WSC 2008

WSC 2008 Final Abstracts

Poster Session Track

Monday 5:10:00 PM 6:10:00 PM
General Posters

Chair: Durk-Jouke van der Zee (University of Groningen)

A non-homogeneous approach to simulating the spread of disease in a pandemic outbreak
Theo Wibisono and Dionne Aleman (University of Toronto) and Brian Schwartz (Emergency Management Unit, Ontario Ministry of Health and Long Term Care)

In the event of a pandemic outbreak, emergency management units must coordinate an effective mitigation strategy to stop the disease spread using limited resources. In order to develop a successful response, it is necessary to have an accurate model of how the disease will spread. Previously presented models largely rely on homogeneous mixing models, which treat every member of the population as having identical infection risk. Intuitively, such an assumption is unrealistic. Certain demographic groups (e.g., healthcare workers, children and the elderly), have higher infection risks. Additionally, behavioral patterns such as use of public transportation impact infection risks. Using contact networks to represent the level of contact between population members and census data to approximate geographic location and travel patterns, we simulate the progression of a droplet-spread disease through the Greater Toronto Area. The results are periodically displayed on area maps using GIS software for visualization and planning purposes.

Modeling of Supply Chain with Variation of Inventory Systems at Nodes
Fernando Rafael Gonzalez Solano, Diana Isabel Davila Ramirez, Marlloly Liseth Sumoza Suarez, and Luis Eduardo Ramirez Polo (Universidad del Norte)

The proposed model is an attempt to analyze the advantages of uses of different systems of inventory in a supply chain involving nodes with different intentions in each one. This model is based on to network of eight nodes distributed between plants, centers of consolidation, distributors and markets with the purpose of simulating the real behavior of to chain of this type.

Verifying the Design of a Cellular Manufacturing System
Benny Tjahjono and Rossella Stama (Cranfield University)

This paper reports a simulation study to evaluate the design of a cellular layout for the production of low volume, high variety of products. The study aims to better understand the production capacity and the potential problems that may arise in the future when more products variety are introduced at smaller quantities. The model was built using a commercial tool but the simulation parameters are stored in a spreadsheet allowing much simpler coding of the routing logic and the data input. The experiments were carried out to reduce bottleneck and to investigate the effect of different batch sizes, machine breakdown and scrap, to the performance of the cellular layout.

SDL Distributed Simulator
Pau Fonseca i Casas (Universitat Politècnica de Catalunya)

Formalisms becomes an important tools since allows the complete understanding of the model and helps in its implementation. However only few simulation tools allows an automatic construction of a simulation model based in a formalization of the system. SDL is a modern object oriented formalism that allows the definition of distributed systems. It has focused on the modeling of reactive, state/event driven systems, and has been standardized by the International Telecommunications Union (ITU) in the Z.100 Since it is a graphical formalism simplifies the understanding of the model. In this paper we show an implementation of a simulation infrastructure that follows SDL formalization language. This infrastructure allows a distributed simulation of the models without any modification to the model definition. Since this infrastructure follows the SDL language formalism it is useful not only for a production use but to teach formalisms and distributed simulation concepts.

An XML-Based Language for DEVS Components
Nicolas Günter Meseth, Patrick Kirchhof, and Thomas Witte (University of Osnabrück)

An XML-based language for simulation components (XLSC) is presented in this paper. The language is designed in a way that the resulting components comply with the DEVS formalism. The objective is to model the static structure of a component as well as its behavior (dynamic structure). For execution of XLSC, an interpreter is prototypically implemented in Java. The use of XLSC enables components to be exchangeable and to be used with any DEVS simulator regardless of its implementation language. Thereby, the interpreter acts as an interface between the model and the simulator as it can directly execute the component’s functions.

A Comparison of Sequential Design Methods for RF Circuit Block Modeling
Karel Crombecq (University of Antwerp), Dirk Gorissen (Ghent University), Luciano De Tommasi (University of Antwerp) and Tom Dhaene (Ghent University)

When modeling complex systems, the locations of the data points are essential to the success of the algorithm. Sequential design methods are iterative algorithms that use data acquired from previous iterations to guide future sample selection. They are often used to improve an initial design such as a Latin hypercube or a simple grid, in order to focus on highly dynamic parts of the design space. In this paper, a comparison is made between different sequential design methods on a real-world electronics problem. Error-based and density-based methods are compared against a novel hybrid technique which incorporates both an error-based measure, using gradient estimations of the objective function, and a density-based measure, using a Voronoi tessellation approximation. The test results indicate that a considerable improvement of the average model accuracy can be achieved by using this new approach.

