最新刊期

摘要：By the use of a novel definition of temporal aberrations,the temporal aberration theory of electron optical imaging system was again investigated by the "Tau-Variation Method".Results show that the expressions in integral form can also be obtained by the "Tau-Variation Method" which usually needs to solve the differential equations for the three of geometrical temporal aberration coefficients of second order.It also proves that the "Direct Integral Method" and "Tau-Variation Method" reach the same goal by different ways.  

摘要：A new low-voltage reflective display pixel for electronic paper based on electrowetting-on-dielectric (EWOD) is proposed.The principle of the display pixel and the influences on the performances were analyzed,and the applied voltage was reduced by using the multiple dielectric layers of a Si₃N₄ film and a fluorocarbon polymer film.The on-switching and off-switching of display pixel were successfully realized by applying the low voltage of 15V,and the response times were less than 1/30s.  

摘要：The image quality suffers from the lack of grayscale expression capability when the 'simple-accumulation’-lighting-mode is applied to alternating current plasma display panels (AC PDP).To solve the problem,an adaptive subfield coding(ASC)driving method was proposed,in which subfield vector is adaptively figured out based on the normalized histogram of input image every TV field.The calculated subfield vector can express the most grayscale information of input image.And then the sustain pulse number of each subfield is adjusted according to the calculated subfield vector.Simulation results show that,the proposed method can eliminate the dynamic false contours in AC PDPs,and overcome the lack of grayscale expression capability caused by the simple-accumulation’-lighting-mode.It can provide good grayscale expression in AC PDPs.  

摘要：For the purpose of stereo image matching,several common stereo image interest point detectors are reviewed in the two aspects of repeatability and information content through experimental analysis in this paper.Based on the image information entropy,clustering is utilized to analyse the texture feafures of images,and a new interest point detect method is then proposed.The most significant characteristic of this method,which improves the Harris detector,is self-adaptive to the image texture.The experiments by using of actual stereo images with different texture features proved that this method guarantees more better repeatability and information content.  

摘要：A support vector machine neural network is proposed for performing VBR video traffic prediction.It takes differential signal as the input of the network.According to the criteria of structural risk minimization of SVM,the errors between sample-data and model-data are minimized and the upper bound of predicting error of the model is also decreased simultaneously so that the ability of generalization of the model is much improved.The simulation results show that the prediction error of SVM neural networks is 0.0018,while the prediction error of GRBF neural networks is 0.0029.In prediction precision,SVM neural networks model extended 40% than GRBF model.  

摘要：It's crucial to optimize global interconnect performance in ultra deep sub-micron System-On-a-Chip designs.In this paper,the optimized method to improve delay-bandwidth performance under crosstalk constraint,suitable to various technology nodes,is proposed from analytical formula based on distributed RLC model.It's verified through HSPICE simulation that,by utilizing this method,the optimization of edge-length global interconnect of mainstream 90 nm technology can achieve delay of 182 ps,data bandwidth of 1.43GHz/ μ m,and near line peak crosstalk voltage within 0.096V_dd.Through cross-section parameters and performance targets determined by this method under various technology nodes,the trend of global interconnect design is reasonably predicted along with continuously scaling down of semiconductor devices.  

摘要：We define the key segment of a popular song as the most impressive part in this song.It can be regarded as a perfect index or summary for this song,therefore it is much helpful for popular song's audition and management.We propose an effective algorithm for automatic key segment extraction based on the analysis of positive and negative samples of key segment extracted manually from a training database.An experiment is carried on a test database including 130 popular songs.Compared with manually extraction of key segments,the proposed algorithm has a rather satisfying performance.  

摘要：The feature of supporting dynamic change is required in the application of workflow management system.While some relational research fell short in terms of expressive capacity and algorithm complexity.Based on synchronization net,a formal workflow model,this paper proposes a definition of process change from logic and semantic view,then classifies elementary change units,and discusses the correctness of change.A whole process change in need is considered as a sequence of those change units.Therefore the complex problem becomes easy and the correctness of change is preserved strictly.Following the description of change,a new method to manage process change is proposed,so as to resolve the migration problem of on-the-fly process instance,centered on synchronizer.  

