Current Issue

• 2013, No. 5　Published:25 May 2013 Last issue    Next issue
  Columns in this issue: 论文

• 论文

  Design and implementation of a distributed-controlled dual-clock FIFO

  YANG Jianxin,HU Xiangdong,LI Yuan

  2013, 35(5): 1-8. doi:

  Abstract ( 198 )   PDF (1485KB) ( 246 )   　　

  Dual-clock FirstInput First-Out (FIFO) structure is very useful for transferring data between modules operating in different clock domains. As more and more independent clock domains are integrated onto a SystemonaChip (SoC), the traditional dualclock FIFO structure increases the design complexity of clock networks, thus blocking the increase of the chip size and the frequency. The paper proposed a distributedcontrolled dualclock FIFO structure. The proposal uses the technology of sourcesynchronous data transfer so as to avoid the design complexity of clock networks, which is induced by distributing the sender's clock into the receiver's clock domain. The paper implemented this design and further introduced its optimization according to the performance and the realizability.

  Design of bufferless resonant clock
  distribution network with low power and Low skew

  XU Yi，CHEN Shuming，LIU Xiangyuan

  2013, 35(5): 9-14. doi:

  Abstract ( 132 )   PDF (725KB) ( 183 )   　　

  As the powerful competitor of the next generation clock distribution technology, resonant clock scheme shows obvious advantages in low power consumption and low skew. But the application of this technology is limited because of the absence of a systematic design methodology. To solve the problem, a  physical design method of bufferless resonant Clock Distribution Network (CDN) was proposed in this paper. The method uses the conventional synchronous design flow and implements resonant clock network efficiently. Based on SPICE analysis and optimization for the design parameters of the resonant CDN, the proposed method makes the physical design converge fast at the target frequency. A multiplier circuit was designed using this method, and the results of post-layout SPICE show that, compared with the synchronous design employing treebased and meshbased CDN, the bufferless resonant CDN based circuit can save CDN power up to 64% and total power up to 16%. Additionally, the clock skew in resonant CDN is less than 2% of clock cycle under different process corners.

  Applying data filling to alleviate the sparsity
  problem for personalized recommendation

  XIA Jianxun1,2,4，WU Fei2,3,4，XIE Changsheng2,3,4

  2013, 35(5): 15-19. doi:

  Abstract ( 151 )   PDF (517KB) ( 172 )   　　

  Till now, collaborative filtering has been the most successful and widely used technology in recommender systems. However, the rating data is extremely sparse so as to affect the prediction accuracy seriously in traditional collaborative filtering. In order to overcome the drawback, in this paper, we proposed three data filling approaches and two recommendation strategies. These data filling approaches for non-rating data in rating matrix are: (1) Filling data using weighed average of row and column data; (2) Filling data using mode average of row and column data; and (3) Filling data using median average of row and column data. One of the recommendation strategies is taking filling data for predicative rating directly, and another is to set the rating matrix filled data as a pseudo rating matrix and collaborative filtering applying Pearson correlation. The experimental results on the MovieLens data set show that all these recommendation strategies can effectively alleviate the trouble of rating data sparseness and gain better recommendation accuracy.

  I/O optimization in ErasureCode distributed
  file system for massive data storage

  YAN Lin1,2,XING Jing1,HUO Zhigang1,MA Jie1

  2013, 35(5): 20-27. doi:

  Abstract ( 148 )   PDF (804KB) ( 199 )   　　

  As the rapid growing of massive data, the storage method of cluster file system is developing from replication to Erasure Code. The storage system based on Erasure Code can provide higher reliability with less storage overhead. However, in the procedure of storing original data and the coded data, storage based on Erasure Code faces more disk I/O conflicts and unbalanced load, which jeopardizes the throughput of the system. Specially, there is a tradeoff between data consistency and data caching in the storage system based on Erasure Code when writing back the parity. And there are limitations for the use of non-datacaching machanism and alldatacaching machanism in the cluster file system. For these two issues, the paper proposed an I/O optimization method, which includes the data layout machanism balancing the load and the parity consistency machanism. In the cluster file system ECFS we developed base on Erasure Code, the throughput of the system can be improved by 95.53% after exploiting the I/O optimization.

  Research of Agent based context quality
  management in pervasive environment

  ZHENG Di1,WANG Jun2,CAI Jianyu3

  2013, 35(5): 28-33. doi:

  Abstract ( 144 )   PDF (967KB) ( 165 )   　　

  With the rapid development of information technology, it is inevitable that the distributed mobile computing has been gradually evolving to the pervasive computing, whose final goal is fusing the information space composed of computers with the physical space where the people are working and living. To achieve this objective, one of the problems is how to continuously monitor/capture and interpret the environment related information efficiently so as to assure high context awareness. Many attentions have been paid to the research of the contextaware pervasive applications. However, most of them just use the raw context directly or take just some aspects of the Quality of Context (QoC) into account. Therefore, proposes an agent based contextaware framework that support QoC management in various layers. By this framework, we can use threshold Agent, quality factor configuration Agent, duplicate and inconsistent management Agent to implement control mechanisms for various layers, thus providing contextaware services and application users with efficient and reliable context information.

