Loading...
  • 中国计算机学会会刊
  • 中国科技核心期刊
  • 中文核心期刊

Current Issue

    • 论文
      ArtemisCOOR:A Platform for Agent Based Dynamic Software Coordination
      CAO Chun,MA Xiaoxing,TAO Xianping
      2010, 32(5): 1-5. doi: 10.3969/j.issn.1007130X.2010.
      Abstract ( 603 )   PDF (513KB) ( 147 )     

      In view of the challenges in the software coordination over the open network environment, an agentbased dynamic coordination model is proposed. Software agents are employed as encapsulations for the traditional software components and services, which enables the coordination logic to be reinterpreted dynamically and autonomously justintime. The concrete interactions between agents can also be multimoded to accommodate the possible heterogeneity among the coordination particapators. The corresponding supporting platform ArtemisCOOR is also introduced, which is featured in its "nonintrusive" intercepting technique, multimode interaction mechanism and the software architecturebased system envolution. We also give a demonstrative application of a hydrographic information system  to show the feasibility and effectivity of this work to meet the initial challenges.

      A Trust Mechanism Based on the Trust Vector in the P2P Environment
      WANG Haiyan,HU Ling,WANG Ruchuan
      2010, 32(5): 6-9. doi: 10.3969/j.issn.1007130X.2010.
      Abstract ( 577 )   PDF (412KB) ( 275 )     
      To address the problem of the cheating behaviors of those malicious peers in the P2P environment aiming to conduct attacks on a high trust value through a short period honest behavior, a trust vector based trust mechanism is proposed in this paper.The accuracy of trust evaluation is increased by the trust vector and the time window. Simulation results demonstrate that the new mechanism can effectively solve the cheating behaviors of those malicious peers compared with some available trust mechanisms.
      An Intrusion Detection Mechanism Based on Immune MultiAgents in WSN
      ZHANG Nan,ZHANG Jianhua,CHEN Jianying
      2010, 32(5): 10-14. doi: 10.3969/j.issn.1007130X.2010.
      Abstract ( 573 )   PDF (609KB) ( 347 )     
      As for the conventional intrusion detection technology in wireless sensor networks which can not adapt to the characteristics of dynamic property and limited resources, the artificial immunity theory and the multiagent technology are combined together, and a new intrusion detection mechanism(IMAIDM) for clustered wireless sensor networks is proposed. The description of an immune multiagent model, the definition of the multiagent function and the relevant algorithms are given. The experiments show that IMAIDM possesses a higher detection rate, better selfadaptability and lower energy consumption feature.
      A Distributed Intrusion Detection and Decision System Based on MobileAgents
      SUN Bocheng,QIU Yanjun,LIANG Shiqing
      2010, 32(5): 15-17. doi: 10.3969/j.issn.1007130X.2010.
      Abstract ( 642 )   PDF (369KB) ( 299 )     

      With the rapid development of network applications,network security has become a major problem. Thus we propose a distributed intrusion detection and decision system based on mobileagents,give the structure of the system, describe the composition, function communication mechanism of agents,and describe a fuzzy geneticsbased learning algorithm.

      A Mobile AgentBased IDS for Wireless Sensor Networks
      ZHANG Hongli,HUANG Shouming
      2010, 32(5): 18-20. doi: 10.3969/j.issn.1007130X.2010.
      Abstract ( 603 )   PDF (410KB) ( 341 )     

      According to the characteristics of wireless sensor networks, with the  intrusion detection technology combined with the mobile agent technology,we propose a mobile agentbased wireless sensor network intrusion detection scheme, with several agent blocks in cooperation and distribution, using a clusteringbased intrusion detection algorithm, in order to improve the security and reliability of wireless sensor networks, and lower the power consumption of intrusion detection.

      Architecture  and Simulation System for SpecialOperational Radio Communication Networks
      GAO Chunrong,BEN Kerong
      2010, 32(5): 21-25. doi: 10.3969/j.issn.1007130X.2010.
      Abstract ( 566 )   PDF (499KB) ( 311 )     

      A mobile agentbased network management architecture for special communication networks that allow a special mission group to reconfigure and rebuild their networks is provided in this paper. With the architecture, a common simulation system is built which can be the integrated platform for the joined operations to configure single devices and the hierarchy of networks, accept highlevel direction from the commanders, cooperate to decide and enforce policies for the control of network configuration, services and traffic and the changing needs of the end users to support the special mission group for their tasks. The architecture can improve the effectiveness and survivability of endtoend network services in a dynamic, hostile, complex and unpredictable military environment, and  provide evidences for using the new and advanced agent technology in specialforce combat fields and other tactical situations.

