Abstract:This paper presents an approach to optimizing performance of middleware applications based on anti-pattern. This approach has three major features: First, a meta-model is offered to build more understandable and formalized representation of anti-patterns; second, the detection of anti-patterns is based on both the static and the dynamic information, which is retrieved at runtime; third, refactorings operate without interrupt the running systems, and is completed in an automated way with the help of the middleware. A prototype based on J2EE has been developed and an e-bookstore is used as a running example to illustrate the ideas introduced in this approach.

Abstract:This paper presents a dataflow-style Java parallel programming model with a runtime profile based thread duplication algorithm to exploit data level parallelism. Furthermore, a new dataflow polymorphism feature is introduced. This model has been implemented in an open source Java virtual machine. Evaluations on real machine show good speedup for benchmark applications.

Abstract:This paper presents RacsCBPS, an admission control scheme for large-scale and scalable content-based publish/subscribe systems. First the key requirements to implement admission control in content-based publish/subscribe systems are identified and how it differs from admission control schemes in the Internet and other research areas is analyzed. A cover-relation based algorithm to compute subscription resource requirements and an admission control algorithm based on subscription routing are presented. The scheme ensures time, space and control decoupling without sacrificing scalability of publish/subscribe systems. Publish/Subscribe systems can seek different balances between system resource utilization and QoS guarantee by choosing different admission control criteria. Experimental results show the effectiveness of the method.

Abstract:A motivation of MDA (model driven architecture) is to use models as programs so as to increase the abstract level of software development. Design patterns provide reusable constructs that can be introduced into MDA to increase the modeling granularity as well as the reusability of model transformation rules. This can be achieved by applying design patterns as the integrated development units in MDA, which raises two problems of modeling and model transforming. First, patterns should be used as integrated modeling units. Second, pattern models should be transformed by applying pattern transformation rules built on the pattern units. To solve these two problems, a solution is to define pattern unit metamodel for each pattern and to provide the EJB model transformation rules basing on each pattern unit metamodel respectively. In this way, patterns can be used as integrated modeling units by instantiating the pattern metamodel units, and then the pattern models can be transformed into the EJB models by applying the transformation rules. This paper also demonstrates how to extend MOF (meta object facility) meta-metamodels to define the pattern specifications and how to define the transformation rules to map the pattern models as well as related business models onto the EJB platform.

Abstract:This paper firstly presents an aspect-oriented call graph (ACG for short), then introduces an AspectJ dynamic advices weaving optimizing method based on the ACG (aspect-oriented call graph) of AspectJ programs. Our method firstly solves a call stack through the ACG and deduces the types of the nodes in the stack, then match the call stack with pointcuts, and finally decides how to weave dynamic advices based on the result of matching. A case study shows that this method has great precision and can identify most of weaving points statically.

Abstract:Due to the increasing speed gap between memory system and processor, cache hierarchies have been implemented into memory system, but additional latency (cache penalty) is introduced. This paper presents an algorithm named as prevent cache penalty by loop partition-unrolling (PCPLPU), which can prevent cache penalty in loops by the combination of loop partition and unrolling. Experimental results show that PCPLPU can prevent cache penalty and improve the performance of programs.

Abstract:AI planner is one of the important representations of AI planning study, the performances of planner, the efficiency and quality of plan, represent the researches of AI planning directly. This paper introduces the architectures of AI planner and StepByStep planner in brief, then describes the methods and strategies adopted by SteByStep in detail, defines the knowledge tree of predicate for extracting domain knowledge. Based on knowledge tree of predicate, the planning tree of predicate is defined and some strategies are applied for constructing the planning tree fast. StepByStep adopts some planning strategies according to the planning trees. Finally, some experiments are taken for eight planners with three representative benchmark domains and their problems, the performances of StepByStep, the efficiency and quality of plan, are analyzed in detail. Experiments show that the strategies of StepByStep controls the process of planning the problems of the three domains well, and validate the guide effect of the domain knowledge in planning process.

Abstract:Topic link detection is a foundational research in the field of topic detection and tracking, which detects whether two random stories talk about the same topic. This paper proposes a method of applying semantic domain language model to link detection, based on the structure relation among contents and the semantic distribution in a story, and also verifies the influence of the strategy incorporating dependency parsing into semantic description. Evaluation on Chinese Corpus of TDT4 show that the semantic domain language model substantially improved the performance of current detection system, whose minimum DET cost is reduced by about 3 percent.

