Current Issue

• 2012, No. 3　Published:26 March 2012 Last issue    Next issue
  Columns in this issue: 论文

• 论文

  A Rough Set Analysis Model of Network Situation Assessment

  ZHUO Ying,HE Ming,GONG Zhenghu

  2012, 34(3): 1-5. doi:

  Abstract ( 292 )   PDF (771KB) ( 343 )   　　

  The rapid development of the Internet leads to an increase in system complexity. Traditional network management can not meet the requirements, and it should evolve to fusionbased Cyberspace Situational Awareness (CSA).As the core of CSA, Network Situation Assessment (NSA) can integrate unit network managements, give a comprehensive macroscopic view of network operation status, and provide decision support. This paper focuses on the characteristics and requirements of network situation and introduces a rough set theory which has the superiority in machine learning, dealing with massive redundancy information, feature selection, and so on. We propose a NSA model based on rough set analysis, gave a formal definition and discuss the assessment procedure in detail, which includes decision table establishment, data preprocessing, situation factor selection, consistency judgment, condition attribute reduction and decision rule reduction. The experiment results demonstrate the efficiency and effectiveness of the prototype system.

  A MobilityBased Cluster Routing Protocol for Mobile Wireless Sensor Networks

  WANG Can，LUO Jian，ZHANG Dafang，LI Wenwei

  2012, 34(3): 6-12. doi:

  Abstract ( 395 )   PDF (771KB) ( 247 )   　　

  Mobility in wireless sensor networks has attracted a lot of attention in the recent years. Under the mobile application scenes, network topology changes frequently. A Mobilitybased Cluster Routing Protocol for mobile wireless sensor networks that can meet the mobility requirements is proposed. It includes a Clusterhead Election Algorithm and an Adaptive LEACHMobile Algorithm. It gives full consideration to node mobility and the remaining energy. The experiments show that MCR can effectively improve network performance in mobile environments.

  A ConnectivityBased Planarization Algorithm for Wireless Sensor Networks

  DONG Dezun1，LIAO Xiangke1，SHEN Changxiang2

  2012, 34(3): 13-18. doi:

  Abstract ( 321 )   PDF (956KB) ( 460 )   　　

  Extracting planar graph from network topologies is of great importance for efficient protocol design in wireless sensor networks. It has become the foundation of many important protocols and applications, such as geographic routing including GPSR, GOAFR, etc. Previous techniques of planar topology extraction are often based on the assumption of accurate node location measurements. It is highly desirable to relax assumptions and extract a planar topology in a locationfree and distributed manner. We present a novel method to construct a planar structure for wireless sensor networks in a distributed manner, merely using connectivity information. Our method can extract a provably planar graph from communication graphs. The extensive simulations also show the effectiveness of this design in randomly generated networks.

  Research on Data Failure Tolerance and Energy Efficiency in MultiEvent Clusters for Wireless Sensor Networks

  XIAO Wei1,2,XU Ming1

  2012, 34(3): 19-23. doi:

  Abstract ( 277 )   PDF (793KB) ( 306 )   　　

  A multievent cluster data failure tolerance schema called MEDFT is proposed for wireless sensor networks, based on the tradeoff achieved by failure tolerance and energy efficiency in event clusters. For this schema, the paper firstly develops a distributed clusterheader election method by calculating the product of both node residual energy and node confident rate. Secondly, it puts forward a node disposal policy for the nodes deployed in the multievent wrapped area, and finally, it establishes a compensation mechanism of multievent cluster data failure tolerance. The simulation results demonstrate that the multievent cluster with MEDFT can significantly increase the network survival period, and obtain event values with less errors and greater data failure tolerance, compared with other algorithms.

  Analysis and Performance Evaluation of the IEEE 802.15.4 MAC Protocol

  ZHANG Linlin1,LI Dongsheng2,YONG Aixia3

  2012, 34(3): 24-28. doi:

  Abstract ( 341 )   PDF (1150KB) ( 316 )   　　

  For the problem of how to improve the network transmission performance in wireless sensor networks, the paper proposes a scheme of using different transmission modes for different network topology structures. Based on the analysis of two different network structures: the “tree” structure and the “network” structure by the NS2 network simulation software, the delivery rate(packet loss rate), the average endtoend delay and the routing overhead are analysed for the two different network structures, and for the beaconless and beaconenabled operations. The experimental results show that the different network topology structures targeted by different transmission ways really can increase the transmission performance.

