Abstract:In this paper, the authors introduce the basic ideas of the 15 AES(advanced encryption standard) candidate algorithms issued by National Institute of Standards and Technology (NIST) of USA. The current analysis results of these algorithms are also presented.

Abstract:VMDP is a visual model for distributed programming based on icons, which is used to illuminate the objects and the controls in a distributed system. With VMDP, a process is represented by a vehicle, a resource by a crossroad, and the coordination among processes by the traffic light. The coordination relations among processes and controls such as synchronization and mutual exclusion are understandable and readable in VMDP. VMDP devotes itself to building distributed system model, network programming, and parallel/distributed programming as well.

Abstract:The interactive segmentation techniques attract more and more attention in that these techniques can handle complicated images. In this paper, an interactive image segmentation method is proposed based on proxy model. The proxy is a task-specific module which carries out certain function. Communications between an operator and computer is realized by means of the control interface and the report interface of proxy. The proposed proxy is constructed on the basis of region features of a to-be-segmented region in an image. This method is applied to solve the segmentation problem in medical images and the result shows that this method is effective in such complicated image as medical image.

Abstract:The segmentation and recognition methods of the omnirange character string which is adhered or intersected with the lines in the engineering drawing vectorization are described in this paper. The directional calculation methods of the string are also supplied on different drawing distribution cases, the segmentation of the adhered string between characters is also realized by calculating the feature vectors of the adhered local string block image and using the iteration approachment algorithm. The authors use the projection method to solve the segmenting problem of the characters in the case of character adhered or intersected with the lines, which made the recognition accuracy of the omnirange string increased. The algorithm has the antijamming property for the local degradation character recognition in the engineering drawings.

Abstract:The main points of protocol performance testing is analyzed in this paper. The QoS(quality of service) of network protocol is described with a formal tool, stochastic Petri nets. Furthermore, the rules from the SPN(stochastic Petri nets) models to test suite of TTCN(tree and tabular combined notation) are proposed, and a performance test case is generated from the SPN models.

Abstract:In this paper, the authors show that assignments can be incorporated into pure functional languages without loss of referential transparency. And the denotational semantics of these assignment operations are given. Using these assignment operations, the authors define an interpreter of a simple imperative language.

Abstract:The maximum entropy approach is proved to be expressive and effective for the statistics language modeling, but it suffers from the computational expensiveness of the model building. An improved maximum entropy approach which makes use of mutual information of information theory to select features based on Z-test is proposed. The approach is applied to Chinese word sense disambiguation. The experiments show that it has higher efficiency and precision.

Abstract:The Sort-Merge-Join algorithm is an effective and widely used algorithm for implementing the important Join operation in database systems. The algorithm is revisited in this paper. It is discovered that sorting both operand relations externally is not necessary in the algorithm. The cost of the algorithm would be reduced greatly if only one operand relation is sorted externally. In order to overcome the shortcomings of the Sort-Merge-Join algorithm, a new Join algorithm called SDC-Join algorithm, is proposed in this paper. The SDC-Join algorithm is a single-relation-sorting based divide-and-conquer algorithm. A parallel version of the SDC-Join algorithm is also presented in the paper. Theoretical analysis and experiment results show that the performance of the SDC-Join algorithm is much higher than that of the Sort-Merge-Join algorithm in both uniprocessor computer systems and parallel computer systems.

Abstract:In order to support parallel programming, almost all of the programming languages incorporate a high-level parallel computing model into the language by providing parallelism and synchronization mechanisms, such as Ada's task with rendezvous and Java's thread with synchronized method. Obviously, such a programming language can only support one high-level parallel computing model. Although the approach of one model is simple and fairly effective to some applications, unfortunately the problems in the real world are always too diverse to be solved by using a single parallel computing model. In this paper, a new object-oriented approach to this problem is proposed. After analyzing the high-level parallel models of the various languages, some novel object-oriented language mechanisms are presented. Based on them, the concept of object-oriented parallel framework is proposed and the method for expressing and using the high-level parallel computing models by parallel frameworks is discussed. As a result, the high-level computing models are separated from languages in some sense and more than one framework could be given within a single language. Therefore, using a single language to support the various parallel models turns to be feasible.

Abstract:The rough set theory is introduced to deal with vagueness and uncertainty. In some aspects, the rough set theory overlaps with the Dempster-Shafer theory of evidence, but the rough set theory uses partitions to specify rough sets, lower and upper approximations, and then to capture uncertainty in knowledge representation. In this paper, directing against the discrepancy in the specification between the two theories, the authors explore their relationship in order for ones to understand them and open the way of applying them. In addition, in evidence theory, the basic aperation to combine evidences is the orthogonal sum, while in the rough set theory, the basic operation is the intersection of partition. Therefore, “Does the evidence combination correspond to the partition?” is the question which may be naturally raised. An example is presented to show that the answer is “no”.