  Design and implementation of an asynchronous bus: PABLE

  ZHANG Guangda,WANG Yourui,SHI Wei,WANG Zhiying,LU Hongyi

  2013, 35(5): 34-40. doi:

  Abstract ( 141 )   PDF (1230KB) ( 186 )   　　

  Asynchronous circuit can resolve clockcaused problems in synchronous circuits, such as clock skew and high energy dissipation, attracting increasing attention. In order to implement full asynchronous communication among asynchronous modules on a chip and take advantage of asynchronous circuits in power consumption and performance, the paper designed an asynchronous bus PABLE (pipelinebased asynchronous bus for low energy), which is partially compatible with the synchronous AMBA AHB protocol and uses the asynchronous pipeline to improve the performance. Asynchronous arbitration circuit was designed, which can eliminate metastability. Finally, the desynchronization method was adopted to implement the PABLE. Results of the experiment show that, under the UMC 0.18um CMOS technology, for a single read or write operation, the read or write latency of the PABLE is lower than the synchronous bus in more than 60% cases and its average power consumption decreases by 41% compared with the synchronous bus.

  Verifiable threshold multisecret sharing
  scheme based on discrete logarithm problem

  WANG Xuejun1,2，GAO Caiyun1，CAO Tianjie1,3

  2013, 35(5): 41-45. doi:

  Abstract ( 106 )   PDF (421KB) ( 155 )   　　

  The traditional secret sharing scheme cannot verify the fraud of participants or secret distributors. The secret sharing scheme proposed by LinWu et al. can verify these frauds, which requires a lot of computation. The secret shares in this scheme must be calculated which can not meet the randomness of shares. It needs to establish secure channels between the secret distributor and each participant, which increases the burden of the implementation of the scheme. Therefore, an improved secret sharing scheme was proposed in the paper. Participants do not need to verify fraud by the secret distributor, which reduces the calculation of participants. The secret shares are chosed by the participants, which achieves the randomness of secret shares. In addition, the scheme can share multisecret sharing and update the group secrets. It also can dynamically add new participants.

  Research on trustworthy model based on user behavior  

  OU Wei1,WANG Yongjun1,HAN Wenbao2

  2013, 35(5): 46-50. doi:

  Abstract ( 104 )   PDF (523KB) ( 169 )   　　

  Traditional authorization and authentication solves the problem of trust of user identify, but not solves the problem of trust of user behavior. Therefore, the paper proposed a new trustworthy model based on user behavior so as to meet the needs of security systems. In this model, users of network service are regarded as subjects, and files and data in computer are regarded as objects. Different subjects and objects are assigned to different levels of confidentiality and integrity. A user' service request on system is mapped into the access from a subject to an object. In this model, security attributes and security strategies and rules used to restrict users' behavior are all defined. Hence, a security and trustworthy computer system is made up.

  Research of IPv6 path reconstruction
  algorithm based on Huffman code 

  HU Qingzhong,ZHANG Bin

  2013, 35(5): 51-55. doi:

  Abstract ( 117 )   PDF (533KB) ( 184 )   　　

  Packet marking is a commonly used IP traceback algorithm, which marks the path information into the IP header and is able to reconstruct the attack path according to the marked information so as to find the attack source. The marked information is not much due to the limited mark area, so several marked packets are required to reconstruct an attack path, resulting in high complexity, low efficiency and poor accuracy of path reconstruction algorithms. To address this problem, a path reconstruction algorithm based on Huffman code was proposed, which marks the link information into the mark area and does not need to store the marked information into intermediate nodes. The proposed algorithm is applicable to IPv6 networks and only requires a single marked packet to reconstruct the entire attack path. The experiment shows that the proposed algorithm is fast and has high efficiency and accuracy.

  A solution of operating systemlevel virtualization
  for Linuxbased mobile handsets and its advantages        

  YE Changchun

  2013, 35(5): 56-60. doi:

  Abstract ( 110 )   PDF (465KB) ( 171 )   　　

  Mobile handsets are more and more widely used, but they must own more security, more availability and more personalization. This paper proposed a solution that constructs virtual environments based on operating systemlevel virtualization for Linuxbased mobile handsets, such as Linuxbased mobile phones, Linuxbased Mobile Internet Devices (MID) and Linuxbased notebooks, whose market share has been enlarging vastly today. It can isolate security threats and software faults from other parts of the entire system, avoid total malfunction caused by an outoforder application, and support personalization by providing operating environments adapted to usage situations. The proposed OSlevel solution does not support different operating systems running on mobile handsets simultaneously, which is allowed by hardwarelevel paravirtualization. However, the OSlevel solution has several advantages over the latter. These advantages are saving software costs for users, not increasing users' learning costs, and least costs and shorter period of research and development.

