data storage and parallel data processing are analyzed and the system security is discussed. Finally the scheme is validated by a practical application and has some reference value in data exchange based on cloud computing.

  Analysis of the impact of gateoxide degradation on
  nanoscale SRAM cell’s critical charge        

  JIN Zuolin,ZHANG Minxuan,SUN Yan,SHI Wenqiang

  2013, 35(8): 20-24. doi:

  Abstract ( 129 )   PDF (1024KB) ( 180 )   　　

  The CMOS technology has entered nanometer era. In nanometer scale, more and more transistors are integrated on one chip and the supply voltage becomes lower, and soft error problem begins to affect ground applications. At the meantime, because of the scaling down of gate oxide thickness, gateoxide degradation does a great harm to the reliability of circuits at high electric field pressure. Soft error and gateoxide degradation become the two main challenges for current and future CMOS technology. The effects of gateoxide degradation on the critical charge of SRAM cell is analyzed through an analytical model and SPICE simulation is performed to validate it. The simulation results show that, as gate-oxide degradation becomes worse, the critical charge of the SRAM cell shows approximate exponential decrease. And with the same gate-oxide degradation degree, Soft error rate of different technology scales shows an approximate linear relationship.

  Implementation of instruction scheduling algorithm
  based on list scheduling for Matrix  DSP

  LUO Jie,CHEN Yueyue,SUN Haiyan,YANG Liu,DAN Xiaoqiang,XIN Naijun,WANG Ji

  2013, 35(8): 25-30. doi:

  Abstract ( 123 )   PDF (682KB) ( 198 )   　　

  Instruction scheduling is an optimization strategy for gcc to achieve instruction level parallelism and to improve performance. At present, the scheduling algorithm that gcc supports are list scheduling algorithm and swing model scheduling. According to the architecture feature of Matrix DSP, an improved instruction scheduling algorithm based on the list scheduling algorithm is implemented. The experiment results show that the improved algorithm can compile the c code into correct assemble codes, improve the performance of instruction parallelism, explicitly mark the parallelism between instructions, and the delay between instructions meets the hardware demands.

  An optimized MP2P cooperative caching method
  based on popularity and minimum access cost         

  ZHOU Xinxin1,2,YU Zhenwei1

  2013, 35(8): 31-35. doi:

  Abstract ( 90 )   PDF (996KB) ( 150 )   　　

  In order to improve the quality of service of mobile peertopeer networks, according to characteristics of P2P networks, a novel distributed collaborative caching method is proposed. By computing the popularity and the cost of accessing data, the data whcih have higher popularity and much lower access cost will be cached first. Thus, the network gets low service delay and communication cost, also the cooperation between mobile nodes and the promotion of availability and efficiency can be achieved. Simulation results show that the strategy can effectively reduce network overhead and delay, and obtain better performance.

  Highefficient identitybased threshold signature scheme         

  ZHENG Guangliang1，WEI Lixian1,2，WANG Xuan1

  2013, 35(8): 36-40. doi:

  Abstract ( 91 )   PDF (1045KB) ( 179 )   　　

  Based on the signature scheme proposed by Guke, a new IDbased threshold signature scheme is designed, which is efficient and provably secure in the standard model (without random oracle). The generation of the threshold key does not require a trusted third party. The new scheme improves efficiency by reducing multiplicative operations and bilinear pairing operations, especially by precomputing bilinear pairing operation. And the new scheme is against existential forgery under the adaptive chosen message attack under the computation DiffieHellman assumption in the standard model. Compared with the current IDbased threshold signature scheme, the new one is more computational efficient.