      A System Model for the United Risk Assessment of Network Security Based on Mobile Agents
      LU Linlin,MA Xin
      2010, 32(5): 26-29. doi: 10.3969/j.issn.1007130X.2010.
      Abstract ( 572 )   PDF (442KB) ( 358 )     

      A system model for the united risk assessment of network security is proposed based on mobile agents(MAURA).This paper analyses the system structure and the function of each part,applies the cooperation method of the contract net protocol to risk assessment, and puts forth a mechanism of united risk assessment. The process of united risk assessment is also studied, which overcomes the problem which is too difficult to share the analytic experience. Meanwhile, MAURA adopts a new algorithm switch policy to aim at heavy loads of detection tasks to increase the system’s adaptive ability. The experimental results show that MAURA is a more adaptive and efficient system.

      A Novel Algorithm to Optimize the Hidden Layer of Neural Networks
      GAO Pengyi 1,CHEN Chuanbo1,QIN Sheng2,HU Yingsong1
      2010, 32(5): 30-33. doi: 10.3969/j.issn.1007130X.2010.
      Abstract ( 688 )   PDF (432KB) ( 356 )     

      This paper proposes a novel algorithm to Optimize the number of Hidden nodes based on Agent(OHA). This approach is completed by two cooperating agents, the RL agent and the NN agent. The RL agent searches better number of hidden nodes according to the reinforcement learning method, and the NN agent optimizes the weights of network with the number by using the separate learning algorithm. After much running, the best solution(weights and hidden nodes) is located. The optimization algorithms and tests are discussed. The test results obtained by using the Iris data set and the risk evaluation data set show the algorithm is better than those by the most commonly used optimization techniques.

      The Unsupervised and Uncorrelated Optimal Discriminant Plane Based on Orthogonal Constraints
      CAO Suqun1,2,WANG Jun1,3 ,WANG Shitong1
      2010, 32(5): 34-36. doi: 10.3969/j.issn.1007130X.2010.
      Abstract ( 642 )   PDF (377KB) ( 303 )     

      An improved optimal discriminant plane(IODP) proposed by Zhao can only be used in the supervised pattern. Based on this point, a novel method is presented to extend IODP to the unsupervised pattern. On the basis of optimizing the fuzzy Fisher criterion to calculate the first optimal discriminant vector, the second optimal discriminant vector with the orthogonal constraint and the conjugated orthogonal constraint of the fuzzy totalclass scatter matrix can be figured out. These two vectors constitute the orthogonalconstraintbased unsupervised and uncorrelated optimal discriminant plane(OUUODP). With these, a novel unsupervised feature extraction method is obtained. The experimental results for the CMUPIE face database demonstrate that this method can extract the features which are conducive to classification and is superior to principal component analysis and independent component analysis when the betweenclass difference is big.

      A Homogeneous Team Learning Model Based on XCS
      TAO Narisu,WANG Chongjun,ZHANG Lei,XIE Junyuan
      2010, 32(5): 37-40. doi: 10.3969/j.issn.1007130X.2010.
      Abstract ( 506 )   PDF (424KB) ( 277 )     

      Homogeneous team leaning is a method to deal with multiagent coordination. But traditional methods adjust the target agent only before or after the system simulation and the computation for the system simulation does not contribute directly to the improvement of the behavior of the target agent. A homogeneous team learning model based on XCS and the idea of cooperative strategy are purposed in this paper. This model overcomes the deficiency of the traditional method mentioned above. We also analyze the influence of some factors, such as rule accumulation, communication and the ability to explore by GA, on the efficiency of multiagent coordination on the proposed model.