  Research and design on video transcoding 
  scheduling algorithm based on LVS

  LIU Gefei,YANG Qing,ZHU Xiaolu

  2013, 35(5): 61-66. doi:

  Abstract ( 143 )   PDF (458KB) ( 155 )   　　

  With the rapid development of the triple play, mobile devices can access various video information on Internet by using 3G networks. However, it is not convenient for users that there are various kinds of video formats. An advanced task scheduling algorithm supporting dynamic loading balance (ADLBA) is  proposed based on the research and analysis of Linux Virtual Server (LVS). We implement it to prove its feasibility and practicality.

  A featureoriented approach to modeling
  nonfunctional requirements in software product line      

  SUN Lianshan,WANG Jinyu

  2013, 35(5): 67-76. doi:

  Abstract ( 130 )   PDF (1063KB) ( 231 )   　　

  Analyzing traditional definitions of NonFunctional Requirements (NFR), this paper redefined NFR as constraints on 'requirements model' that is an available system abstraction at requirements analysis stage. The proposed definition formulates the relationships among NFR and functional requirements. Furthermore, this paper extended a traditional featureoriented software product line requirements modeling approach to explicitly model NFR, functional requirements, the type of NFR, as well as various relationships among them, and to model NFRs variability in the same way of modeling variability in traditional feature model. The proposed approach enables explicitly compilation and customization of NFRrelated assets across projects, and is a solid basis for quantitative evaluation on the quality of the derived products. Furthermore, we presented a multipleview based and featureoriented approach to guide developers to iteratively model both NFR and functional requirements, and meantime to achieve goals including separation of concerns and model complexity management. Finally, a tool prototype was implemented and a case study was conducted.

  Software agent architecture of
  policy-based self-organizing systems and its implementation       

  HOU Fu,MAO Xinjun,GUO Yi,ZHOU Kai

  2013, 35(5): 77-86. doi:

  Abstract ( 101 )   PDF (1005KB) ( 144 )   　　

  Currently, the great challenge of selforganizing software engineering is how to design the proper interactions among agents to satisfy the global macro emergence of selforganization system. To solve this problem, the paper proposed a policybased selforganizing multiagent system development method in which we use policies to regulate agents' behaviors so as to obtain the expected emergent results in a more effective way. However, the key point that supports this development method is how to construct the software agent that can not only perceive and understand the system policy, but also automatically decide and adjust its behaviors conforming to the policy. In this paper, we proposed the BeliefDesireIntentionPolicy (abbreviate BDIP) agent architecture for policybased selforganizing multiagent systems. We also designed the running mechanism as well as the behavior decision algorithm of the BDIP agent to execute the policy. To illustrate our approach and show its effectiveness, we constructed the software antitype of the agent. Case study also demonstrates the effectiveness of the architecture and the running mechanism.

  Bandwidthadaptive tracking algorithm
  based on Mean Shift and Kalman prediction      

  WANG Wenjiang1,HUANG Shan1,2,ZHANG Hongbin2

  2013, 35(5): 87-92. doi:

  Abstract ( 130 )   PDF (666KB) ( 249 )   　　

  As a widely used traditional tracking technique in visual surveillance, Mean Shift algorithm has a deficiency in handling moving targets with high speed or large scale change. In order to sove this problem, a bandwidthadaptive tracking algorithm based on Mean Shift and Kalman prediction was proposed. The algorithm uses Kalman filter to predict the positions of fast moving objects in the successive frame, which are as the initial positions for Mean Shift iteration. Bandwidth trials is utilized to adjust the bandwidth automatically for targets' scale change. The experimental results of pedestrians and vehicle tracking show that our algorithm is effective and robust.

  A new character feature vector similarity function 

  LI Yucheng,TIAN Zhen,YOU Jia

  2013, 35(5): 93-99. doi:

  Abstract ( 117 )   PDF (1089KB) ( 181 )   　　

  To analyze the feature vectors of vehicle plate characters and its extraction method, this paper proposed a linear similarity function definition sim(M,N) based on cityblock distance, discussed the properties of this similarity function, and gave the mathematic explanation of completely dissimilarity. Compared with some common similarity functions,sim(M,N)can describe feature vector differences linearly and the computation is very easy. Meanwhile, the paper discussed the differences between Euclidean distance and sim(M,N) in analyzing feature vector differences, and proved the shortage of Euclidean distance through experiments. In the average character margin experiment of similar vehicle plate characters, sim(M,N) is demonstrated to be better than other similarity functions.         