摘要：Label is denoted by disparity and the energy function is established.Then the problem of matching can be transformed into that of energy function minimization.A network is constructed such that the energies can be related to the capacities of the cuts of the network.Finally,the minimal energy is obtained by the network-flows theory,and hence the disparity data are solved.Comparing with some known algorithms based on graph cuts,the algorithm in this paper extends the label from 1 dimension vector to 2 dimension vector,and adapts vision matching of more general conditions;furthermore the algorithm can gain the minimization in global.Experimental results show that the algorithm has a high accuracy.  

摘要：This paper addresses the issue of nondeterminacy of thread suspension point in user-level process checkpointing,and proposes a solution to it.In addition,the workloads of a process are multithreaded in terms of rollback recovery to exploit effectively system resources for reducing the overhead.The presented schemes are implemented in rollback recovery testbed WINDAR.Experimental results show that the presented scheme improves the performance of rollback recovery,especially for pessimistic message logging protocols.  

摘要：Game theory is applied to study the resource allocation technologies further in this paper.The mechanism of network resource pricing,which shows the relation of supply and demand,was presented based on bidding.Subsequently,we designed the utility function of end system and proved the existence and uniqueness of Nash Equilibrium in resource allocation game.At last,we present a network resource allocation algorithm (RANG) based on the resource allocation game model.The results of experiment and simulation show that RANG algorithm can provide references for the quantity of the resource that users demand and regularize the users' bids.Consequently,the allocation of the whole network resource tends to be more reasonable.  

摘要：An algorithm combining multiple Nave Bayesian (NB) filters based on GMM is presented,which has been successfully applied to e-mail filtering.The method uses the multiple variates statistics analysis to model the relationship between the training data set and their classification by a collection of NB filters.Then a GMM can be learned from the resulting representation.The GMM filters previously unseen e-mails according to the principle of minimizing expected-error-cost,in order to avoid deleting useful e-mails.Experimental results confirm the validity of our method,and show that our approach is insensitive to ratio of feature subset selection.  

摘要：The concept of proposition induced functions is proposed in the present paper,then the concept of truth degrees is introduced by means of infinite propduct of evenly distributed probability spaces and integrated semantics respectively w.r.t.discrete and continuous situations.Next,a graded approximate reasoning theory is established.Finally,theory of consistency degrees of finite theories is also proposed.  

摘要：Clustering is a promising application area for many fields including statistics,pattern recognition,datamining,etc.The effectiveness and efficiency of existing clustering techniques,however,is somewhat limited,owing to the huge amounts data collected in databases.According the theory of fields in physics,a hierarchical clustering method based on data fields is presented.The basic idea is that the field models is introduced to describe the virtual interaction among data objects in data space and the hierarchical partitioning of the original dataset is then performed by iteratively simulating the interaction and movement of the data objects in the fields.Experimental results show that the proposed approach not only enjoys favorite clustering quality and requires no careful parameters tuning,but also has a time complexity approximately linear with respect to the size of dataset.  

摘要：Using weil pairing,an identity-based ring signcryption scheme is proposed.Its concrete algorithm is given.Using this scheme the message sender can anonymously send the message,the confidentiality and authenticity of message are realized at the same time.It is proved that the scheme is secure against adaptively chosen cipher text attack under the difficulty of the Decisional Bilinear Diffie-Hellman problem.Compared with the traditional "signature then encryption scheme",the cipher text of our scheme is rather short.So our scheme is more applicable to systems where cryptogram is sent over a low bandwidth channel.  

摘要：The first semiconductor nerocomputer CASSANDRA-I based on host-PC is analyzed,and the realization and characteristics of the novel semiconductor neurocomputer CASSANDRA-II are described.Finally,the semiconductor neurocomputer CASSANDRA-II is used to greeting speech recognition,the experiment results show that it provides higher discriminability for keyword candidates than HMM model.  

摘要：Security of routing protocols is the most important secure factor in the mobile Ad Hoc networks.AODV protocol is used widely in the Ad Hoc networks for its simplicity and small control overhead,but it does not have any security mechanism.With some potential insecure factors in AODV protocol analyzed,some necessary improvements are made to AODV protocol,including adding attack detection and building up credence mechanism,which would collaboratively guarantee the network security.Simulation results indicate that it can react quickly when some malicious behaviors in the network are detected and effectively protect the network from kinds of attacks.  