      Research  on Some Critical Issues of the Nonlinear Support Vector Machine
      ZHU Shuguang1,QIAN Liyan2, FAN Weibing1,HU Xiaofeng3,WANG Jian1,LIU Qiang1,L
      2010, 32(5): 41-44. doi: 10.3969/j.issn.1007130X.2010.
      Abstract ( 776 )   PDF (383KB) ( 421 )     

      Firstly, some critical issues about the nonlinear binaryclass support vector machine (SVM) are studied. Secondly, the function of the nonlinear mapping used in nonlinear classification, the connotation of the dimension disaster, and the essence of the kernel function are elaborated. Thirdly, a new method to obtain connotative nonlinear mapping of the kernel function is put forward and the expression for the binaryclass SVM is presented. And finally, dataclassification experiments with the binaryclass SVMs are performed.

      Optimized Learning Based on Information Granularity and Connectivity
      WANG Xiuzhen1,2,ZHONG Ning1,3,LIU Chunnian4,GU Weiquan2
      2010, 32(5): 45-47. doi: 10.3969/j.issn.1007130X.2010.
      Abstract ( 631 )   PDF (354KB) ( 252 )     

      As for the problem of cognitive learning on distributed networks and complicated topological structures, the paper proposes a dynamic,information granularitybased and connectivitybased cognitive optimized  learning. The information granules of every network node hold the integrity of information expression under the condition of high degree of polymerization. Selfassembly polymerization of the nodes in the knowledge system and the strong connectivity between every two nodes are the kernel  model in the optimized learning result. The concept of polymerization degree of information granules and the connectivity among information granules, the static reduction of the evolution of the information granules which imitate the  cognitive learning process,and the dynamic imitation of connectivity intensity evolution which correspondes cognitive learning are used here. And the two processes accomplish a whole imitating cognition and reduction expression to every inputting sample in the learning system. This thesis aims to take the distributed topological structure as a theoretical model to propose the cognitive optimization rules for the information granularity of every node as well as the information procession and transmission among the nodes.

      Neural Network Training Based on the Extended Kalman Particle Filter
      WANG Fasheng,GUO Quan
      2010, 32(5): 48-50. doi: 10.3969/j.issn.1007130X.2010.
      Abstract ( 508 )   PDF (374KB) ( 388 )     

      Training neural networks can be viewed as an identification problem for a nonlinear dynamic system. The generic particle filter has been applied with success to neural network training, but the proposal distribution chosen by the generic particle filter does not incorporate the latest observations which can deteriorate the performance of the algorithm. In this paper, we propose to use the extended Kalman filter to generate proposal distribution in the particle filtering framework. The extended Kalman filter can make efficient use of the latest observations, and the generated proposal distribution can approximate the posterior distribution of neural network weights much better, which consequently improve the performance of the particle filter. The experimental results show that the proposed particle filter outperforms the generic particle filter.

      A Method for the Automatic Test Data Generation Based on  the GeneticAnt Colony Hybrid Algorithm
      LI Kewen,ZHANG Zilu
      2010, 32(5): 51-53. doi: 10.3969/j.issn.1007130X.2010.
      Abstract ( 553 )   PDF (416KB) ( 444 )     

      Automatic test can improve the efficiency of software test and reduce the cost of development. Automatic generation of the test data is a very important part in the automated test process. Based on the path coverage criterion, this paper uses the GeneticAnt Colony Mixed Algorithm to search the input domain in order to generate the test data which meet the coverage criterion. The experimental results show the mixed algorithm generates test data effectively.

      A Hybrid Genetic Algorithm Based on DPLL for Solving the SAT Problem
      WANG Xiaofeng,XU Daoyun,TANG Ruixue
      2010, 32(5): 54-56. doi: 10.3969/j.issn.1007130X.2010.
      Abstract ( 589 )   PDF (353KB) ( 289 )     

      When the Optimized Genetic Algorithm based on clustering and ranking selection is  used for the SAT problem, a crossover factor and a mutation factor are imported. Moreover, according to the fitness function and the characteristics of the related issue, threshold δ is adjusted, in order to produce a new population clustering. The algorithm has the effective suppression of delayed convergence, so the satisfiability of formula can be assigned rapidly. Meanwhile,we introduce the DPLL algorithm into the genetic algorithm. It performs the resolution to the variables, and raises the algorithm’s solution efficiency. According to the related experimental data, the performance of the algorithm is obviously better than other algorithms alike, and it has a high reliability.