  A Survey of the Programming Methods for Heterogeneous Systems

  TANG Tao,YANG Xuejun

  2012, 34(3): 29-34. doi:

  Abstract ( 306 )   PDF (399KB) ( 382 )   　　

  Heterogeneous parallel systems integrated with general purpose CPUs and domainspecific accelerators feature higher performance and energyefficiency than their homogeneous counterparts. Nowadays, heterogeneous parallel systems have become an important trend in the high performance computing field. However, the programming model problem becomes one of the most important challenges of heterogeneous systems, and thus attracts much attention from academia and industry. In this paper, we first discuss the source of the programming issue for heterogeneous systems. Then we introduce and classify the researches of the programming models for heterogeneous systems in recent years. Finally, the programming issues to be further studied for heterogeneous systems are discussed.

  Design of the Radiation Hardened DFF Based on Muller_C and DICE

  LI Peng,SUN Yongjie,CHEN Jianjun,LIANG Bin

  2012, 34(3): 35-40. doi:

  Abstract ( 358 )   PDF (1437KB) ( 339 )   　　

  Currently, smaller feature sizes and higher frequencies of deep submicron integrated circuits make ICs susceptible to singleevent upset (SEU) and singleevent transient(SET). In this paper, two kinds of timing sampling latches based on the static circuit and dynamic circuits of Muller_C are designed. Meanwhile, combining with the DICE latch, two D flipflops which can tolerate both of SEU and SET are also designed. Based on the 3D TCAD Mixed mode simulation SET is created and injected into the circuit netlist which obtain the RC parasitic parameters from the layout. The Hspice simulation results show that, these two DFFs can tolerate SEU and SET efficiently. Compared with the SEU and SET tolerant DFFs in ［1］, these two presented DFFs need less area, dynamic power, and static power at the frequency of 500 MHz, and one of the DFFs has a shorter setup time.

   The Embedded Virtualization Technology

  YE Changchun

  2012, 34(3): 41-46. doi:

  Abstract ( 358 )   PDF (461KB) ( 450 )   　　

  The virtualization technology of computer systems is a hot subject of the IT area in recent years. The next step of virtualization technology is embedded systems. What the virtualization of embedded systems does is to insert a software layer called virtual machine monitor between the hardware platform and the operating system, and the virtual machine monitor constructs a few virtual machines, which run several types of operating systems. Many foreign corporations and universities are carrying out researches on the embedded virtualization technology. The advantages of this technology are: to reduce the development cost of embedded systems, to push embedded products faster, to considerate more features, to lower power consumption, to maintain the values of software, and to improve security and reliability. The problems of the embedded virtualization technology include realtime scheduling, diversity of embedded hardware platforms, power management and communication among virtual machines. Embedded virtualization technology will bring the area of embedded systems with huge changes, and deserve our close attention to it.

  Application of Sliding Mode Control in Disk Drive System Based on the Kalman Filter

  LI Honglin1，ZHAO Pengbing2

  2012, 34(3): 51-54. doi:

  Abstract ( 299 )   PDF (723KB) ( 267 )   　　

  According to the random disturbance and noise in disk drive servo system and the traditional PID control has the disadvantages of lag or overshoot, and the traditional sliding mode control has the faults that ask for system uncertainty factors to be known. Then, this paper proposes the exponential approach law sliding mode control method based on the Kalman filter. The method of designing a sliding mode controller based on the reaching condition of the exponential approach law can handle the uncertainty problems in disk drive servo system better, and it is not sensitive to the external disturbances. The simulation experiment of this algorithm is processed based on the MATLAB environment. The simulation results show that this algorithm can weaken the vibration and suppress the random disturbance. This method can be used in the disk drive servo system to significantly improve the dynamic performance as well as it can enhance the stability and control accuracy of the system.

  An Interval Consistency Control Method Based on Estimating the Network Delay in the DVE Systems

  ZHANG Wei1,ZHOU Hangjun1,PENG Yuxing1,TANG Xinchun2

  2012, 34(3): 55-61. doi:

  Abstract ( 390 )   PDF (654KB) ( 338 )   　　

  The interval consistency problem of continuous models is the key issue to determine system functionality and effectiveness in the DVE systems. The existing methods often utilize the strategy that the sender nodes are responsible for choosing the event delay, and impose a restriction relationship on the all receiver nodes, which hampers the process to optimize the system responsiveness. In this paper, we propose a novel interval consistency control method based on estimating the network delay. The method utilizes the strategy that the receiver nodes are responsible for choosing the event delay to relieve the restriction relationship among all the receiver nodes. Moreover, the time stamp consistency and interval consistency are kept simultaneously due to the same delay strategy for all local events and remote events. For adjusting the parameters properly during the run time, we propose a Btree based the network delay estimation method, which adjusts the delay parameter periodically and ensure the effectiveness of our method. The experimental results show that our method can provide interval consistency for all the nodes and improve the overall system performance effectively.