摘要：For a long time,the classification of species depends mainly on morphologic methods.And it is difficult to realize automatic species classification by using computer.This article puts forward a alignment-free sequence comparison method based on whole genomes and draws phylogeny trees for 9 Hepadnaviridae viruses and 14 Caulimoviridae viruses with this method.The results match the standard of virus classification published by ICNV.On this foundation,the article exploringly puts forward an automatic classification method of species based on pattern recognition and uses this method to do the automatic classification for Hepadnaviridae viruses and Caulimoviridae viruses.The correct classification rate reaches 100% and 94% respectively.  

摘要：It is well-known that leakage power dissipation caused by leakage current in CMOS circuits during test periods has become a significant part of the total power dissipation.It is important to reduce leakage power to improve battery life in portable systems employing periodic self-test,to increase reliability of test and to reduce test-cost.This paper first analyzes leakage current,and introduces the transistor stacking effect relevant to it.Then,we present a method based on don't care bits (X) in test vectors and using the genetic algorithm to optimize leakage current in IC test.Experimental results indicate that this method can effectually optimize leakage current of combinational circuits and sequential circuits during test,and it doesn't lose fault coverage.  

摘要：Backbone technology has been proposed to simplify network structure and improve network performance.Without regard to fault tolerance,however,this will degrade the reliability of Ad hoc networks and make communications through constructed backbone susceptible to disabilities of nodes or wireless links.In this paper we explore how to improve network reliability through fault-tolerant backbone.We proposed a localized algorithm (LKFB) to construct the fault tolerance backbone,based on which the maximum extent K fault-tolerant connectivity for any two nodes in the network could be preserved.Simulation results show that by using LKFB network structure can be simplified since the number of gateway nodes has been decreased,and meanwhile by adjusting K network reliability can be controlled.Finally energy-aware fault-tolerant backbone could be constructed through modification of the algorithm's weight function.  

摘要：Presently existing correlation analysis method for multiple data streams were all oriented single dimensions data streams only,which could not identify the real correlation between fields built by multiple variables.To quickly detect correlations between two multiple dimension data streams under constrained resources,a novel correlation analysis algorithm based on canonical correlation analysis (CCA),called StreamCCA,is proposed.Focusing on the computational bottleneck of traditional CCA,StreamCCA introduces a low-rank approximation technique to reduce the dimensionality of product matrix resulted from sample correlation matrix and sample variance matrix,which improves computational performance efficiently on the premise of holding approximate precision.Theoretic analysis and experiments results on synthetic and real data sets indicate that StreamCCA can online detect correlations between multiple dimension data streams accurately.The algorithms proposed herein,are presented as generic forecasting and diagnosis tools,with a multitude of applications on data streams mining problems.  

摘要：A new adiabatic logic circuit adopting two-phase non-overlap power clocks-Clocked Transmission Gate Adiabatic Logic circuit was designed by using the bootstrap effect of NMOS transistors,so that it could charge or discharge output loads in a fully adiabatic manner.Based on this circuit,a novel adiabatic SRAM was designed.So it could recover the charge of large switching capacitances on word-lines,write bit-lines,sense amplified lines and address decoders in a fully adiabatic manner.Using the parameters of TSMC 0.25 μ m CMOS device,the adiabatic SRAM designed was simulated by HSPICE.The simulation results indicate that this SRAM has correct logic function and the character of clearly low power.  

摘要：This paper analyzed optimal coverage for routing in three-dimensional wireless sensor networks.For this NP-hard combinatorial optimization problem,we proposed a new solution used techniques from Computational Geometry,Graph Coloring and established a new model for optimal coverage.We also proposed an energy-efficient distributed heuristic algorithm to find the optimal coverage routing path.Based on the model and algorithm,we presented a three-dimensional optimal coverage routing protocol in a distributed manner.To evaluate the performance of the algorithm,a time complexity analysis was given for each operation.In the end,we analyzed the coverage quality and compared the network lifetime with several typical protocols.Simulation results show that proposed routing protocol has low time complexity and it is considerably effective,scalable and robust.  

