Catálogo de publicaciones - libros

As an important strategy for reducing access time, balancing overload, and obtaining high data availability, replication is widely used in distributed system such as Data Grids. Currently grid researchers often consider replica placement problem from the view of whole systems, i.e, ignoring individual requirements. With the rapid growth of Data Grid applications, fulfilling requirements of every request becomes a big challenge, especially in Data Grids with massive data, multiple users and geographically distribution. We build up a ring-based Data Grid model and define a QoS-guaranteed Replica Management Problem. A QoS-guaranteed replica placement algorithm qGREP is proposed, which takes individual QoS requirement into account and achieves tradeoff between consistency cost and access efficiency. Also a ring-base electing algorithm VOSE is presented to solve update conflicting while minimizing propagating costs. Analysis and experiments show that the model and algorithm are feasible in complex Data Grid environment and achieve good performance under different simulating conditions.

With the development of computer and communication technology, access control of the resources in databases has become an issue focused by both consumers and enterprises. Moreover, the new concept of purpose-based authorization strategies is widely used instead of the traditional one of role-based strategies. The way of acquiring the optimal authorization strategies is an important problem. In this paper, an approach of mining authorization strategies based on purpose in database system is proposed. For obtaining the optimal authorization strategies of the resources in databases for supporting various purposes, an algorithm of clustering purposes is designed, which is based on the inclusion relationship among resources required by the purposes. The resultant purpose hierarchy is used for guiding the initial authorization strategies. The approach provides valuable insights into the authorization strategies of database system and delivers a validation and reinforcement of initial strategies, which is helpful to the database administration. The approach can be used not only in database system, but also in any access control system such as enterprise MIS or web service composing system. Theories and experiments show that this mining approach is more effective and efficient.

Data Grids are today’s emerging infrastructure providing specialized services on handling large datasets that needs to be transferred and replicated among different grid sites. Data replication is an important technique for data availability and fast access. In this paper we present a comparison of various replication models and techniques employed by some major topologies used in data grid environment. We focus on dynamic strategies for replica placement in tree, Peer-to-Peer (P2P) and hybrid architectures. Beside tree model which is being implemented for many Data Grid applications, hybrid and P2P grid models of replication are also emerging for providing scientific communities with better availability and efficient access to massive data.

Although wireless devices are the most suitable device for ubiquitous environment, they have restrictive capacities when using internet services than desktop environments. Therefore this research proposes a wireless internet service method that uses contents-based personalization. The existing websites can easily and promptly access desired news articles and other data through RSS-linked web contents and by the personalization method. The proposed method will make using wireless internet easier while lowering wireless contents production costs. Moreover, personalized mobile web contents that satisfy the preferences of users can be serviced.

In this paper, we propose a system architecture for handling and storing sensor data stream in real-time to support the spatial and/or temporal queries besides continuous queries. We exploit a segment-based method to store the sensor data stream and reduce the managed tuples without any loss of information, which lead to the improvement of the accuracy of query results. In addition, we offer a method to reduce the cost of join operations in processing spatiotemporal queries by filtering out the list of irrelevant sensors from the query range before making the join operation. We then present a design of the system architecture for processing spatial and/or temporal queries. Finally, we implement a climate monitoring application system.

A WFMS(workflow management system) contains two basic elements: the workflow model and the workflow engine. It is important to verify workflow models before they are put to execution. Traditional workflow models mainly describe workflows either from the control perspective or from the data perspective. In fact, the control flow and the data flow are two important aspects for workflow modeling and they are not independent from each other. A new workflow modeling technique, named Dual Workflow Nets (DWF-nets), is proposed to explicitly model the control flow and data flow of workflow processes. Besides, the control/data flow interactions can be captured in DWF-nets. Moreover, the control/data inconsistency, which is neglected by traditional modeling techniques, can be detected by verification of DWF-nets.

Process technology is widely adopted for composition of web services. On-demand composition of web services by a large number of small businesses or even end-users, however, calls for a peer-to-peer approach for process executions. We propose a lightweight process-aware component of middleware for peer-to-peer execution of processes. The approach is of continuation-passing style, where continuations, or the reminder of executions, are passed along with messages for process execution. Conducting the execution of a process is a series of local operations rather than global coordination. Two continuations are associated with an execution: a success continuation and a failure continuation. Recovery plans for processes are automatically generated at runtime and attached to failure continuations.

The purpose of this study was to propose the real time face recognition system using multiple image sequences for network users. The algorithm used in this study aimed to optimize the overall time required for recognition process by reducing transmission delay and image processing by image compression and minification. At the same time, this study proposed a method that can improve recognition performance of the system  by exploring the correlation between image compression and size and recognition capability of the face recognition system. The performance of the system and algorithm proposed in this study were evaluated through testing.