  Research on the SLIQ Parallel Algorithm Based on Cloud Computing

  YANG Changchun,SHEN Xiaoling

  2012, 34(3): 62-66. doi:

  Abstract ( 287 )   PDF (437KB) ( 333 )   　　

  Cloud computing provides efficient solutions to storing and analyzing mass data.It is very important to study the data mining algorithms based on cloud computing from the theoretical viewpoint and the practical viewpoint.The SLIQ algorithm finds the best split point through calculating the scalability indexes one by one.When the amount of data increases,the method is timeconsuming,and the efficiency of the algorithm is very low.In this paper,the algorithms of mining decision rules based on the cloud computing environment are focused on the MapReduce programming model.On the basis,an improved SLIQ algorithm as well as the procedure of the improved SLIQ algorithm on MapReduce is designed in order to realize parallel data mining.

  Evaluation of the Trace Effects in LargeScale Parallel Performance Simulation and Discussion of Some Resolutions

  XU Chuanfu1,WANG Rong2,CHE Yonggang1,WANG Zhenghua1

  2012, 34(3): 67-73. doi:

  Abstract ( 301 )   PDF (1330KB) ( 303 )   　　

  Trace generation is an inevitable step in tracedriven architecture simulation. Traces not only consume largescale storage space, but also may introduce extra intrusions to the execution of benchmarks which can result in errors of the simulation results. The trace effect of parallel tracedriven simulators has its own unique characteristics due to the design and implementation of the simulators as well as particular I/O systems of parallel hosts. We select BigSim, a typical parallel simulator, and several target parallel applications with different computation and communication ratios to evaluate the trace effect on 3 parallel host systems with different I/O modes. Our results show that trace generation has a nonignorable effect on both efficiency and accuracy of parallel simulation. The reasons of the trace effect and some possible resolutions are also discussed. The conclusion of our evaluation is helpful to the design, implementation and use of tracedriven parallel architecture simulators.

  Research and Design of a Disaster Data Processing Scheme Based on Cloud Computing

  HE Guangjun1,GUAN Qun1,WEI Fangqiang2,HU Kaiheng2

  2012, 34(3): 74-79. doi:

  Abstract ( 339 )   PDF (868KB) ( 370 )   　　

  In order to achieve high efficiency of disaster data processing and computer resources integration in research institutions, a solution is proposed based on cloud computing. The resource virtualization based on the resource pool, HealthcheckLoadbalance coordination mechanisms and threeway connection authentication to process control functions are proposed, the debris flow simulation logic deployed on the platform successfully, the cloud computing technology features better engineering practice, and the spare computer resources in the research institutions have been effectively used. The results show that the Cloud Computing technology embodies many advantages in processing data.

  A Common Distributed Middleware for Causal Order Time Management in DVE

  ZHOU Hangjun1,ZHANG Honglei2

  2012, 34(3): 80-85. doi:

  Abstract ( 307 )   PDF (626KB) ( 283 )   　　

  The performance of time management services is directly related to the effectiveness and correctness of DVE established according to the HLA standards. Generally, the realized time management services include two message ordering mechanisms: the receiving order and the time stamping order. The receiving order has a good realtime property but cannot correctly deliver messages that the receiving order is violated to their generating order. The time stamping order can maintain the message delivery in the generating order, while it is hard to meet the realtime property of DVE for its significant overhead. In this article, we propose a common distributed middleware to offer causal order time management services for DVE, which provides standard program interfaces of HLA. The experimental results demonstrate that by utilizing this middleware the delivery order of messages can be preserved correctly and effectively, in the meantime, the small overhead is suitable for the requirement of the realtime property of DVE.

  A Range Particle Algorithm and the Solving of the Inverse Problem of the Line Source

  ZHOU Hangxia,LIU Qian,ZHENG Peng

  2012, 34(3): 86-90. doi:

  Abstract ( 272 )   PDF (739KB) ( 321 )   　　

  Currently the numerical solution of the inverse problem of the line source does not have a mature and effective algorithm. Based on the study of the range search algorithm, a new algorithm is discussed for the numerically solving the inverse problem of the line sourcerange particle algorithm (Range Particle algorithm). First a brief introduction to the characteristics of solving the inverse problem of the line source is given, and a mathematical model of the inverse problem solution is established based on the line source equation. Secondly, based on the mathematical model, a range particle algorithm is designed to solve the problem, and the basic implementation steps and parameter adjustment of the algorithm are also discussed. Finally, the simulated and measured data are used to test the effect of the algorithm. The results show that the range particle algorithm has high precision, fast convergence and computational stability to the inverse problem of the line source, and it is applicable.

