最新刊期

Risk of requirement is ignored in Tropos model.To analysis risks and relationships of requirements,a method analyzing risks in requirement phase is introduced.It's on the base of Tropos model,and proposes a Tropos goal gram-based framework of requirement relationship to identify risk factors of requirements:stakeholder,environment,technology and complexity of requirement and assess risk about feasibility of requirements.Because of subjectivity,these risk factors are assessed by relative experts.It is proved by nexample of train subsystem that the method can identify stakeholders and requirement with risk in time.This makes it earlier to solve risks and saves cost of project.

As ubiquitous computing pattern has been infiltrating people’s daily life nowadays,the context-aware application’s development process is still adopting thetraditional object-oriented programming model and technique,which makes the contextrelated behavior design distributed in the application programs,and the binding of environment context and behavior occurred at coding phase.This will lead to the excessive tightly combination of the environment dependence factor and programming control logic,also hard for the system expanding and maintenance.According to the situation,this thesis aiming at the dynamic adaptation of environment context changing,designed and implemented a context-aware programming model called EIPM which implemented dynamic proxy as the base mechanism,separated thecontext information and programming logic by establishing the mapping rule.Also,it provided the corresponding development platform and execution container as the implementation framework prototype system to support work under the programming model.Use Pervasive Document Access System as application background,test and illustrate the feasibility of EIPM application framework,and the results indicate that EIPM programming model has the dynamic adaptability to the changing environment context.

A planning exploitation graph-bayesian networks model that can be applied in measurement of information security risk frequency is proposed,and the model’s scalability,accuracy and objectivity are achieved.The model graph structure is determined by Planning Exploitation Graph,the local conditional probability distributions are computed by combination ofexpertise knowledge and the maximum entropy prior probability distribution method,and the model parameters are updated with training data by Bayesian networks learning.The analysis of the example shows the model could evaluate the information security risk frequency successfully.

Thermal infrared faces have the characteristics of edge and detail faintness,low contrast,different temperature distribution between faces and their backgrounds.Focusing on these characteristics,a new image segmentation method is proposed.Gray projection is used to coarsely locate face region.Fuzzy connectedness is used to segment background and determine the positions of eyebrows.According to the centers of eyebrows,faces are located exactly and normalized.The experimental results show that the method eliminates the influence of background,reserves more facial information and can effectively locate and segment thermal infrared face images.

The existence and being able to find a large number of reusable components is one of the prerequisites of componentbased software reuse.Software component library is the infrastructure of component management.It can help developers find appropriate components.Component tag based taxonomy is a new method to manage components in component library.It uses component tag to manage component and supports users using tags to find components.This method is efficient in component management because component tag represents the characteristics ofcomponents.However,many components don’t have component tags in component library.Adding tags for these components manually needs a lot of time and workload.It’s almost impossible when there are a lot of such components.So,this paper proposed an automatic component tag extracting method based on classification.This method extractscomponent tags from component descriptions automatically.This paper described atool based on this method and verified the effectiveness of the tool.

Road detection is an important task of remote sensing images processing.With the feature of combining picture and spectrum,hyperspectral images present new useful information resource for road detection.With the requirement of remote sensing road detection,a new method for road detection and extraction based on hyperspectral images is proposed.Firstly,independent component analysis and linear mixture model are used to unsupervised unmix the hyperspectral images,and the unmixed independent component including road information is obtained.Then,the road detection and extraction is achieved by means of ratio operator and Hough transform.The results from simulation experiments demonstrate the validity of the proposed method.

As it is restricted by the filter’s parameters,such as the passband bandwidth,inband ripple,outband attenuation,insertion loss and its volume and so on,designing and chieving broadband IF anti-aliasing filter is difficult.The IF bandpass sampling and antialiasing filter,and the relationship between the key parameters of the LC(inductors and capacitors) and the working frequency,and LC components selection principles are introduced.Then it achieved the designing of the filters,the actual components parameters configuration and characteristic simulation with the computer simulation,and provided the methods to design and realize the higher-order elliptic filters.Lastly,gave the designed circuit and testing results of the 9-order elliptic filter.