Abstract:To emphasize the fuzzy relation among words, latent concepts, text and topics, an information theory based approach to latent concept extraction and text clustering is proposed. Latent concept variable and topic variable are introduced to reveal such relation, and a global objective function is defined in the theme of rate-distortion theory. An anneal-like algorithm is designed to extract the hierarchical tree of latent concept, and to group the texts under corresponding concept hierarchy at the same time. Furthermore, it determines the number of concept and text clustering result with a concept selection method based on minimal description length criteria. It is a soft co-clustering method and outperforms the ones based on the word space, and current text hard co-clustering method based on latent concept by experiments.

Abstract:Behavior sequence segmentation is the first and most fundamental step of behavior analysis and recognition. In this paper, a novel unsupervised algorithm for behavior sequence segmentation is proposed. The algorithm consists of the following steps: (1) The video sequence is coarsely segmented into equal length subsequences with overlapping time window; (2) Segmental-DTW is used to find out matching behavior clips between pairs of video subsequences; (3) The similarity between behavior clips is represented by an adjacency graph, and an efficient graph clustering algorithm is used to generate behavior clusters. The algorithm, based on a coarse-to-fine strategy, is able to satisfactorily segment behavior sequences and cluster typical behavior patterns. The segmentation results can be used for further behavior modeling and recognition. Experimental results show the behavior clips segmented by this algorithm are prototypical and meaningful.

Abstract:In this paper, a feature matching strategy is developed. It is realized by introducing a function whose independent variable is the match matrix, which describes the correspondence of the features. It combines spatial relations and feature similarity organically and makes sure that its global maximum can be reached when the sensed image is aligned with the reference image completely. Thus the feature correspondence can be estimated by finding the maximum of the function. The branch-and-bound strategy is employed in the integer programming problem. A lot of real images are used to demonstrate its performance. Compared with some existing methods, it is automated, robust, and has the highest accuracy.

Abstract:Recognition of complex dynamic gesture is a key issue for visual gesture-based human-computer interaction. In this paper, an HMM-FNN model is proposed for gesture recognition, which combines ability of HMM model for temporal data modeling with that of fuzzy neural network for fuzzy rule modeling and fuzzy inference. Complex dynamic gesture has two important properties: Its motion can be decomposed and usually being defined in a fuzzy way. By HMM-FNN, complex gesture is firstly decomposed into three components: Posture changing, movement in 2D plane and movement in Z-axis direction, each of which is modeled by HMM. The likelihood of each HMM to observation sequence is considered as membership value of FNN, and gesture is classified through fuzzy inference of FNN. In this proposed method, high-dimensional gesture feature is transformed into several low-dimensional features, as a result, computational complexity is reduced. Furthermore, human's experience or prior knowledge can be used to build and optimize model structure. Experimental results show that the proposed method is an effective method for recognition of complex dynamic gesture, and is superior to conventional HMM method.

Abstract:In this paper, evaluation function of MVC (multiview video coding) random accessibility for decoder and server is first proposed based on the methods from random graph theory and hyper-space. Furthermore, a non-linear multi-purpose programming model for MVC performance evaluation is also proposed according to the constraints from practical network bandwidth, restrictions of interaction and other problems. Based on this model, optimal coding strategies for different MVC schemes under variant performance constraints are discussed. It is found that random accessibility of MVC scheme can be improved via this non-linear multi-purpose programming model.

Abstract:A projection function called minimal neighborhood mean projection function (MNMPF) is proposed. The projection function calculates and stores the minimal neighborhood mean of each pixel on each projection line, so that it is able to trace the low grayscale features in image. Compared with traditional projection functions, i.e. integral projection function (IPF) and variance projection function (VPF), MNMPF is insensitive to sheet noise, due to the local selectivity of its minimum operation. During the computation of MNMPF, the image locations of minima are recorded at the same time. This makes MNMPF a 2D operator. All these properties of MNMPF are very suitable for eye location. It can bring precise and robust response to eyes, especially pupils. Experiments on CAS-PEAL and BioID databases show its excellent correct rate and precision over traditional projection functions.

Abstract:Summary-Based retrieval is based on the hypothesis that terms in summary should be more important than other terms not in summary. Recent developments in the language modeling approach to information retrieval have motivated the study of this problem within this new retrieval framework. In the proposed research, two approaches to summary-based retrieval, namely ranking documents directly (SQL) and smoothing documents with summaries (SBDM) are investigated. Results on TREC collections show that, with the proposed models, summary-based retrieval models can perform consistently across collections and significant improvements over document-based retrieval can be obtained. There are two main contributions in this paper. On the one hand, summarization method of retrieval-oriented is examed and effect of this method on information retrieval. On the other hand, the new retrieval model for summary-based information retrieval models is proposed.

