Current Issue

• 2012, No. 11　Published:25 November 2012 Last issue    Next issue
  Columns in this issue: 论文

• 论文

  Categorization and Extraction of Web Pages Based on Hierarchy

  WANG Zhenyu1,TANG Yuanhua1,GUO Li2

  2012, 34(11): 1-6. doi:

  Abstract ( 193 )   PDF (637KB) ( 258 )   　　

  Traditional web crawler provides services based on searching keywords. It cannot extract the categorization information of web pages, thus resulting in efficiency and accuracy problems on text clustering and topic detection. To solve this problem, a method of categorization and extraction of web pages based on hierarchy is proposed in this paper. By building a virtual hierarchy categorization tree and extracting the hierarchies of real web sites, a web page is categorized when it is crawled. For sites which have no categorization information, a page title based categorization algorithm is presented, including building up the domain knowledge base and calculating the semantic similarity based on Hownet. The experimental results demonstrate that this method achieves preferable effects.

  QoS Simulation and Research for Wireless Sensor Networks

  GAO Hongliang1,2,WANG Bingwen1,GAO Chao1,HU Xiaoya1

  2012, 34(11): 7-13. doi:

  Abstract ( 176 )   PDF (697KB) ( 281 )   　　

  Wireless sensor networks (WSN) can be broadly used in military and civilian areas. Due to the limitations of inherent characteristics of sensor nodes and lack of high level protocols and abstract algorithms suitable for the speed of processing large networks, it is crucial for the WSN to satisfy the demand of the user's quality of service (QoS), and the research of QoS in real applications has become a hot issue. The paper firstly introduces the main QoS parameters of the WSN, secondly analyzes the QoS parameters such as energy efficiency, packet loss rate, throughput and transmission latency in three kinds of typical topology structures, thirdly builds up a general simulation platform for the WSN using NS2, and finally simulates and analyzes the main QoS parameters of the WSN on the platform. The simulation obtains effects on QoS parameters under different kinds of topology structures.

  A Lightweight MultiLevel Capabilities Mechanism for Next Generation Internet

  ZHANG Honghao,WANG Jinsong,LIU Tao

  2012, 34(11): 14-20. doi:

  Abstract ( 191 )   PDF (1132KB) ( 247 )   　　

  An antiDoS (Denial of Service) mechanism called LMCM (Lightweight Multilevel Capabilities Mechanism) for next generation Internet is proposed. The LMCM distinguishes the malicious users and the benign users through their behaviors and adopts lightweight validation mechanism to avoid heavyweight operations in the core network. It improves data transfer efficiency but not lowers the overall security, meeting different security requirements. In order to defend DoC (DenialofCapability) attacks caused by the capabilities and guarantee fairly sharing the network resources, the LMCM adopts a hierarchical queue management mechanism. Furthermore, the LMCM improves the flow control mechanism to defend other complicated attack which cannot be defended in TVA（Traffic Validation Architecture） and makes up for the shortcomings and inadequacies of the TVA. In order to get convincing comparative results, we choose some representative topologies in the dataset of the CAIDA (Cooperative Association for Internet Data) as our experiment topologies. Simulation results in dissimilar scenarios indicate that the LMCM is conducive to improving the data transfer efficiency and enhancing the scalability of defense system compared with the TVA.

  A Lossless Robust Watermarking Scheme for Color Image

  JIAN Huan1,CHEN Zhigang1,DENG Xiaohong2,DENG Xiaoheng1,QI Huamei1

  2012, 34(11): 21-27. doi:

  Abstract ( 165 )   PDF (1032KB) ( 278 )   　　

  In order to solve the fragile property of lossless watermarking algorithms in spatial domain, a new robust reversible watermarking method is proposed. Firstly, the original color image is separated into three primary colors channels: red (R), green (G) and blue (B). Then, each channel is divided into nonoverlapping blocks and the arithmetic difference is computed. The watermark information is embedded into channels repeatedly by shifting arithmetic differences according to the given rule. In the extracting procedure, the three layer watermark information can be extracted from three primary colors channels of stegoimages respectively, and the final water information is generated based on voting theory. Experimental results show that the proposed scheme can resist JPEGcompression, the Gaussian noise, salt and pepper noise attack, and has better robustness.

