Catálogo de publicaciones - libros

This paper presents the development of the distributed processing based Security Constrained Optimal Power Flow (SCOPF) and its integration to a Distributed Shared Memory Energy Management System (EMS) in order to enable real time power systems security control. The optimization problem is solved by the Interior Points Method and the security constraints are considered by the use of Benders Decomposition techniques. The SCOPF is initially parallelized using MPI and then integrated to the actual DSM based SCADA/EMS system SAGE, thoroughly used in the Brazilian power system including the National System Operation Center (CNOS). Results obtained on both the MPI and DSM platforms are presented for actual large size Brazilian power systems analyzed over a list of about a thousand contingencies. The results obtained demonstrate the high efficiency and applicability of the developed tool at Control Centers for real time security control.

Cluster Computing, Large Scale Simulations in Engineering, Parallel and Distributed Computing.

The leveraging of existing storage space in a cluster is a desirable characteristic of a parallel file system. While undoubtedly an advantage from the point of view of resource management, this possibility may face the administrator with a wide variety of alternatives for configuring the file server, whose optimal layout is not always easy to devise. Given the diversity of parameters such as the number of processors on each node and the capacity and topology of the network, decisions regarding the locality of server components like metadata servers and I/O servers have a direct impact on performance and scalability. In this paper, we explore the capabilities of the dNFSp file system on a large cluster installation, observing how scalable the system behaves in different scenarios and comparing it to a dedicated parallel file system. Our obtained results show that the design of dNFSp allows for a scalable and resource-saving configuration for clusters with a large number of nodes.

Cluster and grid computing, parallel I/O, parallel and distributed computing.

Top-k queries are attractive for users in P2P systems with very large numbers of peers but difficult to support efficiently. In this paper, we propose a fully distributed algorithm for executing Top-k queries in the context of the APPA (Atlas Peer-to-Peer Architecture) data management system. APPA has a network-independent architecture that can be implemented over various P2P networks. Our algorithm requires no global information, does not depend on the existence of certain peers and its bandwidth cost is low. We validated our algorithm through implementation over a 64-node cluster and simulation using the BRITE topology generator and SimJava. Our performance evaluation shows that our algorithm has logarithmic scale up and improves Top-k query response time very well using P2P parallelism in comparison with baseline algorithms.

The task scheduling problem in heterogeneous system is known as NP-hard. Recently, Bajaj and Agrawal proposed an algorithm TANH (Task duplication-based scheduling Algorithm for Network of Heterogeneous systems) with optimality conditions, which are wider than previous optimality conditions. TANH algorithm combines the clustering technique with task duplication. We propose two postprocessing algorithms, HPSA1 (Heterogeneous Posterior Scheduling Algorithm) and HPSA2, to reduce the schedule length for DAGs which don’t satisfy the optimality conditions of TANH algorithm. Our algorithms reduce the schedule length by exchanging task clusters in which its parent tasks reside. We compare with HCNF (Heterogeneous Critical Node First) algorithm by illustrating an example to show how our algorithms operate.

Motivated by the inadequacy of current parallel programming artifacts, the # component model was proposed to meet the new complexity of high performance computing (HPC). It has solid formal foundations, layed on category theory and Petri nets. This paper presents some important design and implementation issues on the implementation of programming frameworks based on the # component model.

This paper presents Anahy, a programming environment for cluster computing. Anahy is presented in terms of its programming interface (API) and its scheduling mechanism. The main features of this environment are the specification of a POSIX thread-based API and the use of dynamic scheduling techniques based on Directed Acyclic Task Graphs (DAG). The main advantage obtained with these features is the dissociation between the description of the concurrency of an application and its parallel execution. The paper examines how Anahy builds a DAG describing the dependencies among tasks at execution time from a multithreaded program and how this DAG is handled by the runtime to apply dynamic scheduling techniques. The paper concludes discussing three case studies of applications developed in the context of Anahy environment.

This work presents DWMiner, an association rules efficient mining tool to process data directly over a relational DBMS data warehouse. DWMiner executes the Apriori algorithm as SQL queries in parallel, using a database PC Cluster middleware developed for SQL query optimization in OLAP applications. DWMiner combines intra- and inter-query parallelism in order to reduce the total time needed to find frequent item sets directly from a data warehouse. DWMiner was tested using the BMS-Web-View1 database from KDD-Cup 2000 and obtained linear and super-linear speedups.

The work described in this paper tackles the problem of data mining and classification of large amounts of data using the K nearest neighbours classifier (KNN) [1]. The large computing demand of this process is solved with a parallel computing implementation specially designed to work in Grid environments of multiprocessor computer farms. The different parallel computing approaches (intra-node, inter-node and inter-organisations) are not sufficient by themselves to face the computing demand of such a big problem. Instead of using parallel techniques separately, we propose to combine the three of them considering the parallelism grain of the different parts of the problem. The main purpose is to complete a 1 month-CPU job in a few hours. The technologies that are being used are the EGEE Grid Computing Infrastructure running the Large Hadron Collider Computing Grid (LCG 2.6) middleware [3], MPI [4] [5] and POSIX [6] threads. Finally, we compare the results obtained with the most popular and used tools to understand the importance of this strategy.

Grid, Parallel Computing, Threads and Data Mining.

This paper presents the use of the Multimedia C (MMC) language to develop multimedia applications. The MMC language was designed to support operations with multimedia extensions included in all modern microprocessors. Although the idea to extend high programming languages to support vector operations is not novel, we show that integration of multimedia extensions into C is valuable. This is specially true for idiomatic expressions which are difficult for a compiler to identify. The MMC language has been used to develop some of the most frequently used multimedia kernels. The presented experiments on these scientific and multimedia applications have yielded good performance improvements. Although this paper discuses the use of MMC, the key features of the MMC language and implementation of its compiler are also presented.

The Advanced Encryption System – AES is now used in almost all network-based applications to ensure security. In this paper, we propose a very efficient pipelined hardware implementation of AES-128. The design is versatile as it allows both encryption and decryption. The core computation of AES, which is performed on data blocks of 128 bits, is iterated for several rounds, depending on the key size. The security strength of AES has been proven proportional to the number of rounds applied. we show that if the required number of rounds must increase to defeat attackers, the proposed implementation stays efficient.