  State determination algorithm of WSN
  based on twice grey Markov forecasting        

  LIN Wei,LI Bo,HAN Lihong

  2013, 35(8): 41-45. doi:

  Abstract ( 79 )   PDF (725KB) ( 219 )   　　

  According to characteristics of wireless sensor networks, for the sake of saving energy of nodes in wireless sensor networks, a twiceunbiased GreyMarkov forecasting model is proposed. Based on Markov prediction, according to the state determination rule and twiceunbiased Gray prediction, the environment state of wireless sensor networks is determined, and the flag of the state is transferred, so that wireless sensor networks can reduce the energy consumption. Simulation results show that the model improves the prediction accuracy and achieves effective state prediction.

  Improvement of F5 algorithm combined with fast coding     

  WU Song1，ZHANG Minqing1,2，WU Xuguang1

  2013, 35(8): 46-51. doi:

  Abstract ( 150 )   PDF (1658KB) ( 209 )   　　

  Digital steganography is an important way for secret communication, and F5 steganographic algorithm proposed by Westfeld is an excellent steganography in DCT domain. The main steps of F5 steganographic algorithm is introduced, the matrix encoding and embedding process are analyzed. Its defects are pointed out and the caution is analyzed. Then the original matrix encoding of F5 steganographic algorithm is replaced by fast coding which has a lower rate of modification, and the original embedding process is also changed. Experiments show that the new improved algorithm has improvement in both security and capacity.

  State optimization and performance simulation of adaptive
  MIMO-OFDM communication system based on CPSO-PF            

  JIAO Fangyuan,LI Wei,LI Jia

  2013, 35(8): 52-59. doi:

  Abstract ( 110 )   PDF (2200KB) ( 186 )   　　

  In view of the present practical issues that channel noise seriously affects the communication quality of MIMOOFDM communication system since its state parameters cannot be adjusted according to the channel interference and noise, this paper discussed the transmission principle and the performance parameters of MIMOOFDM communication system, studied the performance characteristics of chaotic particle swarm optimization (CPSO) and particle filter(PF), proposed a type of CPSO particle filterbased adaptive MIMOOFDM communication system model, designed a adaptive state optimization algorithm based on CPSOPF, and carried out he system BER simulation by combing typical interference noise. Simulation results show that the state optimization based on CPSOPF can restrain the noise effectively, and improve the system BER performance of adaptive MIMOOFDM communication significantly.

  An ondemand interdomain path building method 

  LIU Liyan

  2013, 35(8): 60-68. doi:

  Abstract ( 114 )   PDF (825KB) ( 134 )   　　

  In inter-domain routing, the optimal path forwarding rule of BGP makes nodes not control what paths they can receive. An ondemand interdomain path building method (OIPBM) is proposed, whose main feature is that nodes can lay control over the path selecting process of their downstream nodes and customize the routes according to their own requirements. Firstly, by extending BGP, a new protocol called P-BGP is proposed, which adds more policy options into routing advertisements. The inserted policies can be referred by the intermediate nodes when selecting paths. Secondly, based on P-BGP, OIPBM is proposed. OIPBM uses BGP to ensure that the source and the destination nodes are reachable, and any source node that requires a special path to destination node can request the destination to start a politic PBGP convergence process to find the satisfying path. Based on theoretical analysis, OIPBM is demonstrated to have good performance by experiments.

  Cryptanalysis and improvement of
  some certificateless signcryption schemes

  ZHOU Caixue

  2013, 35(8): 69-76. doi:

  Abstract ( 95 )   PDF (498KB) ( 305 )   　　

  Certificateless cryptography eliminates the key escrow problem inherent in identity based cryptosystems, and simplifies the certificate management in the traditional public key cryptosystem, which has great superiority. Four certificateless signcryption schemes were analyzed, which shows there exist confidentiality attacks in two schemes and forgeability attacks in three schemes. The four schemes were improved by using the methods of binding receiver in signature part, binding sender in encryption part and including random number in signcryption part respectively. Finally, the improved schemes were proved in the random oracle model, and it shows the improved schemes are secure.