Image segmentation is a classic problem in computer vision,and become a hot topic in the field of image understanding.The research actuality and new progress in image segmentation in recent years are summarized in this paper.Firstly,the traditional methods of image segmentation are introduced summarily.Then,the specific theory for image segmentation,including morphology,fuzzy sets and neural network,support vector machine,immune algorithm,graph theory and granular computing,are presented emphatically.Furthermore,several new representative papers with the application of each theory are analyzed and discussed.Finally,the development trend of image segmentation method is discussed.

This paper focuses on the recurrence component of imitation mechanism and presents humanoid reasoning strategies such as knowledge use and action preview to implement effective behavior imitation.Concretely,by defining the Hausdorff distance between fuzzy sets as the metric of knowledge use,we combined knowledge radius into the distancetype fuzzy reasoning method for the implementation of selective knowledge use;according to the main idea of preview control,we designed the action preview model as high level decision of knowledge use to optimize the parameters of knowledge use strategy.Finally we presented the results of driving behavior imitation,demonstrated the validity of humanoid reasoning strategies indynamic knowledge use,and realized quick learning and imitating performance.

This paper presents an intelligent methodology for diagnosing incipient faults in mine hoist.In this fault diagnosis system,in order to enhance the immune algorithms performance,we propose the improved immunebased symbiotic a new evolutionary learning algorithm.This new evolutionary learning algorithm is based on Discrete Particle Swarm Optimization (DPSO) technique to improve the mutation mechanism.Also to solve the problem that exists in fault diagnosis based on the traditional method using distance discriminant function,an improved method based on immunity strategy with similarity measurement of principle component kernel is presented.The effectiveness of the DPSO based immune algorithms is demonstrated through the classification of the fault signals in mine hoist.Simulation results show that the new scheduling algorithm can deal with the uncertainty situation and be suitable for multifaults diagnosis,compared to the traditional scheduling algorithms.

RFID Discovery Service provides a means to find all dynamic RFID events related to certain objects as the data set,from different enterprises’ information services.To facilitate visual analysis on the discovered data set,a modeling technology for supply chain is proposed.
.The discovered data set of supply chains is specified,which includes the movement of!objects,aggregating with other objects or disaggregating from an integrity,and processing or transforming materials into new products.As well,an innovative distributed RFID Discovery Service is briefly introduced by combining technologies of P2P and parallel processing.Then,a Petri net based modeling tool “SupplyNet” is proposed,and the algorithm for constructing SupplyNet is given.Finally,analysis and experiments on our algorithm show its high efficiency and utility.

For the sake of reducing communication conflicts and channel interference furthest,this paper studies the upperbounds of channels and radios of conflictfree communication in MultiRadio MultiChannel sensor networks.We prove theoretically that when the networks communication radius NCR>3×Dis(P_K)and the Sensor Size of the networks is greater than 2K+1,the upperbound of channels of conflictfree communication is △(CG),in whichK is the power levels of the networks,Dis(P_K)is the communication radius of the largest power P_K of the networks and △(CG)is the largest degree of the conflict graph of the neworks.Based on the above upper bound,a channel assignment algorithm is proposed.Theoretical analysis and experimental results indicate that the channel assignment algrithm this paper proposed can improve the communication efficiency of sensor networks significantly,and then increase the networks throughput.

With limited resources,wireless sensor networks usually use clustering aggregation to decrease traffic.This paper proposes a dynamic clustering algorithm based on aggregation gains.Firstly,we present a nonlinear integer programming model for the overall energy consumption of the network optimization problem,and then propose a low complex and near optimal heuristic cluster head election algorithm.A dynamic clustering algorithm is proposed based on aggregation gains,which can elect the cluster head in a distributed way.Theoretical analysis and experimental results show that the proposed dynamic clustering algorithm can resolve the load balance problem,improve the network energy efficiency,and prolong the network lifetime.

