Many simulations and other applications need to interact seamlessly with distributed data *** Agent-Based Environment for Linking simulations (ABLES) allows the formation of a dynamic "data and simulation cloud&q...
ISBN:
(纸本)9780769518534
Many simulations and other applications need to interact seamlessly with distributed data *** Agent-Based Environment for Linking simulations (ABLES) allows the formation of a dynamic "data and simulation cloud" that links a heterogeneous collection of networked *** consists of three major types of components: user entities that serve as data producers and/or consumers, a brokering system for organizing and linking the various participants, and generic local agents that connect simulations and data resources to the cloud of *** paper presents the design of the distributed brokering system, which is implemented using Java and Sun Microsystems' Jini technology.
Running a large-scale distributedsimulation may needa large amount of computing resources at geographicallydifferent locations. These resources may be from differentorganizations. The simulation may run for a long pe...
详细信息
ISBN:
(纸本)9780769518534
Running a large-scale distributedsimulation may needa large amount of computing resources at geographicallydifferent locations. These resources may be from differentorganizations. The simulation may run for a long periodof time and the availability and the amount of computingresources available may change during the course of thesimulation execution. Therefore, coordinating and managingthe resources for distributedsimulation to complete thesimulation efficiently and effectively is a critical issue. Thispaper describes a load management system for HLA-baseddistributedsimulation. The system is constructed on top ofa Grid Computing environment supported by Globus. Theoverall structure of the system is presented in the paper andhow the system saves and restores a federate is also discussedin detail.
This paper extends our previous work on formalizing even orderings using partial order set and its application in space analysis in distributed *** focus on the time and space trade-off in exploiting event parallelism...
ISBN:
(纸本)9780769518534
This paper extends our previous work on formalizing even orderings using partial order set and its application in space analysis in distributed *** focus on the time and space trade-off in exploiting event parallelism. Even parallelism is divided into inherent (problem) parallelism, event ordering parallelism and effective event ***, we analyze the performance cost of varying even ordering parallelism on memory requirement in open and closed ***, we study the effects of interconnection topology of a physical system on exploitable event ordering *** were obtained from a time-space analyzer that we have developed.
This paper introduces a Coloured Petri Nets formalism-TCPN- which is suitable for modelling and simulation ofcomplex systems. Novel in TCPN is (i) a timing modelwhich accommodates both unordered and ordered (queue)pla...
ISBN:
(纸本)9780769518534
This paper introduces a Coloured Petri Nets formalism-TCPN- which is suitable for modelling and simulation ofcomplex systems. Novel in TCPN is (i) a timing modelwhich accommodates both unordered and ordered (queue)places; (ii) the adoption of Java as the net inscriptionlanguage. TCPN is supported by a graphical tool -RAINBOW - which allows model design, testing andsimulation. A large TCPN model can be split into regions(subnets) whose execution is assigned to differentLP/processors of a networked simulator. distributedsimulation relies on TUTW, an agent-based time Warpalgorithm capable of exploiting temporal uncertainty forimproving simulation performance. The paper describesTCPN and its distributed execution kernel. The practicaluse of TCPN is demonstrated through an example andsome experimental results.
Large rapid-prototyping systems comprising several FPGAs become more and more the tool at hand to verify complete hardware systems at an early stage of development for first time success. Although hardware capability ...
详细信息
Large rapid-prototyping systems comprising several FPGAs become more and more the tool at hand to verify complete hardware systems at an early stage of development for first time success. Although hardware capability is growing rapidly the appropriate software tools are lacking in mapping performance and quality. It is especially difficult to meet certain real-time constraints when a design is distributed among several FPGAs. We propose a macro-based partitioning methodology that significantly improves turnaround times and leads to very compact hardware realizations. We demonstrate the benefits of our approach for a real-time video processing application. In addition, compilation time and hardware resources could be reduced by 35% and 45%, respectively.
simulation has always been a valuable tool forexperimentation and validation of models, architecturesand mechanisms in the field of networking. In the case ofthe DiffServ framework, simulation is even more valuable,du...
