Pengwei Tian

Performance Modeling and Analysis of the Booting Process in a Transparent Computing Environment

In a transparent computing environment, clients load their Operating Systems (OSes) from a remote central server, thus the boot time of clients is a significant performance metric for customers. With the rapid evolution of systems based on transparent computing, an in-depth study on the performance of the booting process in it is urgently required. This paper studies the boot performance of clients in a transparent computing environment for a workload associated with the booting process of clients. To address the performance issues, we propose and solve a closed queuing network model, which is employed to evaluate and predict the boot performance of the system with various hardware configurations, using the obtained workload characteristics. The results from the model are consonant with the actual results measured in our experiments. And then we employ the model to predict and analyze the boot performance.

A novel component retrieval method based on weighted facet tree

Recently, the computing paradigm has switched from mainframe and desktop computing to ubiquitous computing; one of whose motivations is to provide intelligent, personalized and comprehensive services to users. As a solution of ubiquitous computing, Active Services has proposed to generate such services on demand by retrieving, adapting, and composing of existing service components. As the popularity of this paradigm, and hence the number of service components increases, how to efficiently retrieve components to maximally meet user requirements has become a fundamental and significant problem. However, current service components are stored dispersedly on the Internet, nor classified in a unified specification, so that traditional component retrieval methods often fail to search and rank components among several different repositories. To solve these problems, this paper proposes a novel component retrieval method based on a weighted facet tree. We first formally specify the components and user queries under facet classification space in the form of weighted trees, then design a component retrieval algorithm by matching between the trees of query and component, in which the sum of the weights of all the leaf nodes in Query Tree are calculated, and finally rank the matched components by matching degree. The experimental results show promising results in terms of retrieval accuracy and efficiency.

A Probabilistic Semantic Based Mixture Collaborative Filtering

Personalized recommendation techniques play more and more important roles for the explosively increasing of information nowadays. As a most popular recommendation approach, collaborative filtering (CF) obtains great success in practice. To overcome the inherent problems of CF, such as sparsity and scalability, we proposed a semantic based mixture CF in this paper. Our approach decomposes the original vector into semantic component and residual component, and then combines them together to implement recommendation. The semantic component can be extracted by topic model analysis and the residual component can be approximated by top values selected from the original vector respectively. Compared to the traditional CF, the proposed mixture approach has introduced semantic information and reduced dimensions without serious information missing owe to the complement of residual error. Experimental evaluation demonstrates that our approach can indeed provide better recommendations in both accuracy and efficiency.

BASCA: a business area-oriented service component adaptation approach suitable for ubiquitous environment

Service customization always plays an important role in ubiquitous environment, in which the adaptation of service component according to user requirement is a key problem. Component adaptation is a traditional topic in Component-Based Software Development (CBSD) and Service-Oriented Architecture (SOA) research fields. But most of the existing approaches employ high level abstraction to achieve the adaptation process and usually make it a computation-intensive task, which can hardly be utilized in the ubiquitous environment. In addition, the high level abstraction-based approach is always difficult to understand for normal users. In this paper, based on the analysis of existing approaches, a business area-oriented service component adaptation approach is proposed. The approach employs basic functions of the business area to represent user requirement and service components. Three control structures are introduced to model the interconnection of the functions. Then the adaptation is performed based on the functionality relationships between service components and user requirement. The approach usually causes low computation cost and is easy to be utilized in ubiquitous environment.

A key-index based distributed mechanism for component registration

Component registration and organization is the base of component discovery and reuse. With the development of service-oriented architecture (SOA) and component-based software engineering (CBSE), the number of various heterogeneous service components increases rapidly, bringing challenges on component registration and organization, especially the problems of overload both on network and registries for traditional centralized and UDDI registration mechanism. In this paper, a Key-Index (KI) based distributed registration mechanism is introduced. It focuses on reducing the unnecessary but shared information of registered components, by defining a key index tree which extracts key features from original component description, and maintaining the key index trees on all the distributed registry servers, keeping them consistent after each successful registration, without replication of the registered component itself. Experimental evaluation on average registration time, average response time and total traffic cost gives us positive support.

A Unified-Index Based Distributed Specification for Heterogeneous Components Management

With the development of service-oriented architecture (SOA) and component-based software engineering (CBSE), the number of various heterogeneous service components increases rapidly. They are stored in traditional repositories of different organization and can be independently accessed by users. However, it brings difficulty in component discovery and reuse. The utility of massive component resource can not be remarkably improved. In this paper, a unified-index (UI) based distributed specification for heterogeneous components management is introduced. It focuses on creating and maintaining a Unified-Index tree on the UI repository server, which has a management field grouped by some traditional repositories. Each UI repository server maps the components of its children traditional repositories into UI format and stores them in local UI repository. It also actively monitors the adding of new components and removing of invalid ones. Every query of component discovery will be diffused to all of the UI repository servers with transmitting of the root server. So in this desired architecture, users can consistently search and access all the heterogeneous components for further assembly and application, satisfying various personalized and comprehensive user requirements.

A pattern decomposition and interaction design approach

This paper explores and discusses the application of pattern decomposition and interaction design approach in pattern layout design. First we introduce a Pattern Decomposition Representation Model (PDM). In this model, the reusable parts of a pattern are extracted as pattern primitives. At the meantime, a module separated from pattern primitives is defined by the abstract structure of the pattern. Next, the interaction design approach based on the design context and the knowledge-based promotion is proposed and the implementation is presented at the end.

Active Services Model and its Supporting Platform in Grid Environment

Grid is a new paradigm of Internet computing and is developing rapidly. Web service technologies have improved the interoperability of grid protocols and services, as well as providing a unified functional extension mechanism. But current research has not definitely solved the problem of system construction based on Web services in grid environment, especially on the method of dynamically assembling existing resource to realize user-oriented services. This paper introduces an active services model and its supporting platform in grid computing environment, which focuses on relieving the disorder condition in service creating, publishing and deploying and providing a new method to provide user-oriented active services in grid computing environment. Active services are proposed to actively collect, organize, classify and manage the published services to create services resource view for users on application level, and dynamically assemble and execute those services to meet the user demands. The paper also presents the implementation of compositive supporting and management platform (COMP), with an applied case of customized multimedia player generated on it.