Current Issue

• 2014, No. 03　Published:25 March 2014 Last issue    Next issue
  Columns in this issue: 论文
• 论文

  The architecture and state consistency issues of ROIA            

  LIU Dong1,ZHAO Yuelong2

  2014, 36(03): 381-387. doi:

  Abstract ( 149 )   PDF (737KB) ( 288 )   　　

  Realtime Online Interactive Application (ROIA) is a new kind of distributed application, proposed by foreign scholars in recent years, including interactive ELearning system and Massively Multiplayer Online Games (MMOG). Studies on ROIA and MMOG have been widely carried out abroad, but few literatures in this field can be found in mainland China. Firstly, the architectures of ROIA are classified systematically, and the strengths and weaknesses within each framework are analyzed. Secondly, a key issue of ROIA, named the state consistency issue, is discussed. Finally, two basic consistency models of ROIA and the consistency maintenance of C/MS are also analyzed and described.

  Token-based license management in parallel  computer systems      

  CAO Hongjia1,LU Yutong1,2

  2014, 36(03): 388-392. doi:

  Abstract ( 111 )   PDF (525KB) ( 167 )   　　

  A tokenbased license management model is proposed for parallel computer systems. In this model, license using is separated into two steps: license allocation and license checkout, namely. Accordingly license release is separated into checkin and return. License allocation and return may be performed by the resource management system for the application software. Under this model, the use of licenses is controlled by the resource management system totally, avoiding conditions such as license checkout out of resource management system, wrong number of licenses specified for jobs, and application process hang, which results in job failures and resource wasting. Two implementation mechanisms mapping this tokenbased model to currently popular application environment are designed for legacy applications and license managers, which show the feasibility and significance of the model.

  Research of realtime multi-core  architecture for WCET analysis  

  CHEN Fangyuan1，DING Yajun1，ZHANG Dongsong2，WU Fei3， REN Xiujiang1

  2014, 36(03): 393-398. doi:

  Abstract ( 164 )   PDF (503KB) ( 187 )   　　

  With the development of technology and the increasing growth of applications and requirements in embedded realtime systems, multicore processors will be adopted in embedded realtime systems for its high performance and low power features. However, the multicore architecture is hard, even impossible to satisfy the requirement of realtime and the predictability of WCET. Starting with the shared resources, we analyze the influence of onchip shared resources and offchip resources on WCET analysis, and discuss various WCET methods. Two WCET models are introduced: multicore static WCET model and multicore hybrid WCET model. At last, the multicore design rule for the realtime application is proposed.

  Simulation on power integrity of DDR3 system
  based on dynamic target impedance               

  LI Jinwen，CAO Yuesheng，HU Jun，XIAO Liquan

  2014, 36(03): 399-403. doi:

  Abstract ( 147 )   PDF (1173KB) ( 226 )   　　

  DDR3 memory has become one of the mainstream applications in current servers and computer systems. Although many techniques such as dual reference voltage, dynamic ondie termination (ODT), flyby topology and writeleveling, have been adopt by DDR3 in order to improve signal integrity in a certain extent, it is still difficult to design and realize high data rate. Since DDR3 is a typical parallel bus structure, the simultaneous switching noise is couped with the original power noise, affecting the quality of data signals. Taking into account that the chip current is a dynamic changing and frequency related source, the paper proposes a new mixed simulation and design procedure based on target impedance and dynamic target impedance. The constant target impedance is adopted in presimulation, while the dynamic target impedance is used in postsimulation. The tradeoff between speed and accuracy of design optimization is realized, and simulation results prove the feasibility and efficiency of this method.

  An associated storage and retrieval system of massive
  Web data based on multi-attributes                  

  LUO Fang1,LI Chunhua1,ZHOU Ke1,HUANG Yongfeng2,LIAO Zhengshuang1

  2014, 36(03): 404-410. doi:

  Abstract ( 126 )   PDF (678KB) ( 226 )   　　

  Traditional Web Retrievals commonly use the fulltext search method which has good flexibility. However, as the analysis of public opinion usually needs relative information of web attributes and statistics, the traditional retrieval method can not satisfy it well. An associated storage and retrieval system based on the Hadoop platform is designed and implemented, which can offer good basic service for the analysis of public opinion. Firstly, the associated storage of web data based on HDFS is realized by machine learning. Secondly, the problem of small files storage together with the access efficiency of associated data is solved. Thirdly, the mapping between original web data and the extracted attributes is established. Finally, the retrieval of dynamic multiple attributes based on the existed indexing on HBase and the distributed local indexing are realized.