      A Study on the Convergence of the Clonal Selection Algorithm Based on the Interval Sheath Theorem
      LIU Zhandong1,FU Tao2,DAI Yugang1,ZHAO Qinghua1
      2010, 32(5): 57-59. doi: 10.3969/j.issn.1007130X.2010.
      Abstract ( 467 )   PDF (350KB) ( 306 )     

      The clonal selection algorithm arises from an immune optimization algorithm,which is proposed based on the clonal selection theory of immunology.It does this by cloning an operator to operate.Firstly,this article introduces the standard clonal selection algorithm.Secondly,the cloning operator is introducted to improve the standard clonal selection algorithm.Then,this paper analyses the convergence of the clonal selection algorithm,which is based on the knowledge of sequence and that the antibody groups clonal selection’s process is as an object.Finally,the  interval sheath theorem is used to prove the global convergence of the algorithm.

      A Temporal Extended Value Based Argumentation Framework
      ZHAI Haoliang,LI Lei,ZHAO Gansen
      2010, 32(5): 60-63. doi: 10.3969/j.issn.1007130X.2010.
      Abstract ( 444 )   PDF (344KB) ( 243 )     

      Time is an important element of describing the process of argument and the changes of argument activities, and it is active to introduce time into argumentation framework. This paper proposes a temporal extended value based argumentation framework (TEVAF) based on Dung’s standard argumentation framework (AF) and BenchCapon’s value based argumentation framework (VAF). First, we analyze their limitations in describing temporal and values which are related to arguments. Then,according to the requirements, TEVAF is proposed.We present a complete description of the structure of framework,and the  semantics of the framework,and give the fundamental Lemma and some basic results shown by Dung’ argumentation framework,which also hold for TEVAF.

      Optimal Coalition Structure Solving Based on the Bipartite of Integer
      LIU Jinglei,ZHANG Zhenrong,ZHANG Wei
      2010, 32(5): 64-66. doi: 10.3969/j.issn.1007130X.2010.
      Abstract ( 509 )   PDF (366KB) ( 245 )     

      Coalition structure is a partition of the agent set, forming a coalition structure by coalition can make agents cooperate effectively and fulfill the tasks that a single agent can not. In this paper, we propose a BIDP (Bipartite of Integer Dynamic Programming) algorithm to solve the optimal coalition structure generation, which adopts the bipartite of integer to generate bipartite partitions, and takes the bound of integer bipartite as the bound of the search space. And a theoretical analysis proves that BIDP can save 33.3% memory in any case than DP (Dynamic Programming). An experiment analysis shows that BIDP can save 35% and 92% searching numbers on 21 agents when the coalition values meet the uniform distribution and normal distribution. Finally,the paper gives a verdict that the time complexity of the determinant algorithm to solve OCS is between Ω(2n) and O(3n), and the space complexity is Θ(2n).

      An AgentOriented Service Rules Validation Method
      LI Tong,LIU Lin
      2010, 32(5): 67-73. doi: 10.3969/j.issn.1007130X.2010.
      Abstract ( 604 )   PDF (772KB) ( 277 )     

      This paper proposes an approach to validating and simulating service rules based on an agentoriented Service Requirements Modeling Ontology (SRMO). SRMO models service requesters and providers as intentional actors, who carry out rulebased reasoning independently to realize automated service discovery, selection and matchmaking in the service environments. In order to validate the rationality and correctness of the rules, this paper presents an agentoriented service rules validation method, and implements an agentoriented rule verification platform (AORVP). As a result, the current rules in SRMO are validated with this tool, and are revised and improved based on the simulation reasoning results.

      An Algorithm for Coalition Formation with Complex Tasks
      CHEN Yuwu1,CAO Jian1,LI Minglu1,ZHAO Haiyan2
      2010, 32(5): 74-78. doi: 10.3969/j.issn.1007130X.2010.
      Abstract ( 455 )   PDF (435KB) ( 249 )     

      The coalition formation process, in which a collection of agents decide how they might group together to satisfy the tasks, is one of the important aspects in multiagent systems. Although this problem has been investigated for many years, few attentions are paid to the situation where tasks have complex interdependent relationships. To address this omission, we consider the situation where there is a logical interdependent relationship between tasks,and the transfer cost associated with tasks and agents is incurred between interdependent tasks. Based on this background, we present an efficient algorithm based on the graph theory to find the optimal coalition. Considering the tasks with normalized logical interdependent relationships, we transform the problem of finding optimal coalition to the shortest path problem in the graph. We also give the time complexity analysis of the algorithm,and the experiment results show that the algorithm has good performance.