Abstract:Distributed clustering is an effect method for solving the problem of clustering data located at different sites. Considering the circumstance that data is horizontally distributed, algorithm LDBDC (local density based distributed clustering) is presented based on the existeding algorithm DBDC (density based distributed clustering), which can easily fit datasets of high dimension and abnormal distribution by adopting ideas such as local density-based clustering and density attractor. Theoretical analysis and experimental results show that algorithm LDBDC outperforms DBDC and SDBDC (scalable density-based distributed clustering) in both clustering quality and efficiency.

Abstract:In this paper, a method called MABI (moving average based indexing) is proposed to effectively deal with the issue of (-search query in subsequence matching. Two important theorems, distance reduction theorem and DRR(distance reduction rate) relation theorem, are proposed here to be as the basis of MABI. DRR relation theorem has strong capability in "pruning" those unqualified candidate sequences so as to achieve of fast similarity search. Furthermore, by modifying BATON^* introduced by Jagadish, et al., a multi-way balanced tree structure is introduced, to construct the index from time series, which significantly speeds up the similarity search. Extensive experiments over a stock exchange dataset show that MABI can achieve desirable performance.

Abstract:In order to enhance the search results of keyword search in relational databases, semantic relationship among relations and tuples is employed and a semantic ranking function is proposed. In addition to considering current ranking principles, the proposed semantic ranking function provides new metrics to measure query relevance. Based on it, two Top-k search algorithms BA (blocking algorithm) and EBA (early-stopping blocking algorithm) are presented. EBA improves BA by providing a filtering threshold to terminate iterations as early as possible. Finally, experimental results show the semantic ranking function guarantees a search result with high precision and recall, and the proposed BA and EBA algorithms improve query performance of existing approaches.

Abstract:Although nodes in peer-to-peer networks are viewed as equal entities in function, some of them have much more importance than others in the aspect of overlay topology. This paper introduces the concept of "partition node" to describe the topologically-critical nodes, whose failure may potentially lead to overlay partitioning. Then a simple, effective and distributed method to detect and avoid partition nodes, is proposed. The results of simulation show that the proposed method can optimize the overlay topology and remarkably improve the fault tolerance of unstructured P2P systems under a dynamic environment.

Abstract:In Ad Hoc networks, there exist two fundamental characteristics: Multi-Hop wireless-based transmission and node mobility. The former results in different contention relationship among flows from that in wireline networks, and the latter leads to the time-varying network situations. Firstly, on the basis of the link's interference set depicting the characteristics of the contention relationship, the congestion control problem for small time interval is formulated as a nonlinear optimization problem. Secondly, by using the dual decomposition theory, a price cooperation approach (PCA) is proposed to solve this optimization problem. In PCA, a price framework based on the link's interference set is built. Meanwhile, to implement PCA in realistic ad hoc environment, three deployment techniques are proposed: Queue Size Monitoring, Neighbor Set Approximation, and HELLO-based Message Piggybacking. Otherwise, the network status detection and receding optimization is introduced to deal with the uncertain changes of network situations, and an adaptive optimization strategy (AOS) is proposed correspondingly. The simulation results in MATLAB environment show that AOS has better performance of adaptation to time-varying network situations than PCA. The simulation result in NS2 environment show that PCA and PCA+AOS significantly outperform TCP, ATCP, and ATP in many important performances, including throughput, packet drop ratio, and fairness, under a variety of scenarios and mobility models.

Abstract:Current anonymous routing protocols do not provide anonymous mechanism for multicast in MANETs and have only had ad-hoc anonymity analysis. This paper proposes a new scheme called probabilistic multicast with universally composable anonymity. One-Time key pair is used to keep a route record in privacy during route discovery processes. Gossip-Scheme, secret DH path and Bloom Filter are adopted to realize anonymous source multicast during transmission of data packets. Finally, the protocol is analyzed based on the UC (universally composable) framework and its performance is evaluated by simulation. The analysis and simulation results show that the proposed scheme provides both anonymity and reliability for multicast in MANETs.