  Analysis and optimization of test and diagnosis
  process for digital microfluidic biochip                    

  ZHANG Ling1,2,KUANG Jishun2,MEI Junjin1,LIN Jing1

  2014, 36(03): 411-415. doi:

  Abstract ( 125 )   PDF (1111KB) ( 212 )   　　

  To achieve the test and diagnosis solution for digital microfluidic biochip, the test and diagnosis process is modeled, and the corresponding diagnosis test cost is computed. The test and diagnosis cost is analyzed by the tools of Matlab, and some important observations are obtained: Firstly, with the increase of the number of subarrays divided for parallel testing, the test and diagnosis cost nearly does not change at all. In other words, the number of subarrays has less impact on the test and diagnosis cost. Secondly, as the faulty probability of the single biochip cell increases, the test and diagnosis cost increases dramatically. In addition, in the process of the diagnosis, the faulty domain must be diagnosed according to the corresponding faulty probability until all the faults are located. The test and diagnosis cost of the last faulty location is the maximum, dominating the total test and diagnosis cost of the biochip. These observations provide the theory evidence and guidance for digital microfuidic biochip test and diagnosis.

  Research on dielectrics parameters calibration
  method of transmission line model

  LI Pengchong，ZHANG Keke，ZONG Yanyan

  2014, 36(03): 416-419. doi:

  Abstract ( 130 )   PDF (678KB) ( 163 )   　　

  The impact of PCB dielectrics on transmission line when the highfrequency signal travels through the PCB is described and discussed.  Furthermore, a new PCB dielectrics parameters calibration process is proposed. It collects the design data by simulation and validation based on PCB test board, then it gets the Dk and Df using this new proposed process. The test results demonstrate that such process is effective and can improve the model accuracy of transmission line.

  Single event effects applied research
  based on a simplified resistor-capacitor circuit        

  DENG Quan1,WANG Tianqi2,LI Peng1,ZHANG Minxuan1,XIAO Liyi2

  2014, 36(03): 420-425. doi:

  Abstract ( 104 )   PDF (1022KB) ( 178 )   　　

  As technology scales down to nanometer, the effect of microscopic particles on semiconductor devices becomes more and more influential. In recent years, studies of device reliability gradually attract the attention of people and a lot of researches are carried out. The paper pays attention to the effect of Single Event Upset and uses the simplified RC circuit model to study the application of simplified circuit on the basis of traditional simulation. It also summarizes the law of resistance and capacitance values that changes the electrical properties at the sensitive node electrical properties of the equivalent circuit and explores the accuracy of SEU estimation using curve IdVd. A method of predicting the adjacent position upset in single test is proposed. According to the characteristics of experiments’ IdVd curve, a prediction can be made by classifying them. Simulation results are compared with the prediction results, proving that they are the same and the prediction is accurate and valid.

  Design and implementation of output scheduling
  IP core based on packet descriptor         

  ZHAO Guohong,HU Yongting,LI Tao,SUN Zhigang

  2014, 36(03): 426-432. doi:

  Abstract ( 102 )   PDF (1100KB) ( 198 )   　　

  Output scheduling is an important QoS technique. The packetbased output scheduling technique is widely used for QoS component due to its simplicity. However, large onchip packet buffer required by the technique increases the chip area and production cost. Aiming at the problem, a novel output scheduling technique based on packet descriptor is proposed. The descriptorbased output scheduling technique can reduce storage space requirements without degrading system performance. The design and implementation of the output scheduling IP core based on this descriptor-based technique is also introduced, including the external interface of the IP core, the parameter configuration and the internal processing procedure. The experiments carried out on FPGA demonstrate the feasibility of the IP core.