  Study on the Personalized Services Based on Identities and Context

  XIAO Laiyuan，YIN Ming

  2012, 34(11): 28-33. doi:

  Abstract ( 174 )   PDF (566KB) ( 232 )   　　

  Because of the diversity of their environment, people tend to require for more personalized services which has become an important subject in information age. In order to achieve the purpose of the recommendation, personalized service collects and analyzes user interests and behaviors. In this paper, we present an implementation case of financial services called easy golf. Through easy golf, users can easily order discount information to meet their own demands. They can also make reasonable financial plan according to the optimal suggestions provided by the system.

  Research on the Asynchronous Network Security Communication

  SHANG Xinna,SUN Lianying,PENG Tao,LIU Chang

  2012, 34(11): 34-37. doi:

  Abstract ( 164 )   PDF (568KB) ( 347 )   　　

  Due to the limitation of network,the data communication security between clients and servers is an important issue in the asynchronous network system.The paper deeply analyzes the common encryption technologies,such as RSA,DES,MD5,and put forward a new communication security model based on their limitations.The proposed model applies multiple encryption technologies comprehensively to improve the quality of raw data,the reliability of data sources and whereabouts,thus increasing the security of data communication.

  Security Measurement Method of LAN Based on Attack

  LI Xiaoyan1,2,WANG Qingxian1,YANG Lin2,ZHU Yun3

  2012, 34(11): 38-45. doi:

  Abstract ( 168 )   PDF (977KB) ( 267 )   　　

  For making more scientific and comprehensive quantitative measure on the security of the LAN, the idea of measuring the defense performance byusing attack effect is proposed. The malicious code, network attacks, information sabotage attack, information attack, and other content security attack scenarios are designed. Attack target set containing attack duration, control duration, interference duration, infection rates are proposed.  Based on the attack target set, a network defense performance model using AHP analysis method is presented. Simulation results show that the proposed measurement method and model can better measure the security of network protection systems.

  An Approach for Dynamic Role Transition in Kylin OS

  LI Wenbo，LIAO Xiangke，HE Lianyue，LI Shanshan，FAN Xiaokang

  2012, 34(11): 46-49. doi:

  Abstract ( 167 )   PDF (441KB) ( 226 )   　　

  In the paper, based on the Role Based Authorization framework (RBA) in secure Operation Systems (OSs), we propose an approach to dynamic role transition in Kylin OS, using the way of trustworthiness measurement for the uncertainties in authentication systems. In the model, through quantifying the uncertainty of authentication and constraining the bound of role transition, the process of role transition can be completely transparent to users, thus guaranteeing the safety and stability of OS.

  A (α, k)Anonymity Method Based on Social Networks

  ZHANG Xiaolin，WANG Ying，LI Yufeng

  2012, 34(11): 50-54. doi:

  Abstract ( 164 )   PDF (543KB) ( 275 )   　　

  With the rapid development of Internet technology and the establishment and use of online social networks,more and more people attend in social networks to share information.In this interactive process,a large number of data are produced.Some of these data are a perosn’s privacy so that they should not be known by other individuals.The privacy protection of the released data in the social network becomes a emerging research subject.The paper proposed the (α, k)Anonymity method of the social network. It uses the clustering method to protect the node’s attribute and the relationship between nodes. Each cluster have at least k nodes,and in each cluster the percentage of any sensitive attribute value associated with the node is not higher than α.Theoretical analysis and experimental results show that the proposed social network (α, k)Anonymity method can effectively protect the privacy protection in the condition that the information loss is as small as possible.

  An AIK Certificate Generation Protocol in Remote Attestation Based on ZKV

  PENG Xinguang,JING Yanying

  2012, 34(11): 55-61. doi:

  Abstract ( 178 )   PDF (579KB) ( 205 )   　　

  There are some defects,such as platform information exposure and trusted third party efficiency bottleneck, in AIK certificate generation protocol.This paper proposes a method that combines three technologies,zero knowledge proof, Kerberos and Virtualizing Trusted Platform, to solve this problem, we call this method as ZKV. We build a prototype system, remote proof process, and verify the ZKV efficiency.