  Survey of multiobjective simulated
  annealing algorithm and its applications    

  LI Jinzhong,XIA Jiewu,ZENG Xiaohui,ZENG Jintao,LIU Xinming,LENG Ming,SUN Li

  2013, 35(8): 77-88. doi:

  Abstract ( 261 )   PDF (620KB) ( 620 )   　　

   Multi-Objective Simulated Annealing (MOSA) algorithm has been widely studied and applied to various fields successfully as a simple and effective multi-objective intelligence optimization algorithm. A systematic survey and discussion of the development of MOSA algorithm and its application in the recent twenty years are introduced. Firstly, the generic framework of MOSA algorithm is briefly described. Secondly, several typical MOSA algorithms are discussed, calculation methods of acceptance probability functions for those algorithms are emphatically addressed, and these algorithms are classified and analyzed. Thirdly, some typical applications of MOSA algorithms are introduced. Finally, some promising directions and challenges for future research in the area of MOSA algorithm are proposed according to the present studies. This paper can provide a comprehensive reference for future study of MOSA in algorithm improvement and its practical applications.

  Improvement and performance research of hybrid
  algorithm based on compound chaoticartificial fish swarm         

  YI Xinbing1,2，YANG Kai1

  2013, 35(8): 89-95. doi:

  Abstract ( 112 )   PDF (1714KB) ( 163 )   　　

  When artificial fish swarm algorithm is close to the optimal point during its optimization process, the convergence rate declines so that it is difficult to get exact solutions. Besides, the algorithm is easy to fall into local minima in complex issues. Aiming at the aforementioned disadvantages, a hybrid algorithm is proposed, which combines the compound chaotic search technology and the improved artificial fish swarm algorithm. It adopts the mapping combination with more ergodicity to generate the local search method. The method can avoid that artificial fish are into local extremum area for a long time, so that it reaches the global extreme points more precisely. Meanwhile, the artificial fish swarm algorithm is improved by introducing feedbackswallowed behavior of artificial fish. The improved algorithm reduces optimization complexity at late stage, improves accuracy and guarantees convergence efficiency. Experimental results show that, under the same parameter conditions, the proposed hybrid algorithm outperforms the basic artificial fish swarm algorithm in convergence rate, optimization accuracy and global optimization ability. Experiments demonstrate the efficiency of the proposed method.

  Prediction of improved BP neural network
   by Adaboost algorithm     

  LI Xiang,ZHU Quanyin

  2013, 35(8): 96-102. doi:

  Abstract ( 130 )   PDF (1122KB) ( 219 )   　　

  The traditional BP (Back Propagation) neural network is easy to fall into local minimum and has lower accuracy. According to this problem, a method that combines the Adaboost algorithm and BP neural network is proposed to improve the prediction accuracy and generalization ability of the neural network. Firstly, the method preprocesses the historical data and initializes the distribution weights of test data. Secondly, it selects different hidden layer nodes, node transfer functions, training functions, and network learning functions to construct weak predictors of BP neural network and trains the sample data repeatedly. Finally, it made more weak predictors of BP neural network to form a new strong predictor by Adaboost algorithm. The database of UCI (University of California Irvine) is used in experiments. The results show that this method can reduce nearly 50% for the mean error absolute value compared to the traditional BP network, and improve the prediction accuracy of network. So this method provides references for the neural network prediction.

  Research on RMB paper currency
  number recognition algorithm         

  WANG Yan1,LIU Yang1，SONG Baichun2

  2013, 35(8): 103-108. doi:

  Abstract ( 109 )   PDF (951KB) ( 285 )   　　

  In order to improve the fitness of RMB paper currency number recognition algorithm, based on structural feature of characters, a novel RMB paper currency number recognition algorithm is proposed, which is based on multiple mutual compensation structure characteristics. And the image preprocess of RMB paper currency number is studied including image positioning, image binarization,  single character segmentation, character size normalization, etc. Algorithms of English letter arrangement, Arabia digital arrangement and their mixed arrangement of RMB paper currency number are designed. These algorithms are introduced by taking partial flow chain of mixed arrangement recognition program as an example. An ARM embedded system is used to validate three recogntion programs. Some suggestions are proposed to solve the issues such as font diference of characters, and character interference in the experiment. The statistical results of accuracy of three recognition programs show that the algorithm has high precision and meets the practical needs.