Abstract:In this paper, an efficient grid-based sense scheduling method for event detection in Wireless Sensor Networks is proposed. After grid-partitioning sensors in target environment, the sensors within each grid elect one representative to carry up the sense task in turn, which has low communication overhead, balanced energy consumption and good scalability. It is shown by analysis and simulation that the grid-based sense scheduling can achieve energy saving proportional to the node density and better detection delay than random scheduling. The method also provides means to balance network lifetime and detection quality by adjusting size of grids and duty time of nodes.

Abstract:Based on previous observation, a dynamic adaptive cycle into gossip-based topology management is introduced to replace the traditional fixed cycle and a quick topology convergence method based on adaptive cycle is also proposed . In this method, those nodes featured with local topology stability send fewer gossip packets, while nodes in frequently changing environment send more packets. This dynamic adaptive method improves data exchange efficiency, saves network resources, enables faster local data exchange, and consequently speeds up the overall topology convergence. In detail, by using logistic curve as the basic control function of the adaptive cycle, rules for accidental events is defined accordingly. Simulation is presented to show the validity of this approach and that it is especially suitable for dynamic network environment.

Abstract:This paper proposes a distributed energy efficient clustering protocol for target surveillance in sensor networks (EECTS). The protocol selects cluster heads according to a hybrid of node's residual energy and distribution of its neighbors. In addition, for the sake of reducing the energy dissipation of the cluster head, a minimum spanning tree with root of the base station is constructed among the cluster heads. Then it sends the gathered data to its upstream node along the spanning tree. The chief task of surveillance sensor networks is to sensing the moving target. So an intra-cluster node schedule method named EECTS-1 that can senses the most part of the network and it's enhanced method EECTS-2 are introduced. The two methods can obtain the high continuous surveillance degree when the moving target enters into the network. EECTS produces a linear network lifetime in the number of nodes and keeps good continuous surveillance degree simultaneously. The simulation results show that with the same performance of surveillance the EECTS-1 achieves the same lifetime as that HEED obtains and outperforms DEEG with up to 35% improvement. The EECTS-2 outperforms HEED significantly with prolonging the network lifetime about 70%~80%. Therefore the EECTS is suitable for military target surveillance and has high reliability of the sensing information.

Abstract:RIPEMD-128 is a cryptographic hash function proposed in 1996 by Hans Dobbertin, Antoon Bosselaers and Bart Preneel. It consists of two different and independent parallel parts, with which the results in each application of the compression function. This paper presents a practical attack for finding collisions for the first 32-step reduced RIPEMD-128 with complexity of 2²⁸ 32-step reduced RIPEMD-128 operations. This is the first published analysis for the first 32-step reduced RIPEMD-128.

Abstract:In this paper, based on the statistical smoothing strategy, a image region feature generative probability estimation method is proposed by exploiting maximum weight matching algorithm. By further analyzing and measuring the semantic correlations between words based on the training set, a novel image annotation algorithm for adopting the generative model is presented. The first annotation keyword is obtained by using the proposed image region feature generative probability estimation algorithm. Then, a heuristic iterate function is proposed to exploit the keyword semantic correlation. Finally, the semantic correlation between the annotation and the image can be improved by our annotation algorithm. The proposed annotation approach is tested on a real-world image database, and promising results are achieved.

Abstract:This paper proposes a strategy for retrieving multi-texture images based on contour and texture segmentation. Firstly, the contour of each texture primitive is extracted from an image and its Fourier descriptor is calculated. Thus, the contours of the texture primitives in the original image are clustered according to these shape descriptors. Then Gabor wavelet transform is applied to extract the features of texture primitives for each group, so the image can be represented by a set of feature vectors in feature space. Finally, an improved and noise insensitive Hausdorff distance is used to calculate the distance between two feature vector sets. Furthermore, the retrieval of multi-texture images can be implemented. A large amount of experiments show that this method has higher retrieval precision, compared with the state-of-arts methods.

Abstract:To support large-scale users to share a virtual environment, multi-server architecture is applied to the virtual environment system and each server handles a partition of the virtual environment. The unpredictable movements and interactions of avatars may cause some servers to be overloaded. Existing load balancing algorithms try to redistribute load among servers and incur too much overhead, which may reduce the interactivity of system. In this paper, a layered iterative dynamic load balancing algorithm is proposed. Setting the overloaded region as the center, the algorithm chooses a limited number of neighboring regions as the load adjusting goals and iteratively spreads the overload outward from the center. The load balancing state can be achieved through iterations. Based on the virtual environment wherein avatars are scattered under skewed and clustered distribution, the algorithm is built and comparison with the existing load balancing algorithms are made. The results show that this algorithm can adjust load efficiently and incur less overhead.