  Research on PeertoPeer Network Trust Models

  WANG Qinglian,WU Guodong,SUN Junjie

  2012, 34(11): 62-67. doi:

  Abstract ( 182 )   PDF (529KB) ( 321 )   　　

  Due to the openness,anonymous and other characteristics of P2P networks,there is a considerable security risk in the communication and services between the nodes.As a solution,trust model has been extensive researched and developed in recent years.In this paper,we summarize and analyze the four types of the trust model for P2P networks:super-peer,recommendation,feedback and game theory,and give their features,advantages and disadvantages.At last we put forward the research prospect of the P2P trust model.

  Research on Sequential Pattern Mining in Intrusion Detection System

  LI Liang

  2012, 34(11): 68-71. doi:

  Abstract ( 257 )   PDF (455KB) ( 237 )   　　

  The Intrusion Detection System (IDS, Intrusion Detection System) is an important component of the computer security architecture. With the increasing data flow of network, the intrusion detection system that is combined with the data mining has become a research hotspot. An improved PrefixSpan algorithm based on time constraints and attributerelative is developed to fit into the network audit data of computer intrusion detection. The experiment of dealing with a computer network audit database gives the application and result analysis of the algorithm.

  The Memory Access Characteristics of the Cell Processor

  ZHENG Yi，DENG Lin，DOU Yong

  2012, 34(11): 72-76. doi:

  Abstract ( 189 )   PDF (1196KB) ( 274 )   　　

  The Cell processor is a typical heterogeneous multicore processor with powerful computing ability. However, we are facing the challenges of 'memory wall' in developing parallel applications, such as, limited capacity of local memory, limited memory bandwidth for multicores and the large latency for data communication. The DMA transfer mechanism is often used to hide the large latency and improve the utilization of memory bandwidth. In the paper, we start with a series of DMA experimental tests in the context of the Cell processor architecture, and perform mathematical analysis to setup a unified formula on the average bandwidth of DMA by means of exponential fitting, which describes that SPE amount and DMA block size take main effects on DMA bandwidth in quantity.

  A Hybrid Time Management Approach for Social Network Simulation

  ZHANG Yingxing,YAO Yiping

  2012, 34(11): 77-82. doi:

  Abstract ( 163 )   PDF (1086KB) ( 228 )   　　

  Parallel Simulation Technique plays a more and more important role in researching large scale social networks. Synchronization is the crucial factor to affect the parallel simulation performance. Nowadays, it is an important topic of synchronization study to exploit application characteristics so as to develop an effective synchronization algorithm. Due to the larger differences of individual behaviors in the large scale social networks, the conservative approach may become limited by sensitivity to lookahead; due to the complex individual interaction, the optimistic approach appears to be prone to cascading rollbacks. In this paper, we propose a hybrid time management approach where a logic process can autonomously choose to be optimistic or conservative, and optimize the selection by utilizing the community detection. The experimental results demonstrate the efficiency of our methodology.

  Approximation Algorithms:A Measure Perspective

  WANG Gang,LUO Zhigang,LI Cong,HUANG Xuhui

  2012, 34(11): 83-90. doi:

  Abstract ( 170 )   PDF (519KB) ( 247 )   　　

  Standard approximation, differential approximation, and domination analysis are three different measures of approximation algorithms. The standard approximation ratio measures the relative error between the optimal solution and the approximate solution. Differential approximation pays attention to the position of the approximate solution in the interval defined by the optimal solution value and the worst solution value. Domination analysis considers the rank of the approximate solution among all feasible solutions. We survey concepts and main results in this domain, and also advantages and disadvantages of the three measures. Special focus has been placed on inapproximability results such as those based on the PCP theorem and the Unique Games Conjecture.

  Cable Deformation Simulation Based on B-Spline Curve

  MA Hongquan1,ZHANG Jinrong2,ZHANG Xueying1,YE Tingting1

  2012, 34(11): 91-95. doi:

  Abstract ( 214 )   PDF (552KB) ( 326 )   　　

  Cable Deformation Simulation is the core problem for both computer aid design and maintenance process simulation of cables. It is one of the research hotspots in virtual reality techniques. B-Spline curve is one of significant achievements in computational geometrics. It has been widely used in computer aided design. The paper firstly introduces the concept of BSpline  And analyzes the advantages of simulating the cable deformation with Cubic uniform BSpline curve. Secondly, the GlobalEnergyOptimization method and the HighlyCoordinateReset method are combined in order to search for the position of the control points, and a cable model with two given end points and one giving passingthrough point are established. Finally, based on that, other cable models with constraints of given tangents are built up by the Lagrange Multiplier method. This model describes the cable deformation under stable status, featuring small amount of calculation, lifelike model and flexible control of deformation. It can be used in cable design guidance and virtual maintenance process simulation.

