Catálogo de publicaciones - libros

In parallel and distributes systems we have to consider aspects of communication, synchronization and data movement besides the actual algorithmic solution. In order to enable the clear separation of these aspects [29] we have to separate the structure or shape of data structures from the actual data elements themselves. While the latter can be used to describe algorithmic aspects, the former enables us to derive the communication [27].In this paper we formalize the notion of shape and data elements and how they can be separated as well as used to reconstruct a valid data structure again for arbitrary container types, i.e. parameterized, nested algebraic data types and arrays. The shape function removes the data elements from the container type and replaces them by a dummy element * (or alternatively, their identifier). The data function extracts the list of data elements contained in the data structure; while a re_cover function can be used to reconstruct a data structure from its shape and a consistent list of data elements. All these functions can be characterized as special cases of a general traversal function operating on the data structure or its shape. We have used this as the semantical basis for handling overlapping data distributions not only over arrays as in commonly used approaches but over arbitrary (container) types.

Modeling agent negotiation is of key importance in building multi-agent system, because negotiation is one of the most important types of agent interaction. Negotiation provides the basis for managing the expectations of the individual negotiating agents, and it enables selecting solutions that satisfy all the agents as much as possible. Thus far, most negotiation models have serious limitation and weakness when employed in large-scale multi-agent systems. Yet, large-scale multi-agent systems find their use in major domains of human development such as space exploration, military technology, disaster response systems, and health technology. This paper presents an agent negotiation model which extends the capabilities of the model associated with the Agent Negotiation Engine for Collaborative Decision Making, to address the negotiation issues associated with large-scale multi-agents systems. The model utilizes Qualitative Reasoning and Game Theory algorithms to track the negotiation process, and a similarity criteria algorithm to manage the large amount of negotiation information associated with large-scale multi-agent systems. For completeness sake, the paper also presents the negotiation models from which the negotiation model for large-scale multi-agent systems evolved, as well as how and why the modifications were made.

The environment in which a parallel application is executed has high impact on the performance of the application due to interference caused by various factors in the execution environment. A detailed understanding of the sensitivity of the application to the parameters describing the execution environment can be of great help in (a) predicting a suitable target machine model for the application, (b) predicting its performance on the target machine, and (c) any algorithmic bottlenecks. In this paper, we analyze a suite of parallel applications for their sensitivity to local and non-local interference arising due to various factors in a parallel environment. We create a test bed consisting of five different parallel applications taken from different sources and analyze their sensitivity to single node and multi-node perturbations and show that parallel applications can behave very differently under different conditions of interference in the environment in which they are running. The main contributions of this paper are: (a) studying a suite of parallel algorithms for their sensitivity to local and non-local interference, (b) demonstrate that an application can behave differently to different interference levels in the environment, (c) demonstrate that the sensitivity of an application can be quantified as its absorption ratio at a given interference level.

Virtualization is becoming an increasingly popular method to achieve software-based solution for sharing hardware infrastructure in a completely isolated manner. Xen virtual machine monitor (an open source software) is becoming a popular resource to implement virtualization for managing multiple operating systems instances within one physical computing node. Current research has focused on migrating these instances between nodes during runtime. This capability is useful for numerous activities, such as fault management, load balancing, and low-level system maintenance. In this paper, we evaluate the performance of Xen Virtual Machine Monitor for high performance computing (HPC) systems and discuss its suitability for 1) easy management of applications (e.g., automatic load balancing of MPI application processes), and 2) easy management of the HPC architecture (e.g., automatic migration of OS instances away from physical resources that have been predicted to fail in the near future

High Resolution (HR) means pixel density within image is high. Along with pleasing picture, the high resolution image offers additional information that may be vital in analyzing image precisely in applications like, military, medical imaging, consumer electronics and so forth. Super resolution image reconstruction is used to restore a high resolution image from several low resolution images. Super resolution image reconstruction is three stage process: Registration, Interpolation and Restoration. In this paper we suggest a wavelet based interpolation that decomposes image into correlation based subspaces and then interpolate each one of them independently. Finally combine these subspaces back to get the high resolution image. We propose it for super resolution imaging along with results to put forth that it produces best results qualitatively analyzed using subjective quality measure. The concepts related to super resolution imaging; interpolation and wavelet are covered as background theory.

This paper presents in summary the implementation of a Web Content Management System (WCMS) at Intercollege. The system was developed to be used by the Research Office of Intercollege. As a Content Management System (CMS) the system presents a tool that can essentially by used to manage content, which materializes here as research activities and knowledge obtained, in various ways, so that it can be quickly and easily published and accessed. The purpose of this system, just like other CMSs, was to record, analyze, and be able to re-use accumulating information on research activities of Intercollege employees.

In this paper, we present a condensed survey of multi-agent systems, with special emphasis on cooperation coordination, conflict resolution and closely related issues; issues that are critical for the development of large-scale, distributed complex software systems. Then we present three different cooperative MAS architecture types, discuss their drawbacks and propose the need for a service driven framework for the development of cooperative multi-agent systems.

Recently Data Warehouse System is becoming more and more important for decision makers. Most of the queries against a large Data Warehouse are complex and iterative. The ability to answer these queries efficiently is a critical issue in the Data Warehouse environment. If right Index Structures are built on the columns, performance of the queries, especially ad-hoc queries will be greatly enhanced. In this paper, we have concentrated on various implementation issues of Simple Bitmap Indexing and their analysis.

This paper discusses when selfish agents begin to cooperate instead of defect, taking a specific task of maintenance of themselves. The network cleaning problem where a collection of agents capable of repairing other agents by overwriting its content can clean the collection will be discussed. With this problem, cooperate corresponds to repairing other agents and defect to not repair. Although both defect is a Nash equilibrium: no agent is willing to repair others when only the repair cost is involved in the payoff, agents may cooperate with each other when system reliability is also incorporated in the payoff and with certain conditions satisfied. Incentive for cooperation will be stronger when further system wide criterion such as availability is involved in the payoff.

During the execution of a project (investment, innovation etc.), three important parameters must be kept in mind: we have to execute the project as soon as possible, with minimal total cost and not to exceed resource (manpower, materials, engines etc.) availabilities.Why does it important to execute the project as soon as possible with minimal total cost? If more than one company compete for the execution of an investment project, usually the chance of winning the tender will be higher if a company can execute the project with minimal total project time (TPT) and minimal total project cost. This problem could already be handled in the 60s and 70s with network planning (CPM, MPM, PERT etc.), scheduling (Gantt Diagrams, LOB etc.) and other related cost-minimizing (CPM/COST, MPM/COST etc.) techniques. The most difficult problem was to handle the resources. During the execution of a project we must keep in view the resources, because these resources are usually straitened. There are well-defined number of labours, engines and so on.If we would like to execute the project with minimal TPT and minimal total project cost and optimal use of the resources (manpower, materials, engines etc.) the problem becomes easily so hard to solve (already at 5000-10000 activities) that computers available today cannot find the solution within a reasonable time. The real problem is more complicated, because before the execution of the project we can only estimate the duration time, (variable) cost and resource need of activities. In real life it is common that the duration time of project activities cannot be estimated correctly. In this paper a novel algorithm is introduced by which an optimal resource allocation with minimal total cost for any arbitrary project could be determined. Moreover, this algorithm also handles the competences of the human resources.A distributed problem solving environment is also introduced that implements the above mentioned optimal resource allocation algorithm with a parallel branch and bound method. The system is built on the Jini technology . It is a dynamic, service-oriented infrastructure that utilizes spare cycles of networked workstations in an efficient way and solves computation intensive problems more easily due to the parallelization.