摘要：To improve the speed and the problem of high false positive in micro-calcification detection,a novel micro-calcification detection method based on support vector classifier model with rejection feature and Iterative Rank-order Filter Subspace is proposed.In the training step,iterative rank-order filter is used as coarse detector,and the non-micro-calcification training samples are selected from the constrained subspace.So it overcomes the effect of number difference between the negative and positive samples,and at the same time it reduces the computing load in training stage.In the detection step,coarse detection is applied at first to find suspect micro-calcification region,then Firs the first layer of support vector classifier (SVC) with maximum margin between two classes will be used for classification the input pattern;and the sphere support vectors of true micro-calcification points describing the distribution of the sample are obtained by searching all the sphere boundaries containing the samples of this class.So the input pattern of no-object classes can be rejected by the second support vector domain description (SVDD).Lastly the results of SVC and SVDD classifier are integrated to obtain the accurate results.Experimental results demonstrate that the new calcification detection method performs better in achieving lower false positive rate (FPR) and fast speed.  

摘要：There have been so far three prestigious schools of AI theory respectively established through the structural,functional and behavioral approaches to human intelligence whereas there have been also,however,controversies among them over the theme of which one of the three should be the one who dominates.It is the author's belief that all the three schools did not cover the essence of intelligence.A new approach,mechanism approach,is thus proposed in the paper to cope with the situation.It is discovered that the mechanism of intelligence formation should be a series of transformations——from information to knowledge and further to intelligence that are discussed in the paper.Interestingly but not surprisingly,the three existed schools of AI theory can well be unified under the framework of the mechanism approach.The mechanism approach and the unification of AI may hopefully provide a new foundation for the future development of AI theory and applications.  

摘要：With people rely more on computer networks,the study on the dependability of networks is increasingly significant.First the paper provides insight about the main definitions and attributes related to dependability,and classifies and compares the general models for dependability.On the basis of the synthesis,the modeling and analysis methods using Stochastic Petri Net (SPN) for the network system dependability are mainly investigated.The service failure and fault tolerance SPN models and the calculating methods of main dependability attributes are examined.The existing problems and applications of SPN are concluded that concern the dependability of networks. And directions for future research are also indicated.  

摘要：Abstract:Many kinds of statistical word sense disambiguation (WSD) methods and technologies in home and abroad are analyzed,and lots of literatures are referred to in this paper.Key questions of statistical WSD research are pointed out,and the research progress is illustrated around them.In the final section the problems and future research emphases in WSD are explored in this paper.  

摘要：A real-time system is required to complete its work and deliver its sevices on a timely basis.We lays much emphasis on hard real-time characteristics of the system,because hard real-time performance is one important performance requirement of special system such as avionics network.The periodic task model is a well-known deterministic workload.With its various extensions,the model characterizes accurately many traditional hard real-time applications.Switched fabric is one of basic fibre channel topology.Fixed-length packet scheme is mainly used in fibre channel switches.Taking the study on fibre Channel switch fabric under hard real-time constraints as background,an optimization design of typical message set is proposed according to the weighted round-robin algorithm.It is derived and formally proved that the achievable utilization using the optimization design can be doubled at least.  

摘要：Optimal sensor placement is one of the important methods for optimal design of mechatronic BIT system.Effective detection is the important constrain for optimal sensor placement.In fact the reliability of sensors must be considered while optimal designing the system because it would influence the effectiveness of system.Analyzed the detectable model,a mixed integer nonlinear programming (MINLP) model is proposed for optimal placement of sensors in this article.MINLP is based on detectable model of fault mode of equipment and sensors of BIT system.The objects of optimization are minimization of failure probability and cost of selected sensors.The constraints of this model involves the influence of failure probability of sensors on the FIR,FDR and FAR of mechatronic BIT systems.The solution can provide a optimal scheme for sensor placement.MINLP is applied to a detection system for a Servo system and its practicability is analyzed.  

摘要：A new method for measuring the noise parameters of a FET/pHEMT device is proposed.By measuring the matched (50 Ω ) noise figure F₅₀,based on the intrinsic H-parameters and the intrinsic chain noise matrix C _<em>A^INT,the linear equation of the gate noise temperature T_g and drain noise temperature T_d is obtained.The statistic analysis of the T_g-T_d figure is performed at all frequencies range,and then the measured noise temperatures T_g and T_d are determined.In terms of noise network synthesis theory,the full chain noise matrix C _<em>A is computed,and the corresponding noise parameters(F_min,R_<em>n and Γ _opt) are determined.The measurements for three FET/ pHEMT devices show that the results are perfectly in accordance with these by means of Garcia and Lázaro.  