  Rapid structural road elements detection based on pixels    

  BI Jianquan1,2,LU Geng1,JIANG Yongheng1,WANG Jingchun1

  2013, 35(5): 100-105. doi:

  Abstract ( 105 )   PDF (967KB) ( 170 )   　　

  The basis of the autonomous driving system of the microvehicles based on simple visual is the road element detection. Based on the lack of the microvehicles processing capacity and the realtime requirements of the detection and control algorithms, the paper proposed a new feature extraction and tracking algorithm based on the pixel data blocks. We use the fixed OTSU threshold image binarization to reduce the impact of light. The detection experiments show that the rapid structural road elements detection(RSRED) algorithm can extract the lane markings, road signs, zebra crossings and other road elements against the interference of light, shelter, pollution and the vehicle speed definitely influence over detection rate of road elements. Finally the Lanekepping and Overtaking experiments show that the algorithm is efficient and stable.

  Color image scrambling based on Zigzag Transformation     

  LIU Gen1，JIANG Tianfa1,JIANG Wei2

  2013, 35(5): 106-111. doi:

  Abstract ( 177 )   PDF (1100KB) ( 169 )   　　

  Most of the traditional digital image scrambling algorithms belong to space domain scrambling, which disorder the positions of pixels by some transformation. The space domain scrambling is very simple but practicable. It only disorders the positions of pixels, but can not change the colors of every pixels.The scrambling algorithms based on color domain can change the statistic feature of image, but can not change the permutation feature of image. These algorithms are not secure, because it is easy to guess the outline of original image. In order to avoid these problems, the paper proposed a scrambling algorithm based on dual Zigzag transformation, which belongs to both space domain scrambling and color domain scrambling. Experiment results show that this algorithm can not only change the outline of image, but also change the statistic feature of image. In addition, the algorithm has a good image scrambling effect, high safety, and it is simple and easy to realize.

  Modular nonparametric feature analysis for face recognition     

  CHING Qiang,CHEN Weiqi

  2013, 35(5): 112-117. doi:

  Abstract ( 98 )   PDF (586KB) ( 165 )   　　

  Based on the nonparametric feature analysis (NFA) method, the paper proposed a modular NFA algorithm and applied it on face recognition. Firstly, the modular NFA algorithm divides the original images into modular subimages. Secondly, the NFA method is employed to extract the features of the modular subimages. There are two advantages: 1) local feature of the images can be extracted efficiently, and it has outstanding effect for the images that have large variations in facial expression and illumination; (2) Singular value decomposition of matrix may be avoided in the process of feature extraction, which is simpler than that of other technologies such as NFA. Meanwhile, it overcomes the small sample size problem. Moreover, NFA is a special case of modular NFA. Experimental results on ORL and YALE face databases show that the performance of modular NFA is superior to NFA and modular NSA.              

  Image denoising based on K-Means
  clustering and binary tree decision        

  LIU Yongxia1,ZHANG Zhaohui1,2,ZHANG Yanmin1

  2013, 35(5): 118-123. doi:

  Abstract ( 129 )   PDF (1412KB) ( 152 )   　　

  In this paper, a new filter algorithm was proposed for pepperandsalt noise suppression. Firstly, the neighborhood of each given pixel is partitioned by Kmeans clustering according to the local grey level distribution. Secondly, the recognition rules for noisepolluted pixel detection are constructed, and the noise pixel can be detected based on multilayer binary tree decision. The proposed algorithm only filters the recognized noise pixels without changing those nonpolluted pixel values. Experimental results show that the proposed algorithm can efficiently preserve informative details when filtering image noise. For those images with strong noise pollution, the proposed algorithm outperforms both median filter and the algorithm proposed in［7］.

  Edge detection of IHC tissue images based on local color entropy    

  XU Qing,FU Rong

  2013, 35(5): 124-129. doi:

  Abstract ( 115 )   PDF (1270KB) ( 189 )   　　

  The immunohistochemical (IHC) image segmentation often appears some problems caused by uneven or overlapping stain, resulting in bad edge detection. This paper proposed a local color entropy method based on the properties of R，G，B color channels in IHC images. For IHC images, the better result of entropy for edge detection in this paper can be obtained by adding color information for image entropy in order to target areas which have a different color but approximate grey scale, and a good edge detection can be got by OTSU. By comparing the result of Matlab simulation experiment and that of traditional image entropy for edge detection, this method can be proved that it efficiently extract the contours with a high ratio for detection of targets.