The thesis addresses the problem of automatic and dynamic tuning of parallel and distributed applications. Our objective is to help developers in the process of improving the application performance. The work presents a whole solution that deals with the issues of automatic and dynamic application improvement. In this approach, an application is monitored, its performance bottlenecks are detected, solutions are given and the application is modified on the fly
Villagarcía Wanza, Horacio A.
One of the major breakthroughs in VLSI technology has been its ability to implement signal processing algorithms and architectures onto cost-effective ASIC and DSP chips.
Works in this area are highly interdisciplinary; they involve interaction among areas such as VLSI and circuit design, computer architecture, signal processing, communications and computer-aided design.
The editors collect in two volumes much of the research efforts and findings that have made the high performance implementation of signal processing possible in the '90s.
Zhu, Haiping; Li, Peigen; Zhang, Guojun
Traditional workflow models have obvious shortcomings in describing complex workflows. Such complexity is due not only to the hierarchical property of business process, but also to the complicated dependencies among tasks. In this paper, we first introduce a conceptual task model based on UML class graph. Then we apply colored Petri nets to model task interfaces and composite task versions which form Task State Nets (TSNs) and Task Version Nets (TVNs). The main advantage is that they explicitly express all task interfaces´ internal states and thus they can easily describe the complex control dependencies. Next we further combine TSN/TVN with...
Corbalán, Leonardo C.; Lanzarini, Laura Cristina; De Giusti, Armando Eduardo
Evolving neural arrays (ENA) have proved to be capable of learning complex behaviors, i.e., problems whose solution requires strategy learning. For this reason, they present many applications in various areas such as robotics and process control. Unlike conventional methods "based on a single neural network" ENAs are made up of a set of networks organized as an array. Each of them represents a part of the expected solution. This work describes a new method, ALENA, that enhances the solutions obtained by solving the main deficiencies of ENA since it eases the obtaining of specialized components, does not require the explicit...
Cancela, Héctor; Robledo, Franco; Rubino, Gerardo
System survivability is the ability to give service in spite
of failures of some of the components. To assure survivability is an important goal when designing a communications network backbone, to ensure that it can resist to failures in the switch sites as well as in the connection lines. Previous work has employed a Greedy Randomized Adaptive Search Procedure (GRASP), based on path algorithms, to build low cost network topologies which comply with heterogeneous node-connectivity requirements, which can model the survivability goals. In this work, we present another variant of the GRASP procedure, based on a tree search, which obtains good...
Gómez, Sergio Alejandro; Chesñevar, Carlos Iván
Many classification systems rely on clustering techniques in which a collection of training examples is provided as an
input, and a number of clusters c1,...cm modelling some concept C results as an output, such that every cluster ci is labelled as positive or negative. Given a new, unlabelled
instance enew, the above classification is used to determine to which particular cluster ci this new instance belongs. In such a setting clusters can overlap, and a new unlabelled instance can be assigned to more than one cluster with conflicting labels. In the literature, such a case is usually solved non-deterministically by making a...
Reynoso, Luis A.; Genero Bocco, Marcela; Piattini Velthuis, Mario
UML models quality is clearly a crucial issue that must be evaluated at the initial stages of object-oriented software development, in order to get software products with better quality. This fact is corroborated by the huge amount of metrics that have appeared in the literature related to the measurement of UML diagrams at a high level design stage. Most of these metrics are focused on the measurement of internal quality attributes such as structural complexity, coupling, size, etc. However, none of the proposed metrics take into account the added complexity involved when a UML model are complemented by expressions written...
Baraldi, Waldemar A.; Scarpa, Diego; Ponzoni, Ignacio; Ardenghi, Jorge Raúl
Facing the different approaches to process activation and global termination detection on distributed systems, this paper performs a practical comparison between the mainly opposed centralized and distributed models. Firstly, they were implemented to solve both the activation and termination detection problems. Based on processing performance and communication overhead metrics, both systems were evaluated running on various parametrized environments. As a result, we present the values obtained and the final conclusions. The overall design was made following an object oriented methodology, which was coded using C++. The inter-process communication was carried out using the PVM libraries.
Singh, Awadhesh Kumar; Bandyopadhyay, Anup Kumar
A mathematical model for the specification and verification of a data link layer protocol is proposed. The weakest precondition calculus, developed by Dijkstra, originally for sequential programs, has been chosen for this purpose. It is demonstrated that the wp-calculus provides a basis, not only for the modeling but also, for a straightforward and thorough analysis of large and complex distributed systems like data link layer protocol. This analysis contributes to the understanding of the system and could lead to an improvement in the design. The technique has been illustrated by describing the sliding window protocol.
Luo, Jianshu; Tang, Zeying; Lu, Hanqing
In this paper, an improvement of deformable template matching algorithm for polygonal objects in grayscale images using two-dimensional deformable templates along orthogonal curves is presented. In the process of pre-computing extensions of the deformable template along orthogonal curves, the novel matching approach incorporates adapting knowledge-specific template discretization techniques appropriate for different polygonal objects and minimizing the improved internal and external energy terms containing inter-shape information of polygonal objects. In our application, this energy optimization problem of the deformable template is efficiently solved by a genetic algorithm (GA). Our algorithm has been successfully applied on synthetic images and real images. The...
Yu, Shengsheng; Zhou, Jingli; Jiang, Gang
In Bluetooth networks, devices are organized into small piconets and large scatternets, and each node acts as the role of master, slave or gateway. Due to dynamic topology changes, different bandwidth available and unpredictable interference of media in Bluetooth networks, the congestion of data flow will inevitably emerges on the link, and the gateway has to switch between piconets on a time division basis, so its presence in the different piconet has to be controlled by scheduling mechanism such as inter- and intra -piconet scheduling. However, the time division in gateways will limit the network capacity and introduce bottleneck points...
Marin Cahiuan, Juan Mauricio
The Web has became an obiquitous resource for distributed computing making it relevant to investigate new ways of providing efficient access to services available at dedicated sites. Efficiency is an ever-increasing demand which can be only satisfied with the development of parallel algorithms which are efficient in practice. This tutorial paper focuses on the design, analysis and implementation of parallel algorithms and data structures for widely-used text database applications on the Web. In particular we describe parallel algorithms for inverted files and suffix arrays structures that are suitable for implementing search engines. Algorithmic design is effected on top of the...