  Metric reconstruction of revolution surface     

  ZHANG Quanbing,SUN Qing,WANG Lei,CHENG Hong

  2013, 35(8): 109-113. doi:

  Abstract ( 98 )   PDF (763KB) ( 212 )   　　

  Threedimension reconstruction based on images is one of important researches in computer vision. The reconstruction of single uncalibrated image of revolution surface is studied. The algorithm introduced by Colombo etc, which obtains threedimension reconstruction and texture acquisition at the same time, is improved and a new approach of reconstruction is proposed. Firstly, plane transmission is used to compute the equation of the ellipse with respect to the point of image apparent contour line. Secondly, the construction of the object is recovered by computing the location and size of each cross section of the revolution surface according to the property of quadratic curve. The method of the reconstruction process is simple and keeps the overall structure of the revolution surface more correct. Simulation and real image experiments demonstrate the correctness and validity of the algorithm.

  A novel type of threshold functions for wavelet shrinkage  

  HE Xiping1,2,YANG Jin2,3

  2013, 35(8): 114-119. doi:

  Abstract ( 92 )   PDF (1342KB) ( 175 )   　　

  A novel type of threshold functions with a variable parameter is proposed based on the wavelet shrinkage architecture presented by Donoho  D L and Johnstone I M, and the effect of different parameter on signal shrinkage is also analyzed. This new type of threshold functions has many advantages over the soft and hard threshold functions. It is continuous and simple in expression, and has any order derivative which makes some kinds of mathematical processing very simple and convenient. It also overcomes the shortcomings of the soft threshold method where there is an invariable dispersion between the estimated wavelet coefficients and the decomposed wavelet coefficients. Further more, the new type of threshold functions is more flexible than the softthreshold and hardthreshold functions and integrates their excellence features. All these advantages make it possible to construct an adaptive signal denoising algorithm. Comparative experiment results show that the wavelet shrinkage method adopting the new threshold function is more capable of eliminating the noise from the signal, and results in less MSE and higher SNR than the traditional method adopting hard and soft threshold function.

  Variable step size sparsity adaptive
  iterative hard thresholding image reconstruction 

  DUAN Shifang,MA Shexiang

  2013, 35(8): 120-124. doi:

  Abstract ( 131 )   PDF (765KB) ( 231 )   　　

  Aiming at the shortcomings that iterative hard thresholding（IHT）reconstruction algorithm of compressed sensing theory requires the sparsity of original signal is known, a variable step size sparsity adaptive iterative hard thresholding（VSSSAIHT）algorithm was proposed. When the sparsity of original signal is unknown, the proposed algorithm according to the differences between adjacent residuals chooses appropriate step size to increase the number of support set of the reconstructed signal, approximate the sparsity of original signal gradually and restore signals by gradual iterations. Simulation results show that the VSSSAIHT algorithm, compared with the IHT algorithm, improves the quality of the reconstructed image, and reduces running time.

  Modular design of virtual experiment system
  based on MVC design model                

  ZHANG Yuan1,ZHANG Zhao1,LIU Rui2

  2013, 35(8): 125-129. doi:

  Abstract ( 83 )   PDF (597KB) ( 178 )   　　

  Based on the investigation of experimentation and aerodynamics theory, live interactive web aerodynamics virtual experiment platform is built up. Using MVC (ModelViewController) design pattern, the system has better scalability and ease of maintenance. The platform chooses ActionScript3 as a development tool, focusing on the accumulation of reusable control modules with class binding technology. It's important for the system function expansion and development of other project. The system has allaround functions, strong sense of immersion, hence satisfying the demands of experiment operation trainning and the experiment courses online study.