  Frequency scaling policies for energy efficient routers     

  SHI Xiangjun,SONG Tian

  2014, 36(03): 433-440. doi:

  Abstract ( 125 )   PDF (1835KB) ( 202 )   　　

  With the development of the network scale, energy consumption problems of network infrastructures cannot be ignored. Studying and applying energy efficient routers is one of the important steps to develop green network. Based on the analysis of network traffic and the proof of the huge energy saving potential of finegrained frequency scaling, the system architecture of energy efficient routers is designed, which can dynamically scale the frequency. Two kinds of router frequency scaling strategies are mainly proposed, which are called microperiodical scaling policy and threshold scaling policy. And four specific frequency scaling methods are analyzed in detail. We simulate the frequency scaling policies under different traffic conditions, analyze the performance of policies under various application scenarios according to the experimental statistical data. The experiments on real traffic of access router show that about 40% of energy can be saved by using our power scaling methods, providing a reference for designing and applying of energy efficient methods on routers.

  High-efficient signcryption scheme based on identity           

  DENG Lunzhi，WANG Xiangbin,QU Yunyun

  2014, 36(03): 441-445. doi:

  Abstract ( 114 )   PDF (434KB) ( 223 )   　　

  Signcryption offers authentication and confidentiality simultaneously with a lower cost. A new identity based signcryption scheme is proposed, which is proved to be secure under the random oracle model. The proposed scheme does not require pairing operation in signcryption, and only have 2 pairing operations in unsigncryption. Compared with the previous schemes, the computation cost of the proposed scheme is less.

  An algorithm to solving minimum double-cost flow problem     

  MA Yubin,XIE Zheng,CHEN Zhi

  2014, 36(03): 446-451. doi:

  Abstract ( 110 )   PDF (1026KB) ( 162 )   　　

  Multi-objective optimization is a critical part of network optimization. The paper derives a minimum doublecost flow problem in double-cost network model with limitations on capacity from a practical case. According to the thought of hierarchy and the feature of double-cost network, the corresponding minimum double-cost flow problem is introduced and a two-layer primaldual algorithm is proposed to solve it. The correctness of our algorithm is proved and its complexity is estimated as  O(n2v0). Besides, the algorithm is improved to solve the minimum kcost flow problem in k-cost network. Finally, a case is used to exhibit how the algorithm works.

  An efficient certificateless proxy blind signature scheme          

  WEN Jiajun,ZUO Liming,LI Biao

  2014, 36(03): 452-457. doi:

  Abstract ( 125 )   PDF (408KB) ( 165 )   　　

  Considering that multiple references found that the existing scheme is not efficient, based on reference［5］, it proposes a certificateless proxy blind signature scheme with high efficiency. The scheme is based on the appropriate modification of partial key to reduce the operation time of bilinear pairings. Meanwhile, due to the continuous improvement of the network service, replacing the operation time by the communication time can improve the efficiency. By comparing the proposal with various signature algorithms and verification algorithms, it is proved that our scheme improves the efficiency greatly and satisfies unforgeability, nonrepudiation, identification, blindness, etc.

  An optimal routing algorithm for a class of  directed double loop network             

  LIU Wangfei，CHEN Baoxing，YUE Hao

  2014, 36(03): 458-462. doi:

  Abstract ( 102 )   PDF (759KB) ( 164 )   　　

  Directed double loop network G(N;1,h), where N is the number of its nodes, 1 and h are its steps, is an important interconnection network. Some properties of G(N;1,h) are given. As two applications of these properties, a diameter formula for this network is given. An optimal and simple routing algorithm for a class of directed double loop network is also obtained.

  A single population shuffled frog leaping algorithm       

  WANG Lianguo1,2，GONG Yaxing2

  2014, 36(03): 463-468. doi:

  Abstract ( 102 )   PDF (685KB) ( 184 )   　　

  Aiming at the shortcomings of shuffled frog leaping algorithm(SFLA) such as ease of trapping into local optimum, low optimization precision and slow speed when it is used to optimize some functions, a Single Population Shuffled Frog Leaping Algorithm (SPSFLA) is proposed. Without sorting the whole population, this new algorithm adopts single population. The individuals are updated by learning from the global best individual and the global middle position. If the current individual is not improved, the global best individual will be mutated and the current individual will be replaced by the new one. Those enhance the running speed, the convergence rate and the optimization precision of SPSFLA. The simulation results show that the improved algorithm has better optimization performance.