ISBN:
(纸本)9780769518534
simulation has always been a valuable tool forexperimentation and validation of models, architecturesand mechanisms in the field of networking. In the case ofthe DiffServ framework, simulation is even more valuable,due to the fact that an analytical approach of mechanismsand services is infeasible because of the aggregation andmultiplexing of flows. In this work, we have extended thefunctionality of a widely used simulation environmenttowards the direction of realistic traffic generation and aseries of mechanisms defined by the DiffServ *** modules created are being presented and a casestudy of a simulation scenario that exploits thefunctionality provided by them is described.
In this paper we discuss and compare three differentcausality inconsistency tracking mechanisms in support ofpreemptive rollback in optimistic parallel simulation onmyrinet clusters. These mechanisms exhibit different...
ISBN:
(纸本)9780769518534
In this paper we discuss and compare three differentcausality inconsistency tracking mechanisms in support ofpreemptive rollback in optimistic parallel simulation onmyrinet clusters. These mechanisms exhibit different communication/processing overhead and also different effectivenessin revealing causality inconsistency of the currentlyexecuted simulation event. By the results of an empiricalstudy on a classical simulation benchmark we have foundsome trade-offs between these mechanisms, pointing out indicationsof application contexts for which each mechanismis expected to be well tailored.
This paper introduces a technique that reduces thenumber of state savings and the maximum memoryneeded for the time Warp. We present a technique todetermine if an event is safe or not. If an event executionis safe, no...
详细信息
ISBN:
(纸本)9780769518534
This paper introduces a technique that reduces thenumber of state savings and the maximum memoryneeded for the time Warp. We present a technique todetermine if an event is safe or not. If an event executionis safe, no state saving is carried out. The techniquediscards some saved states even though the time stampsare larger than the GVT. We prove that the technique iscorrect under both aggressive and lazy cancellationscheme. This technique can be implemented withminimal additional overhead. Benchmark results oncircuit simulation show that the mechanism can reducethe number of state savings and maximum memory sizesignificantly. The technique is applicable to hardwaresimulation, network simulation and other systems thathave fixed interconnection.
As the Internet evolves toward the global multiservice network of the future, a key consideration is the support for services with guaranteed quality of service (QoS). The current Internet has a number of barriers to ...
详细信息
As the Internet evolves toward the global multiservice network of the future, a key consideration is the support for services with guaranteed quality of service (QoS). The current Internet has a number of barriers to QoS support for real-time data delivery. The proposed differentiated services (Diffserv) framework is seen as the key technology to achieve this. This paper examines standard transport layer protocols, i.e. UDP and TCP, suited for real-time data delivery over Diffserv-capable networks where bandwidth guarantees are provided. We designed a streaming video application that buffers a portion of data on the receiver before starting the playback and operates in a streaming mode. Streaming media applications currently used in the Internet adopt UDP because of the general belief that TCP's reliable delivery mechanisms, such as congestion control and retransmission, are unsuitable for time-sensitive delivery. However, TCP shows the better simulation results than UDP when packets are lost over Diffserv networks. In our simulation, TCP can adapt its sending rate to the guaranteed bandwidth and the buffer on the receiver allows TCP to retransmit lost packets before the playback. As for TCP, the employment of SACK option that can recover lost packets efficiently contributes to a better performance.
Cycles and knots in directed graphs are problems thatcan be associated with deadlocks in database andcommunication systems. Many algorithms to detect cyclesand knots in directed graphs were proposed. Boukercheand Trop...
ISBN:
(纸本)9780769518534
Cycles and knots in directed graphs are problems thatcan be associated with deadlocks in database andcommunication systems. Many algorithms to detect cyclesand knots in directed graphs were proposed. Boukercheand Tropper have proposed a distributed algorithm thatsolve the problem in a efficient away. Their algorithm hasa message complexity of 2m vs. (at least) 4m for theChandy and Misra algorithm, where m is the number oflinks in the graph, and requires O (n log n) bits ofmemory, where n is the number of nodes. We haveimplemented Boukerche and Tropper's algorithmaccording to the construction of processes of the CSPmodel. Our implementation was done using JCSP, animplementation of CSP for Java, and the results arepresented.
暂无评论