WSC 2005 Final Abstracts

Simulation of Grid Computing Infrastructure: Challenges and Solutions
Sugato Bagchi (IBM T. J. Watson Research Center)

Abstract:
Recent advances in middleware technologies such as grid computing have provided IT architects with the ability to design infrastructures that are more flexible and less dedicated to specific application workloads. However, the capabilities of design and analysis tools that IT architects use have not kept pace. In this paper, we describe our progress in developing an IT infrastructure modeling environment that supports an extensible set of analysis tools. We focus in particular on a discrete-event simulator for analyzing the performance of computational workloads that are running on a grid. The unique modeling requirements and challenges presented by the grid computing infrastructure domain are discussed. Efficient event queue management and other simulation techniques to address these challenges are developed. Finally, we position the role of simulation analysis in the larger context of estimating the business and financial returns from grid computing investments.

Time Management in Distributed Factory Simulation, a Case Study Using HLA
Ke Wang, Sheng Xu, and Leon F. McGinnis (Georgia Institute of Technology)

Abstract:
Distributed simulation is attractive for modeling large scale manufacturing systems. However, synchronization requirements present distinct challenges in designing and implementing a suitable modeling structure. Our experience in building a distributed simulation model for 300mm wafer fab using the High Level Architecture (HLA) shows that time management should be considered at both the design and the implementation stages. This paper proposes an optimistic-conservative synchronization scheme by using model specific information at the implementation stage. Computational results are then presented to show that properly implemented distributed factory simulation can achieve significant reduction in model execution time.

Data Consistency in a Large-Scale Runtime Infrastructure
Buquan Liu, Huaimin Wang, and Yiping Yao (National University of Defense Technology)

Abstract:
In order to support large-scale distributed simulation, we have developed a RTI called StarLink+ with particular architecture which is compliant with IEEE 1516. StarLink+ is composed of a Central RTI server and multiple Local RTI servers. Each Local RTI server manages multiple federates. Data consistency has great influence on RTI's performance and scale. In StarLink+, only a small portion of data must be globally consistent for all Local RTI servers. However, a great amount of data is not consistent for different Local RTI servers. This paper focuses on the research of data consistency about a variety of data in StarLink+. On the one hand, we introduce the fully consistent data such as object name designation and handle assignment; on the other hand, we also study the partly consistent data such as publication and subscription, ownership transfer, and time management.

Initialization of Online Simulation Models
Andre Hanisch and Juri Tolujew (Fraunhofer Institute IFF) and Thomas Schulze (University of Magdeburg)

Abstract:
Online simulation is a relatively new control strategy for short-term decision-making for the control and management of processes in existing systems. In contrast to traditional “non-terminating” simulation, online simulation cannot use a transient phase to tune the models because the simulation models need to run very quickly and also need to deliver results right from the start. In this context, the initialization of such online simulation models represents a special problem. It requires mapping between the system-describing variables in the model and the available data in the real system. This paper examines two different methods of initialization. Special emphasis is placed on explaining the approach of parent model synchronization. Both initialization approaches are transferred to the context of analyzing and forecasting pedestrian flows in a public building. A first prototypical implementation in SLX is also briefly presented.

Addressing Complexity Using Distributed Simulation: A Case Study in Spaceport Modeling
Jaebok Park, Reinaldo Moraga, and Luis Rabelo (University of Central Florida), Jeffrey W. Dawson (University of central Florida) and Mario Marin and Jose Sepulveda (University of Central Florida)

Abstract:
As the size, complexity, and functionality of systems to model and simulate continue to increase, benefits such as interoperability and reusability enabled by distributed discrete-event simulation are of interest, especially for distributed manufacturing and enterprise engineering. The High Level Architecture (HLA), a standard distributed simulation environment, is one technology that enables the interconnection of distributed model components. Many applications in industry are developed by a variety of Commercial Off-The-Shelf (COTS) simulation tools, which require some form of gateway to integrate the models into the HLA component-based simulation. This paper summarizes a study conducted to integrate COTS simulation models using gateway tools and visualization of the system states running as part of other simulation models under HLA. The study focused on the prototype of a virtual engineering environment, called the Virtual Test Bed, designed to analyze operations of current and future space vehicles, spaceports, and ranges as a distributed simulation environment.

Flexible Integration of XML into Modeling and Simulation Systems
Mathias Roehl and Adelinde M. Uhrmacher (University of Rostock)

Abstract:
As the effort towards standardization of formalism representations increases so does the need for verifying whether models do or do not follow a standard. Data binding allows to systematically exploit XML and its associated technologies for modeling and simulation purposes. Based on the schema definition of a formalism, a binding compiler generates model classes that support the user in constructing models according to the formalism. Most constraints can be checked automatically, few require separate efforts by the designer of the simulation system. Although simulators could be build for these declarative model descriptions, they would be hardly efficient. To this end, a separate transformation component is required. In this overall process, both model specifications that are consistent with a formalism definition and models that can be executed efficiently are supported equally.