Catálogo de publicaciones - libros

The deployment of J2EE systems in Grid environments remains a difficult task: the architecture of these applications are complex and the target environment is heterogeneous, open and dynamic. In this paper, we show how the component-based approach simplifies the design, the deployment and the reconfiguration of a J2EE system. We propose an extended architecture description language that allows specifying the deployment of enterprise systems in enterprise Grids, driven by resources and location constraints. With respect to these constraints we present a deployment process that instantiates propagatively the application, taking into account resources and hosts availability. Finally, we present an autonomic solution for recovery from failures.

The peer-to-peer (P2P) systems have grown significantly over the last few years due to their high potential of sharing various resources. Analyzing the workload of P2P system, however, is very challenging as it involves with the cooperation of many peers. Researches have shown that P2P systems become very effective when dividing the peers into two layers, SP (Super-Peer) and OP (Ordinary-Peer). In this configuration, SP based P2P systems have to deal with a large volume of queries from OPs. Therefore, it is important for SPs to keep their workload stable to provide quality service to the OPs. In this study, we present a collaboration strategy for workload balancing based on SP’s workload characteristics and status. Through the SP’s load balancing mechanism, the message response time is decreased and the workload of P2P system becomes more stable.

Imbalanced workload-distribution can significantly degrade performance of grid computing environments. In the past, the theory of divisible load has been widely investigated in static heterogeneous systems. However, it has not been widely applied to grid environments, which are characterized by heterogeneous resources and dynamic environments. In this paper, we propose a performance-based approach to workload distribution for master-slave types of applications on grids. Furthermore, applications with irregular workloads are addressed. We implemented three kinds of applications and conducted experimentations on our grid test-beds. Experimental results show that this approach performs more efficiently than conventional schemes. Consequently, we claim that dynamic workload distribution can benefit applications on grid environments.

A framework for managing pervasive computing is presented. It enables end-users to easily and naturally build visual interfaces for monitoring and customizing context-aware services. It is built on an exiting a symbolic location model to represent the containment relationships between physical entities, computing devices, and places. It supports a compound document framework for visualizing and customizing the model. It provides physical entities, places, computing devices, and services in smart spaces with visual components to annotate and control them and to dynamically assemble visual components into a visual interface for managing the spaces. It can visualize and configure the spatial structure of physical entities and places and the status and attributes of computing devices and services, e.g., the location in which context-aware services are available. By using the framework, end-users can monitor and customize pervasive computing environments by viewing and editing documents.

is a software system that makes it easy for scientists and industrials to deploy and execute their parallel and distributed applications on a public-resource computing infrastructure. The objective of is to develop a real High Performance Peer-To-Peer platform with a distributed scheduling and communication system. The main idea is to build a completely symmetric model where nodes can be providers and consumers at the same time.

This paper describes the different “components” of an infrastructure and the new features proposed by this platform compared to other similar Global Computing projects. It also describes the porting, the deployment and the execution of a phylogenetic CPU time consuming application on an experimental platform.

One of the most important collective communication patterns used in scientific applications is the , also called . Although efficient algorithms have been studied for specific networks, general solutions like those available in well-known MPI distributions (e.g. the MPI_Alltoall operation) are strongly influenced by the congestion of network resources. In this paper we present an integrated approach to model the performance of the All-to-All collective operation, which consists in identifying a contention signature that characterizes a given network environment, using it to augment a contention-free communication model. This approach, assessed by experimental results, allows an accurate prediction of the performance of the All-to-All operation over different network architectures with a small overhead.

Clustering allows hierarchical structures to be built on the nodes and enables more efficient use of scarce resources, such as frequency spectrum, bandwidth, and energy in wireless sensor networks (WSNs). This paper proposes an energy efficient clustering algorithm for self-organizing and self-managing high-density large-scale WSNs, called SNOW . It introduces region node selection as well as cluster head election based on the residual battery capacity of nodes to reduce the costs of managing sensor nodes and of the communication among them. Each sensor node autonomously selects cluster heads based on a probability that depends on its residual energy level. The role of cluster heads or region nodes is rotated among nodes to achieve load balancing and extend the lifetime of every individual sensor node. To do this, SNOW clusters periodically to select cluster heads that are richer in residual energy level, compared to the other nodes, according to clustering policies from administrators. To prove the performance improvement of SNOW , the ns-2 simulator was used. The results show that it can reduce the energy and bandwidth consumption for clustering and managing WSNs.

In this paper, we propose a novel application on grid, the biometrics grid, to promote the development of both biometrics technology and grid computing. The biometrics grid aims to overcome/resolve some main problems of existing biometric technology using grid computing. The most important service provided by the biometrics grid is an algorithm testbed for biometrics researchers on single biometric or multimodal biometrics. We give a case of two respective biometrics recognition processes in voiceprint and face on grid to show that it is feasible in deploying different biometrics applications on a testbed for performance evaluation.

In a pervasive multimodal multimedia computing system, the user can continue working on a computing task anytime and anywhere using forms of modality that suit his context. Similarly, the media supporting the chosen modality are selected based on their availability and user’s context. In this paper, we present the infrastructure supporting the migration of a visually-impaired user’s task in a pervasive multimodal multimedia computing environment. Using user’s preferences which quantify user’s satisfaction, we derive the user’s task feasible configuration. The heart of this work is the machine learning-derived training to acquire knowledge leading to configuration optimization. Data validation is presented through scenario simulations and design specification. This work is our continuing contribution to advance research on making informatics more accessible to handicapped users.

This paper presents a new Service Discovery Protocol (SDP) suitable for Wireless Sensor Networks (WSN). The restrictions that are imposed by ultra low-cost sensor and actuators devices (basic components of a WSN) are taken into account to reach a minimal footprint solution.

The WSN communication model we use is based on the approach [1] which implements a lightweight middleware for WSN on top of standard object oriented middlewares using a small set of interfaces. The proposed SDP uses also this set, so it supposes the minimal overhead for devices and communication protocols, allowing, at the same time, the deployment of a valuable set of services.