A Granularity-Aware Parallel Aggregation Method for Data Streams

0 IntroductionDtartaaf fsitcr ecaonmtsroiln , mwaenayth aeprpfloicraectiaosnts e ,tcs u,ccho maes fwroebm sdeirsvtircieb-,uted data sources at different geographiclocations .Distributedprocessing is the inevitable development trend for managingdata streams . Shared-nothing clusters can scale up to thou-sands of computers ,increase the available main memory,pro-cessors ,diskspace and bandwidthalongthe way,andtherebyprovide potential for high throughput and lowlatencies . Yetto date(As yet) ,th…     

A Distributed Data Mining System Based on Multi-agent Technology

Distributed Data Mining is expected to discover preciously unknown, implicit and valuable information from massive data set inherently distributed over a network. In recent years several approaches to distributed data mining have been developed, but only a few of them make use of intelligent agents. This paper provides the reason for applying Multi-Agent Technology in Distributed Data Mining and presents a Distributed Data Mining System based on Multi-Agent Technology that deals with heterogeneity in such environment. Based on the advantages of both the CS model and agent-based model, the system is being able to address the specific concern of increasing scalability and enhancing performance.     

Architecture Design of Global Distributed Storage System for Data Grid

This paper describes the architecture of global distributed storage system for data grid. It focue on the management and the capability for the maximum users and maximum resources on the Internet, as well as performance and other issues.     

Data Location Mechanism in Agent-Based Autonomic Storage Systems

This paper introduces agent-based methodology to build a distributed autonomic storage system infrastructure, and an effectively negotiation mechanism based on agent is applied for data location. We present Availability-based Data Allocation （ADA） algorithm as a data placement strategy to achieve high efficient utilization of storage resources by employing multiple distributed storage resources. We use Bloom filter in each storage device to track the location of data. We present the data lookup strategy that small size of read request is handled directly, and large size of read request is handled by cooperation with storage devices.The performance evaluation shows that the data location mechanism is high available and can work well for heterogeneous autonomic storage systems.     

WA-IDS： A Distributed Intrusion Detection System Based on Data Mining

Aiming at the shortcomings in intrusion detection systems (IDSs) used in commercial and research fields, we propose the MAIDS system, a distributed intrusion detection system based on data mining. In this model, misuse intrusion detection system (MIDS) and anomaly intrusion detcction system (AIDS) are combined. Data mining is applicd to raise detection performance, and distributed mechanism is employed to increase the scalability and efficiency. Host- and network based mining algorithms employ an improved Bayesian decision theorem that suits for real security environment to minimize the risks incurred by false decisions. We describe the overall architeeture of thc MA-IDS system, and discusss pecific design and implementation issue.     

An Efficient Real-Time Fault-Tolerant Scheduling Algorithm Based on Multiprocessor Systems

In the context of real-time fault-tolerant scheduling in multiprocessor systems, Primary-backup scheme plays an important role. A backup copy is always preferred to be executed as passive backup copy whenever possible because it can take the advantages of backup copy de-allocation technique and overloading technique to improve schedulability. In this paper, we propose a novel efficient fault-tolerant ratemonotonic best-fit algorithm efficient fault-tolerant rate-monotonic best-fit （ERMBF） based on multiprocessors systems to enhance the schedulability. Unlike existing scheduling algorithms that start scheduling tasks with only one processor. ERMBF pre-allocates a certain amount of processors before starting scheduling tasks, which enlarge the searching spaces for tasks. Besides, when a new processor is allocated, we reassign the task copies that have already been assigned to the existing processors in order to find a superior tasks assignment configuration. These two strategies are all aiming at making as many backup copies as possible to be executed as passive status. As a result, ERMBF can use fewer processors to schedule a set of tasks without losing real-time and fault-tolerant capabilities of the system. Simulation results reveal that ERMBF significantly improves the schedulability over existing, comparable algorithms in literature.     

An Efficient Query Rewriting Approach for Web Cached Data Management

0 IntroductionWitmhentth ef oerxp slhosairvineg gr ionwfotrhm oaft itohne iWsW aW k,eythe p rroeqblueirme-.When we deal withthe unprecedented amount of data,we areeasylost in browsing them. Thus , handling these data effi-ciently becomes more difficult than before.Consider the application domain publication retrieval ,which are abundant onthe Web and usually demandlongti meto generate the end-results through the query interfaces pro-vided by search engines .It oftenrequires three steps to de…     

Quasi Serializable Concurrency Control in Distributed Real—Time Database Systems

This paper formally defines and analyses the new notion of correctness called quasi serializ-ability,and then outlines corresponding concurrency control protocol QDHP for distributed realtime databases.Finally,through a series of simulation studies,it shows that using the new concurrency control protocol the performance of distributed real-time databases can be much improved.     