  A Line Clip Algorithm of Based on Cell and R-Tree Spatial Indexes against Arbitrary Polygon Window

  LI Nan1,WU Xincai2,MA Jinjin3,WANG Zhong4

  2012, 34(11): 96-103. doi:

  Abstract ( 167 )   PDF (1057KB) ( 335 )   　　

  There are three main problems in line clip:reducing the number of segments intersection,reducing the time complexity for computation property of intersections and determining whether nonintersection lines is in clip polygons.This paper presents an improved line clip algorithm against general polygon window.The algorithm uses RTree and Cell structures to calculate segment intersections between segments in lines and segments in polygon edges.Then,based on uniform subdivision,a method named local radial is introduced in order to reduce the time complexity for pointinpolygon.The algorithm has three improvements.Firstly,segmentintersections of two spatial index structures can be calculated simultaneously.Secondly,a ‘local radial method’ is presented to reduce the time complexity for computation property of intersections,whose time complexity is,and the method avoids judging the orientation of clip polygons.Finally,the algorithm can be used in the clipping window that is general polygon.

  The Automatic Image Mosaic Based on Harris and SIFT Algorithms  

  LI Huanhuan，HUANG Shan，ZHANG Hongbin

  2012, 34(11): 104-108. doi:

  Abstract ( 201 )   PDF (537KB) ( 290 )   　　

  Image mosaic technology is widely used in many fields such as remote sensing image processing, computer recognition, medical image analysis and artificial intelligence, etc. Considering the shortcomings of scaleinvariant feature transform (SIFT) algorithms, i.e. the complexity and timeconsuming of the feature extraction actions, and the Harris algorithm can fast extract the features in the image, this paper presents an algorithm combining the advantages of the Harris and the SIFT and uses the algorithm in automatic image stitching. we firstly use the Harris algorithm to improve image feature extraction, secondly use the SIFT algorithm to describe the feature points, thirdly use the Euclidean distance to match the obtained feature vectors, and finally implement the automatic image stitching. Experimental results show that the method can effectively improve the efficiency of matching in the SIFT algorithm, and better do automatic image stitching.

  A Blind Watermarking Algorithm of Color Image Based on SVD

  GUO Qiaodan,WU Xisheng

  2012, 34(11): 109-113. doi:

  Abstract ( 175 )   PDF (1311KB) ( 313 )   　　

  In order to protect the copyrights of digital products, a color watermarking algorithm based on lifting wavelet transform (LWT) and singular value decomposition(SVD) is proposed. The LWT is quick and it is suitable for any size of image. The SVD has geometrical invariances and it shows the intrinsic characteristics of the image. This algorithm makes use of the characteristics of the LWT and the SVD, and chooses the blue channel of the color image to embed watermarking. The algorithm firstly uses multilevels LWT on the original image, secondly applies the SVD on low frequency subband and middle frequency subbands, thirdly uses Arnold to scramble the grayscale watermarking, and finally divides it into four parts that are embedded into the singular values matrices .It does not need the original image when the watermarking is extracted, so it can be better applied into the practice. The experiment shows that the algorithm has good transparency and robustness to common attacks and geometric attacks.

  The Micro Image Threshold Segmentation and Automatic Classification of Leafy Herbal Medicine

  ZHANG Cuiping1,YANG Shanchao2

  2012, 34(11): 114-119. doi:

  Abstract ( 227 )   PDF (2733KB) ( 284 )   　　

  According to the characteristics of the micro image of leafy herbal medicine, the paper proposes a threshold based segmentation method for the micro image of leafy herbal medicine, and measures the stomata index. The edge detection methods are used to elevate the threshold segmentation, and the morphology methods and region descriptor are employed to segment cells and count regions. Suppose that the objects in the image has normally distributed，we segment the stomata in the image where the cells have been removed. In this step the characteristic of the colorspace is used. The experiments show that the proposed threshold based segmentation method is effective for the Micro Image of Leafy Herbal Medicine.