Dempster9;s rule of evidence combination is computational expensive. this paper presents a parallel approach to evidence combination on a qualitative Markov tree. Binarization algorithm transforms a qualitative Mar...
详细信息
ISBN:
(纸本)0780378407
Dempster's rule of evidence combination is computational expensive. this paper presents a parallel approach to evidence combination on a qualitative Markov tree. Binarization algorithm transforms a qualitative Markov tree into a binary tree based on the computational workload in nodes for an exact implementation of evidence combination. A binary tree is then partitioned into clusters with each cluster being assigned to a processor in a parallel environment. the parallel implementation improves the computational efficiency of evidence combination.
the proceedings contain 179 papers. the special focus in this conference is on Support Tools, Environments, Performance Evaluation, Prediction and Load Balancing. the topics include: A grand challenge for computing re...
ISBN:
(纸本)354040788X
the proceedings contain 179 papers. the special focus in this conference is on Support Tools, Environments, Performance Evaluation, Prediction and Load Balancing. the topics include: A grand challenge for computing research;a java-based grid programming environment;a hardware counters based tool for system monitoring;a portable, extensible, and scalable tool for parallel performance profile analysis;on utilizing experiment data repository for performance analysis of parallelapplications;flexible performance debugging of parallel and distributedapplications;exposing and observing communication behavior of parallel cluster applications;a race detection mechanism embedded in a conceptual model for the debugging of message-passing distributed programs;a framework for rule-based autonomic management of distributed scientific applications;a modular tool architecture for parallel program analysis;symbolic performance prediction of speculative parallel programs;a reconfigurable monitoring system for large-scale network computing;presentation and analysis of grid performance data;distributed application monitoring for clustered SMP architectures;static load-balancing techniques for iterative computations on heterogeneous clusters;impact of job allocation strategies on communication-driven coscheduling in clusters;dynamic load partitioning strategies for managing data of space and time heterogeneity in parallel SAMR applications;performance-based dynamic scheduling of hybrid real-time applications on a cluster of heterogeneous workstations;recursive refinement of lower bounds in the multiprocessor scheduling problem;cooperating coscheduling in a non-dedicated cluster;predicting the best mapping for efficient exploitation of task and data parallelism and scheduling strategies of divisible loads in din networks.
In this paper, a new parallel Turbo encoding and decoding technique is introduced. In this technique, a long information data frame is first divided into sub-blocks which are then encoded with trellis terminating and ...
详细信息
ISBN:
(纸本)0780378407
In this paper, a new parallel Turbo encoding and decoding technique is introduced. In this technique, a long information data frame is first divided into sub-blocks which are then encoded with trellis terminating and decoded by parallel multiple SISO modules. It is shown that, at a slight increase in hardware complexity and a slight loss in the transmission efficiency due to the extra terminating bits appended, the proposed scheme can effectively reduce the decoding delay, and at the same time achieve noticeably better error performance compared withthe regular schemes, especially in high code rate situation.
Flexible intrusion detection and response system (ID&R) needs to maximize security while minimizing cost and making response automatically. (CID)-D-2&R, the Cost-based Intelligent Intrusion Detection and Respo...
详细信息
ISBN:
(纸本)0780378407
Flexible intrusion detection and response system (ID&R) needs to maximize security while minimizing cost and making response automatically. (CID)-D-2&R, the Cost-based Intelligent Intrusion Detection and Response System, is proposed in the paper, which is originally developed as a facility to deal with network-based attacks and to make effective response automatically and intelligently. the networking environment deployed withthe (CID)-D-2&R consists of two major parts: Guard, which runs on the specific Guarded Host (GH), and Spy, which runs in Guarded Network (GN). the components of the (CID)-D-2&R are introduced, which include intrusion detection, attack classification, damage analysis, attack path rebuilding, resources automatically safeguarding, calamity recovery, and security Officer. the several kinds of data flow in (CID)-D-2&R are discussed, too. While (CID)-D-2&R is only a prototype, some experimental results are also presented.
Resource allocation and reallocation are very important issues affecting the performance of gang scheduling. Currently most gang scheduling systems adopt regular allocation policies. For many existing clusters of PCs,...