考虑系统特征的异构计算负载平衡

异构系统上调度的目标是平衡计算负载，以获得最短的并行计算时间，提出了一个最大异构调度算法，充分考虑了系统对于程序的处理特征及系统的结构和状态，试图在合理的前提下充分利用所有的处理器资源。实验中，在不同的负荷状态下测定了各原语对应于各种调度方案的执行时间。结果表明，该算法在缩短程序响应时间方面取得了很好的效果。     

基于MPI笛卡尔拓扑结构的切片性能分析

通过对MPI并行进程的拓扑结构及其通讯特点进行分析,提出了网络拓扑结构切片的思想,并实现了在MPI中对一般网络拓扑结构进行任意切片的MPI库函数。通过对MPI笛卡尔拓扑结构进行切片实验性能分析,得出构造切片通讯子进行通讯的方法要优于使用原始通讯子的方法。     

基于消息传递的网络并行计算环境

在分布式计算机系统中,由于处理机间无共享内存,因此采用消息传递的方式实现处理机间的数据交换。文章介绍了消息传递接口标准和几种典型的并行计算环境,讨论了基于消息传递的网络并行编程环境的特点,并对其开发进行了初步探讨。     

并行分布式图形生成系统

文章介绍了基于计算机网络的分布式并行图形生成系统，该系统符合ＯｐｅｎＧＬ规范．系统提出了利用非均匀分解实现负载平衡的分解算法，设计了具有自适应能力的体系结构，其研究结果可直接在微机和普通工作站组成的计算机网络上应用，也可用于开发高性能三维图形系统上．     

PDD算法在对称多处理器高性能计算机上的并行实现

使用基于MPI并行编程方法，对PDD算法进行了并行处理及fortran编程，给出了并行代码在上海大学计算机学院自强2000集群式高性能计算机上的运行结果，以及与串行程序相比较的并行加速比。     

MPI并行编程环境及程序设计

通过对MPI原理和特点的研究,给出了并行MPI程序的基本设计思路和执行过程,并实现了向量相加的并行计算.     

G-MPI:一种基于网格的MPI框架及其实现

针对异构网格的复杂环境,提出了一种基于网格平台的新型消息传递接口(MPI)程序设计和运行框架Grid MPI(G-MPI),实现了网格中集群内部节点和集群外部节点的有效通信,从而提供了一个更广意义上的虚拟集群环境.G-MPI作为一个网格中间件,提供给网格平台上有并行任务需求的用户使用.G-MPI提供了C/C 编程接口,使得在集群环境下用这两种语言开发的并行程序,不必经过改动,就可以移植到网格环境下运行.网格环境下并行程序的开发,也变得相对容易.G-MPI加强对并行任务运行状态监控的同时,针对某些特殊的并行任务,给出了新框架下任务异常时的二次调度方案.应用实例验证了新框架的有效性.     

高效率FDTD网络并行计算研究

在一套Beowulf型网络并行计算机系统上,对FDTD并行化处理中的区域分割、邻近子区域之间的边界数据交换、数组操作和进程同步等关键步骤进行了研究和优化,提出了采用MPI并行函数的FDTD并行计算方案.通过三维FDTD并行计算举例,验证了并行计算的正确性,通过测试得到了较高的并行效率.     

分布式计算系统中的动态调度策略研究

在分析已有调度策略的基础上,针对异构分布式计算系统提出了具有良好可扩展性的全局动态分布调度策略.即,利用处理系数比较好地反映了计算系统的异构性,同时调度在各个处理机上分布执行,充分利用了系统的全局负载信息,以较小的调度开销获得较高的调度效率.     

网络并行计算在轨道结构动力分析中的应用

针对车辆-轨道结构动力分析问题，提出了一种基于网络并行计算环境的并行求解方法．考虑车辆和轨道结构特征，建立车辆-轨道结构动力分析模型，推求动力响应微分方程．针对推导的微分方程，改进串行求解算法，提出一种宏流水并行算法．针对算法的性质分析了该算法的适用范围，并从理论上分析了最佳节点机个数和最大加速比．考虑车辆通过轨道不平顺情况，实现了求解轨道结构动力响应的宏流水并行程序．分别在工作站和个人计算机组成的网络并行环境中，测试了提出的宏流水算法性能．实验结果表明，宏流水并行算法可以获得良好的加速比．