  Quantitative safety evaluation of aviation equipment
  embedded software based on S-curve model                

  WANG Hongpo1,2,ZHOU Hongjian1,WANG Honglei2

  2014, 36(03): 469-474. doi:

  Abstract ( 118 )   PDF (689KB) ( 179 )   　　

  Software testing is an important phase during the lifetime of aviation equipment embedded software. It is difficult to evaluate the effect of software testing. Based on software testing，discusses the occurrence law of the safety defect existing in aviation equipment embedded software. It is believable that the occurrence law of the embedded software also satisfies the Scurve rule. AML model is introduced to build up the Scurve model of a kind of aviation equipment embedded software based on the defect data gathered during the software testing period. The total defect number of the software is predicted. Finally, ChiSquire verification and prediction error metric are applied to verify the goodness of fit of the software defects. Results show that AML model can model the real safety defects finding process. It eliminates the effects of randomness of the data, so the model is reliable. The predicting data can be trusted to some extent. It can be used to model and analyze the safety of aviation equipment embedded software.

  An improved automatic CSS analysis method      

  XU Ke,ZHANG Dafang,ZHAO Dan

  2014, 36(03): 475-480. doi:

  Abstract ( 107 )   PDF (624KB) ( 164 )   　　

  CILLA is an automatic Cascading Style Sheets (CSS) analysis method by obtaining Document Object Model (DOM) tree dynamically. It can find out the effectless and useless CSS code snippets in Web applications. However, there is a high false positive rate when using CILLA to analyze CSS with PseudoClasses, CSS with attribute selector, CSS declared in the attribute of page element, or CSS generated automatically. An improved CILLA method is proposed to solve these problems. The experimental result shows that the improved method can reduce 20% false positive rate on average.

  Similarity analysis of malware’s function-call graphs                 

  LIU Xing,TANG Yong

  2014, 36(03): 481-486. doi:

  Abstract ( 159 )   PDF (1470KB) ( 249 )   　　

  The similarity analysis of malware is an important part of the current automatic analysis of malware. The paper proposes a new method of similarity analysis of malware based on functioncall graphs. This method uses the similarity distance of malware’s function-call graphs (called SDMFG) to measure the similarity of two malwares’ function-call graphs, and then analyzes the similarity of the two malwares. This method improves the accuracy of similarity analysis of malware, providing a strong support for analysis of the homology and evolution characteristics of malware and malware detection and prevention.Key words:

  Research of controlling host security
  login system using smart phones           

  GONG Jianping1,XU Yajuan1,GONG Bingchen2

  2014, 36(03): 487-490. doi:

  Abstract ( 109 )   PDF (643KB) ( 173 )   　　

  Controlling host security login system using smart phones is a new generation signon technology that combines intelligent terminal and password mechanism. Firstly, the relationship between Winlogon and GINA and the basic Windows system interactive logon principle are introduced. Secondly, the paper propose a system that a smart phone（based on Android platform）is used as an authentication factor carrier to control the host logon system under the windows platform. And the system's critical technique and advantages are described. The system uses the powerful functions of the application interface and the bluetooth communication of smart phones to realize the twofactor authentication mechanism with both the phone and PIN, largely improving the safety of the system as well as avoiding extra investment and inconvenience of carrying the phone.

  Image edge detection method of underground
  objects based on improved Canny operator             

  SHANG Changchun,MA Hongwei,AN Jingyu

  2014, 36(03): 491-496. doi:

  Abstract ( 117 )   PDF (1020KB) ( 252 )   　　

  In view of the shortage of the traditional Canny algorithm in detecting the low intensity edge capacity, the  improved edge detection method is carried out from the following three aspects: (1) Using a new fourorder partial differential equations of the noise reduction algorithm for image denoising, it can further improve the noise reduction effect and better preserve image details in the noise reduction process, so that the underground objects are more easily detected. (2) Using an adaptive threshold method for the image edge detection, the method realizes the adaptive dual threshold extraction, which can effectively extract the real edge. Especially in the edge extraction of low contrast image, this method has more advantages.  (3) Using the theory of fuzzy decision, an effective edge connection method is proposed based on the traditional Canny algorithm. Finally, in order to verify the effect of Canny edge detection operator, Prewitt, Robert, Sobel, and traditional Canny algorithms are used to perform the underground image edge detection test. The results show that the new method can detect more lowintensity edge and inhibit the noise at the greatest degree. It gives the foundation for the robot image recognition in coal mine.