详细信息
ISBN:
(纸本)0780378407
Resource allocation and reallocation are very important issues affecting the performance of gang scheduling. Currently most gang scheduling systems adopt regular allocation policies. For many existing clusters of PCs, however, processors are interconnected via a fast Ethernet or a crossbar switch network and the location of processors is not a significant factor when communication costs are considered. Adopting regular allocation policies may cause unnecessary problems of fragmentation. In this paper we introduce a job-relocation method to alleviate the fragmentation problem.
Optimal scheduling of tasks of a directed acyclic graph (DAG) onto a set of processors is a strong NP-hard problem. Because list scheduling has shown good performance and is less difficult to design, it has been studi...
详细信息
ISBN:
(纸本)0780378407
Optimal scheduling of tasks of a directed acyclic graph (DAG) onto a set of processors is a strong NP-hard problem. Because list scheduling has shown good performance and is less difficult to design, it has been studied and used widely. In this paper we present a new list scheduling scheme to schedule tasks of a DAG onto a homogeneous multi-processors system. the primary objective of this scheme is to minimize schedule length and scheduling time itself. We analyzed three typical list scheduling algorithms-MCP algorithm, ETF algorithm and BDCP algorithm, and find that they all can not guarantee the earliest schedules of the CPNs. In this paper, we propose a better list scheduling algorithm based on critical path, and its time-complexity is O(pv(2)). this algorithm always it is the most important purpose that schedule CPN as soon as it is a ready task, which makes the nodes that have the greatest influence to the scheduling length of the task graph be scheduled first. this greatly shortens the scheduling length of the task graph. We compare the performance of this algorithm with existing scheduling scheme through analysis and the result of experiments. the scheduling length generated by this algorithm is shorter than that of other scheduling algorithm.
Many applications in different fields need to access large amounts of data. the difficulty of handling out-of-core data limits the performance of supercomputers. this paper first describes an out-of-core computation m...
详细信息
ISBN:
(纸本)0780378407
Many applications in different fields need to access large amounts of data. the difficulty of handling out-of-core data limits the performance of supercomputers. this paper first describes an out-of-core computation model on cluster of workstations. then an object-oriented method is offered, which makes the out-of-core programs easier to write. Some optimal strategies, such as distribution of Local Array Files, data sieving and prefetching, also make an improvement in the performance of out-of-core parallel computations.
In this paper, we discuss performance of partition algorithms for parallel solution of large-scale nonlinear systems of equations. We describe first a block Broyden algorithm for solving a nonlinear system in which a ...
详细信息
ISBN:
(纸本)0780378407
In this paper, we discuss performance of partition algorithms for parallel solution of large-scale nonlinear systems of equations. We describe first a block Broyden algorithm for solving a nonlinear system in which a diagonal matrix is used as an iterative matrix. then, we analyze the parallelism of the algorithm and discuss in details different partitioning schemes. Finally, we give some numerical results and analyze performance of the partitioning schemes. the numerical results show that the algorithms combining block Broyden method with partitioning techniques are effective, and that they can be used in the large-scale problems arising from scientific and engineering computing.
In this work, a new methodology for computing unsteady flow problems is presented. this method treats the temporal direction identically to the spatial directions, transforming a 2- or 3- dimensional time-marching pro...
详细信息
Airborne SAR remote sensing image has the characteristic of large data volume and computation burden, so the processing needs very large computer memory and stronger computation ability. Based on the introduction of t...
详细信息
ISBN:
(纸本)0780378407
Airborne SAR remote sensing image has the characteristic of large data volume and computation burden, so the processing needs very large computer memory and stronger computation ability. Based on the introduction of the SAR image processing procedure, this paper studies the SAR image processing using computer parallel computation technology. the parallel processing mechanism is based on the parallel computer cluster operation and the large virtual shared memory technology. In the Client/Server architecture based SAR image parallel system, agent based network communication plays an important role in the computer performance monitor and burden distribution. In the end, the application of the SAR image parallel processing system in the disaster monitor and assessment system is introduced. the application result illustrates the high efficiency of the system and the feasibility of our research.
暂无评论