In this paper,we discuss the minimum coverage breach and maximum network lifetime problem in directional sensor network.In our directional model,each sensor may have several sensing directions,but only one direction can be activated at the same time.In wireless sensor network,maximizing the network lifetime and minimizing the coverage breach are two conflicting objectives.To make a trade-off between them,we study Minimum Coverage Breach under Lifetime Constraint (MCBLC) problem and Maximum Lifetime under Coverage Breach Constraint problem (MLCBC).For MCBLC problem,we first formulate it as Integer Programming and then propose greedy algorithm (MCBLC-G) algorithm.For MLCBC problem,based on MCBLC-G algorithm,we use binary search technique to get a solution.Extensive simulations have been presented to demonstrate the performance of these algorithms.

Game theory is alternative framework to model confliction between independence players,which can be used to solve the problem.However,classic game theory couldn’t be applied to network directly.Game theory based routing approach MPS,which can “encourage merit and punish evil”,is presented.It rewards congestion avoiding users and punishes misbehaving users.A Nash Equilibrium could be reached when players are rational.

This paper introduces a particle estimation algorithm using Shcorrelation coefficient (PE) for nonlinear system state.It consists ofprediction,update,and smoothing.It modifies the weights of the particles using the Sh correlation coefficient between the observations of the estimated state and the observations of the particles.The simulation results are presented to demonstrate the improved performanc e of the SCPF over those known particle filters including the sequential importance resampling algorithm,the auxiliary particle filter,the regularized particle filter,the Gaussian particle filter,and the Gaussian sum particle filter.

To increase the memory reliability,the technology of memory built-in self-repair is growing rapidly.During the course of repairing,the BIRA SRAM has to be accessed continually,therefore the power dissipation is dramatically high.To solve this problem,a memory built-in self-repair method based on address partitioning strategy has been proposed.This method can decrease the power dissipation by simplify the address comparison course.Experimental results show that the proposed method can repair the defective cells in embedded memories and the power dissipation can be decreased simultaneously.

This paper proposes a method for the system dependability requirement elicitation based on experiential knowledge about how the software often causes issues to the environment.The basic idea of this approach is to capture deniability requirements by identifying the issues that the system may bring to the environment and determining the countermeasures to tackle them.To identify the potential issues and determine the countermeasures,we propose to utilize the accumulated knowledge about how the system fails and brings issues to the environment.By taking advantage of this kind of knowledge,the proposed approach can help the analysts to find more dependability requirements for the software to be developed.

This paper present the process of GPU-based volume rendering algorithm.Aiming at multiorganization for medical imaging calibration problem,we propose a two dimensions table based texture transfer function.For the problems of CPU and GPU resource allocation and tasks coordination,we propose a GPU based acting geometry algorithm to generate texture coordinates and cooperating with GPU to finish Volume Rendering.Finally,experiments have been carried out with a large number of medical imaging data and the result indicates that the algorithm proposed in this paper can be a good solution for multiorganization.The algorithm makes the calibration and reconstruction speed so efficient that the speed reaches a millisecond-level.The speed fully satisfies the clinical needs.

Internetware system is a kind of complex system that is typically situated in dynamic and open environments and adapts to the changes of environment.How to effectively develop such systems has become a great challenge in software engineering community.This paper introduces an agent-based integrated environment SADE for developing Internetware systems.It consists of a number of technologies for Internetware such as agent-based abstraction and construction,dynamic binding mechanism for self-adaptation and selfevolution,ODAM methodology,self-adaptive strategy description language SADL,etc.The technical framework of SADE and itscomponents are introduced in details,including ODAMTools to analyze and design Internetware,the programming toolkits to code Internetware and runtime infrastructure.A case study is illustrated to show how to develop Internetware system with our approach.

Inspired by the research in physiology,a novel algorithm for extracting bottom-up attention information (integration of contrast sensitivity and Markov chain,ACSMC) is proposed in this paper.In our algorithm,the original image is weighted with a contrast sensitivity formula which is a function retinal eccentricity to simulate the mechanism of retinal ganglion.A Markov chain is defined on feature maps.The equilibrium distribution of this chain is taken as saliency values.The average of algorithm cost time and area under receiver operating characteristic curve (AUROC) based on the research of neurobiologist demonstrate its effectiveness.