摘要：Some models have been established to calculate the static dielectric constant of aqueous electrolyte solution.Others models based on the fitting parameters in Debye's equation by a lot of experiments and consequently were used to calculate the real and imaginary part of the complex effective permittivity of few aqueous electrolyte solutions,such as saline water and seawater.Here,the effective permittivity for various aqueous electrolyte solutions at microwave frequency has been gotten by the experiments,and we introduce the complex impact factor of different ion based on the experimental data and proposed a new model to calculate the complex effective permittivity of aqueous electrolyte solutions at microwave frequency.The calculated results are compared with the measured results at 915MHz and 2450MHz.The good agreement can be observed.This study offers base to calculate the distribution of electromagnetic field and temperature in the chemical reaction in dilute solution irradiated by microwaves and high potential applications of the method in the fields of geophysics,remote sensing,and inspection of water pollution.  

摘要：For typical topology of buck converter,only one terminal of the inductor is connected to the chip.So it is impossible to parallel a resistor with the inductor to damp the ring as it is used in boost converter.A novel anti-ringing circuit is presented to solve this problem.A time varying resistor network is connected between one terminal of the inductor and the power supply or the ground.In the beginning of the ring,a small resistor is used to damp the ring rapidly.But the DC current is large.With the weakening of the ring,the resistance is increased to reduce the DC current gradually.And when the ring disappears and the DC current is small enough,the network is opened.By this way,anti-ringing is achieved and no DC current exists after the ring damping is ended.A DC-DC converter with this anti-ringing circuit has been implemented in Hynix 0.5 μ m CMOS process,the result indicates that it works well and effectively.  

摘要：Internet is characterized as a uncontrollable,variable and unpredictable infrastructure,this brings the difficulty of developing,debugging and evaluating network applications.Network emulation system constructs a virtual network environment which has the characteristics of controllable and repeatable network conditions.This makes it possible to predict the correctness and performance of proposed new technology before deploying to Internet.In this paper we present a methodology for predicting the correctness and performance of applications based on the PARNEM,a parallel discrete event network emulator.PARNEM employs a BSP based real-time event scheduling engine,provides flexible interactive mechanism and facilitates legacy network models reuse.PARNEM allows detailed and accurate study of application behavior.Comprehensive case studies covering bottleneck bandwidth measurement and distributed cooperative web caching system demonstrate that network emulation technology opens a wide range of new opportunities for examining the behavior of applications.  

摘要：If the substrate of larger electromechanical coupling coefficient (such as YZ－LiNbO₃ substrate) is used for wavelet type of interdigital transducer (IDT),its frequency characteristic curve is not smooth,but its insertion loss is smaller.If the substrate of smaller electromechanical coupling coefficient (such as X 112°Y－LiTaO₃ substrate) is used for wavelet type of IDT,its frequency characteristic curve is smooth,but its insertion loss is larger.The less the finger pairs N of surface acoustic wave (SAW) element,the larger its bandwidth Δ f/f,and the larger its insertion loss,so the finger pairs N is appropriately taken,and is not too small.In wavelet transform element of SAW type,the finger pairs N of the transmitting IDT is 57,the finger pairs N of the receiving IDT is between one-fourth of 57 and one-third of 57.  

摘要：This paper presents a novel unsupervised image segmentation algorithm based on hidden Markov random field (HMRF) model.For each order model segmentation the proposed algorithm makes use of the correlated information between adjacent models.Therefore the algorithm avoids the drawback about that mean field algorithm is restricted by initial condition.Furthermore,in order to solve the model selection problems of unsupervised image segmentation,the sum of squared error criterion with penalty term is proposed.The experiment results testify that the proposed criterion is superior to the Pseudo-likelihood Information Criterion (PLIC),and it is shown that the performance of the segmentation is satisfied.  

摘要：QoS (Quality of Service) of control networks,especialy the deterministic delay,plays a decisive part in the performance evaluation and design of the Network Control Systems (NCS).Network calculus gives a method to get the end to end QoS parameters,such as delay,backlog,etc.,from the network resources.We extend the calculus model of computer networks and present a general calculus model,which is suitable for control networks better.At last,a deterministic delay of Control network with Line and Star topologies is derived.