Panoramic Screen-Based Simulation with Dynamic Background
Samsun Lampotang and David E. Lizdas (University of Florida), John J. Tumino (Army National Guard) and Nikolaus Gravenstein and Harshdeep S. Wilkhu (University of Florida)

A national focus group evaluating a preliminary screen-based simulation found navigation between multiple screens representing different operating room (OR) entities challenging. Consequently, we re-designed the simulation without using multiple screens. To address the challenge of aesthetically depicting both extensive scope and minute detail of an OR on a single screen without using a patchwork collage of graphics, we developed a novel approach using a panoramic photograph captured from an anesthesia provider‘s typical vantage point. We used Director to render the panoramic background interactive and dynamic. In spite of extensive computation to allow smooth 360o panning around the OR, image de-warping, animation and sonification of the panoramic background and projecting mathematical models up to 3 hours forward in time, the simulation runs in real time with imperceptible delay on a regular Windows notebook computer. Users can jump forward and backward in time. Clinicians are currently using the simulation for training.

An Agent-Based Simulation Study of the Dynamics of Mobile Viral Advertising
Jiang Wu, Bin Hu, and Shengping Dong (Huazhong University of Science and Technology)

In the mobile era, mobile advertising is essential and has been developing very fast. Marketers are eager to turn to mobile viral advertising for benefiting from initially targeting customers. In this paper, we propose a computational model to reconstruct the spreading of advertisements in social networks. Using this computational model as a test-bed and running a series of virtual experiments, we acquire observations and implications about how to choose an initial set of people to maximize the performance of spreading advertisements. Also, we observe and analyze the impacts of the network structures including topology, size and density and the initial selected number of targeted people on the dynamics of mobile viral advertising. The virtual experiments also help us to examine the suitable policies for combining viral adverting with mass marketing in mobile commerce. In addition, we run virtual experiments (simulations) in a real mobile-online social network to validate the model and to provide an example for practitioners to apply this computational model. In practice, we use the attributes of people and the interest groups they are members of in social networks to infer the spreading probability between people.

Optimal Service Channel Reconfiguration Based on Multi-Agent Simulation
Jin Yan Shao, Ming Xie, Li Xia, Wen Jun Yin, and Jin Dong (IBM China Research Lab)

Solving the problem of long queues becomes increasingly urgent in many on-site service outlets. For example, in a bank branch or a government agency, customers may have to wait for a very long time to be served. How to optimally reconfigure the capability of service channels in such outlets is a key for the service providers to improve efficiency and quality of service. In this paper, we propose a method to optimally configure service channels at a given service outlet. We integrate customer experience metrics with cost and profit into a unified objective function for optimization. Multi-agent simulation is employed to model the stochastic service processes and customer behavior, and to evaluate the objective function in optimizing service channel capacity. Some real-life data collected from bank branches provide significant empirical support to the method and demonstrate that the presented method is both effective and efficient.

Better Confidence Intervals for Importance Sampling
Halis Sak and Josef Leydold (Vienna University of Economics and Business Administration)

It is well known that for highly skewed distributions the standard method of using the t statistic for the confidence interval for the mean does not give robust results. This is an important problem for IS as its final distribution is often skewed due to a heavy tailed weight distribution. On the poster, we first explain the Hall's transformation to correct the confidence interval of the mean and then evaluate the performance of this method for two numerical examples from finance, which have closed form solutions. Finally, we assess the performance of this method for the credit risk examples. Our numerical results suggest that Hall's transformation can be safely used in correcting the confidence intervals of financial simulations.

Decision-Analytic Models for Breast Cancer: Do Currently Published Models Meet the Requirements of Personalized Medicine?
Beate Jahn, Nikolai Muehlberger, Johannes Wurm, and Uwe Siebert (UMIT - Private University for Health Sciences, Medical Informatics and Technology)

Objective: To give an insight into the structure and methodology of simulation approaches used for evaluating interventions in breast-cancer. Research increasingly focuses on personalized strategies in medicine. Therefore, conventional clinical studies, genomic and proteomic (–omics) and biomarker studies need to be translated into patient-tailored prediction rules and treatment decisions. Recommendations for future comprehensive breast-cancer simulation are derived based on published simulations and requirements arising from personalized medicine, Methods: A systematic literature review is performed and information on study design, simulation framework etc. are extracted. Strengths and limitations are assesed. Results: The review shows how currently available models encompass the underlying biologic disease progression, treatment effects including complications and economic outcomes. We expect that these models are limited with respect to information from –omics and biomarker studies. Discrete-event-simulation and agent-based simulation approaches are discussed. Conclusions: This work provides an essential basis for further simulation approaches to evaluate personalized breast-cancer treatment.