Abstract:Current software reuse techniques usually utilize a repository to organize standard and generic reusable assets, such as class libraries and generic component repository in object-oriented programming environments. However, this repository-centered reuse ignores the project related information to some extent, which describes application context of reusable assets. Application context is useful to the understanding and utilization of the assets. In this paper, an alternative object-oriented reuse approach is presented, which utilizes the documentation of a project to organize reusable assets. This so-called project-centered documentation reuse is very useful to the development of a family of software in the same domain. In the paper, some limitations of current object-oriented methods for supporting documentation reuse are discussed and a new method for enhancing documentation for reuse is provided and analyzed. Then, aiming at browsing and maintaining reusable assets in documentation, the needs for navigation in documents are discussed and enumerated.

Abstract:The maximum clique problem is a wellknown NPcomplete problem. Previous algorithms are either applicable only to some particular graphs or in need of exponential time cost. In this paper, an algorithm is presented, which computes the maximum clique(s) based on the notion of interval representation of a graph. It can be applied to any simple graph and only needs polynomial time under the certain conditions.

Abstract:The solution on complex-valued wavelet basis and the corresponding construction of complex-valued filter bank are discussed in this paper, the real part of the complex-valued filter bank has linear phase, and its length is even which can be obtained from the complex-valued filter bank. Furthermore, the complex-valued filter bank is compared with other real filter using the same quantizer on coding of image, the complex-valued filter bank achieves well compressing performance.

Abstract:Communication subsystem is a major factor that affects the overall performance of a workstation cluster system. In this paper, the authors analyzed and compared the performance of three popular networks, and pointed out that the high level protocol processing is the bottleneck of communication in workstation cluster system. A user level high performance protocol, called RCP(reduced communication protocol), had been built on a cluster of 8 Sun SPARC workstations connected with 640Mbps Myrinet. Low overhead and high efficiency of the protocol are achieved by simplifying the protocol's redundant functions, reducing the times of data copying, and operating directly on hardware buffer. The round trip latency of RCP is much lower than TCP/IP(200μs vs 1 540μs), the application available bandwidth is much improved(178Mbps vs 34 Mbps) and the bandwidth usage of Myrinet on RCP reaches to 80.5%. A very simple interface to application or PVM is provided as well.

Abstract:Approximate reasoning with the incomplete information is one of the difficulties that the knowledge engineering has faced. A precedent logic program theory with the property of nonmonotonicity is proposed in this paper. The synthesis evaluation for the interpretation of knowledge can be taken with the theory, such that the optimal selection of interpretation is made possible which becomes the best approach to the current knowledge. The theory completion in the significance of optimal selection is achieved and the requirement of completion and consistency of knowledge are avoided. To acquire the precedent logic programs in the applications, based on an inductive logic programming, learning algorithm is presented which incorporates the multiple inductive methods and has greater ability of induction. The presented theory and the algorithm have been applied in an expert system and gained satisfactory results.

Abstract:The computational and compositional features are very important in the construction of software for the workstation clusters. However, due to the lack of suitable supporting environment of software development, most existing distributed parallel software systems are weak in these two aspects, especially in the compositional feature. In this paper, a distributed object based framework for parallel programming is proposed. The goals of the framework are: first, getting high parallel computing efficiency; second, constructing a mechanism to encapsulate and reuse parallel program. The framework is tested by some parallel algorithms, the results indicate that the framework is helpful.

Abstract:In this paper, the authors propose a fast B wavelet transform method in spectrum domain and step spectrum analysis theory for extracting completed magnetic resonance(MR) spectrum data from the MR truncated data, and the reconstruction image. In this method, the feature information is abstracted from the available low-frequency MR data using B wavelet method, and then high-frequency components are exerted from the feature information and whole spectrum data is recovered by using step spectrum analysis theory(SSAT), at last MR image is reconstructed from the recovered completed spectrum data by FFT(fast Fourier transform). Experiment and simulation results show that this method gives much better reconstructed image than the available method.

Abstract:In this paper, two classes of directed networks——ORC-networks and IRC-networks are defined, and a polynomial time algorithm is presented for computing their rooted communication reliability, i.e. the probability that a specified vertex, root vertex, can communicate with all other vertices. The complexity of the algorithm for ORC-networks and IRC-networks is O(｜E｜) and O(｜V｜·｜E｜) respectively, where ｜V｜ and ｜E｜ are the number of vertices and of edges of networks respectively.

Abstract:The simplification of Bézier curves in fitting and editing is discussed, three kinds of node-reduce operators are defined, and the algorithms based on them are given in this paper. The simplest Bézier expression can be gotten through the algorithms, and a strict proof is given. These algorithms have been already used in the developing software.

Abstract:First, the two notions: weakly definable principal congruence and definable subdirectly irreducible class are introduced in this paper. The authors prove that if a variety generated by a finite algebra has both weakly definable principal congruence and definable subdirectly irreducible class, its equational system is finitely axiomatizable. Further discussion shows that the results are new, and are significant generalization of the known results.