  MultiClass Object Recognition in Natural Scenes

  WU Shilin1,2,3,4，ZHU Feng1,3,4

  2012, 34(3): 91-95. doi:

  Abstract ( 295 )   PDF (659KB) ( 345 )   　　

  In this paper, a conditional model (CM) is used to incorporate different feature potentials including texture, texture environment and location features of objects for multiclass object recognition and segmentation in complex natural images. Besides, we model the relationship between different objects by the scene of images and propose a new scenebased conditional model called the sCM model. We investigate the performance of our model in the classbased pixelwise segmentation of images on the Oliva & Torralba database and compare its result with other methods. The results show that our themebased RCRF model significantly improves the accuracy of objects in the whole database. More significantly, a large perceptual improvement is gained, i.e. the details of different objects are correctly labeled.

  The Realization Technology Research of the AMFB Model Based on OpenCASCADE

  CHEN Hongyu,LIU Zijian,AI Yandi

  2012, 34(3): 96-102. doi:

  Abstract ( 293 )   PDF (1319KB) ( 320 )   　　

  According to the AMFB’s product information modeling principle, the architecture design and application framework design of the OpenAMFB system are finished by applying a geometrical modeling platform of OpenCASCADE. And, on this basis, the AMFD model application system(OpenAMFB) is developed by using the OCAF development framework as the software architecture, the feasibility of the realization technology is verified, and reference is provided for the development of the modelcentric product’s information integration system. An OpenAMFB system shares data and information with the CAE, CAM, CAPP systems, and has important theoretic reference values and pratical guidance significance of information, machinery and other engineering technology research,and plays a certain role for the integration of CAD/CAE/CAPP.

  Research and Implementation of  FPGA-Based High-Speed 2D-DCT

  LIU Qing1,2，CHEN Jinqiang1，YU Peiling1

  2012, 34(3): 103-107. doi:

  Abstract ( 371 )   PDF (638KB) ( 352 )   　　

  This paper presents a highperformance 2D discrete cosine transform (DCT) processor based on distributed algorithms(DA),which uses the field programmable gate array (FPGA) to implement the JPEG encoder. This processor has none multiplier. An improved DA structure is designed and the pipelining style is applied to the data flow. The FPGA implementation and simulation results show that this design can achieve a higher operation speed than the traditional fast DCT algorithm.

  Realization of a Web Camera Supporting the  B/S Pattern

  LI Gang,MI Yuangen

  2012, 34(3): 108-112. doi:

  Abstract ( 322 )   PDF (607KB) ( 304 )   　　

  This paper introduces the realization of a web camera supporting the  B/S(Browser/Server) pattern based on S3C2440 CPU and the mjpgstreamer software. The basic hardware structure consists of the S3C2440 which is the main processor of the network camera, the camera with USB interfaces and the DM9000 net chip, which is used to take and transmit pictures. The uboot and Linux are used as the bootloader and the embedded operation system respectively in the software system. The mjpgstreamer is the software to manage picture stream, which focuses more on the method of compiling, function and porting. In the end, the function by which the  realtime video pictures taken by distant network cameras can be watched through a browser is realized.

  An Improved BM Pattern Matching Algorithm Based on Double Character Sequence Checking

  WANG Hao,ZHANG Lin,ZHANG Qing

  2012, 34(3): 113-117. doi:

  Abstract ( 317 )   PDF (644KB) ( 327 )   　　

  The BM algorithm is a  more efficient single pattern matching algorithm. The common method of improving the BM algorithm is to start with increasing the probability of the first failure character matching and the maximal moving distance of the matching window. At the same time, the higher cost of accessing memory counteracts the actual efficiency of the new algorithm. Optimizing the number of the lookup table and the times of accessing memory when it moves the key distance of the matching window, and DCSBM makes good use of the double character’s first failure matching probability at the cost of reducing the key steps properly. It is tested that DCSBM obviously increases the average moving distance of the matching window. In larger length of the text or pattern, the real efficiency of DCSBM is higher than BM, BMHS, or BMN algorithm.