Symbiotic Simulation Systems: An Extended Definition ... - CiteSeerX

14 downloads 0 Views 355KB Size Report
Symbiotic Simulation Systems: An Extended Definition Motivated by Symbiosis in Biology. Heiko Aydt. Stephen John Turner. Wentong Cai. Malcolm Yoke Hean ...
22nd Workshop on Principles of Advanced and Distributed Simulation

Symbiotic Simulation Systems: An Extended Definition Motivated by Symbiosis in Biology Heiko Aydt

Stephen John Turner Wentong Cai Malcolm Yoke Hean Low School of Computer Engineering, Nanyang Technological University Singapore 639798 {aydt, assjturner, aswtcai, yhlow}@ntu.edu.sg Abstract

writing there are two established definitions. The first considers mutualism, in which the relationship is beneficial for both partners, as the only form of symbiosis. The second considers symbiosis as it was originally intended by Anton de Bary who coined the term symbiosis in 1879 [7]. His definition is wider and considers several subcategories, including mutualism [8, 31]. Because there is no single definition of symbiosis, subcategories other than mutualism are often ignored [35]. This is presumably also the reason why mutualism is the only form of symbiosis which has been considered in the context of symbiotic simulation systems.

Although various forms of symbiosis are known in biology, only mutualism has been considered in the context of symbiotic simulation systems. In this paper, we explain why the original definition of symbiotic simulation systems is narrow and why it is important to consider other forms of symbiosis as well. As a consequence we propose an extended definition of symbiotic simulation systems motivated by symbiosis in biology. By using this extended definition, we identify five different types of symbiotic simulation systems which can be applied in various applications. We describe how single systems can be combined and propose a hybrid symbiotic simulation system in the context of semiconductor manufacturing.

1

In this paper we will use symbiosis as it was originally intended by Anton de Bary and draw analogies between the various forms of symbiosis in biology and symbiotic simulation systems. We also explain why the current definition of symbiotic simulation system is too narrow and why it is necessary to consider other forms of symbiosis as well. This leads to a number of distinct types of symbiotic simulation systems which can be used either independently or in combination with each other. We give an example of how various different types of symbiotic simulation systems can be combined together to give a hybrid symbiotic simulation system in the context of semiconductor manufacturing.

Introduction

The parallel and distributed simulation working group at the Dagstuhl Seminar on Grand Challenges for Modeling and Simulation in 2002 introduced the term symbiotic simulation systems to refer to a new paradigm for discrete event simulation [12]. This paradigm emphasises a close relationship between a simulation system and a physical system which is mutually beneficial for both. The simulation system benefits from real-time measurements of the physical system, continuously provided by sensors. These measurements are used to initialise and drive highly accurate simulations of the physical system. In turn, the physical system benefits from its symbiosis with the simulation system from decisions made in near real-time. Decision making is based on the outcome of what-if experiments which involve the simulation of several scenarios, each representing a different decision alternative. Symbiosis has its origins in biology. However, there is no universally agreed-upon definition and at the time of

1087-4097/08 $25.00 © 2008 IEEE DOI 10.1109/PADS.2008.17

This paper is structured as follows: In Section 2 we give an overview of related work. In Section 3 we give examples and explain why the original definition of symbiotic simulation systems is too narrow. In Section 4 we give an overview of symbiosis in biology before drawing analogies and extending the definition of symbiotic simulation systems. In Section 5 we introduce five distinct types of symbiotic simulation systems based on our extended definition. The various types can be combined to give a hybrid symbiotic simulation system. In Section 6 we propose such a system in the context of semiconductor manufacturing for real-time equipment control. In Section 7 we present our conclusions.

109

2

Related Work

to assume that actively steering the measurement process can also be advantageous for symbiotic simulation systems. A symbiotic simulation system involves a control feedback to the physical system. This feedback is meant to control the physical system rather than to control the measurement process. Similarly, DDDAS does not explicitly mention a control feedback to the physical system but is not in conflict with it either. It is reasonable to assume that DDDAS can be used to control a physical system with the same intention as a symbiotic simulation system. Although very similar, the primary focus of these paradigms is different. However, if a symbiotic simulation system does also steer the measurement process it can be considered as a DDDAS.

A concept which is closely related to symbiotic simulation systems, is that of on-line simulation systems. In [2] an on-line simulation is defined in the context of manufacturing as a “computerized system capable of performing both deterministic and stochastic simulations in real time (or quasi real time) to monitor, control, and schedule parts and resources in a discrete-part manufacturing environment”. In a similar context, on-line simulation is used as part of an online planning and control system which uses on-line simulation to evaluate several control policy scenarios [6]. Multiple scenarios are also used in an on-line simulation system for military networks [32]. A slightly different definition of on-line simulation is given in [15], where an on-line simulation is described in the context of UAV path planning as a simulation that runs in real-time and in parallel with a physical system and does not necessarily include a feedback to the physical system. This definition of on-line simulation should not be confused with real-time simulation, as defined in [13], where advances in simulation time are paced by wallclock time. Although on-line simulation has also been used in various other works [10, 11, 37], none of them has provided a proper definition. In addition, the term “online simulation” can also be found in the context of gaming and education, where it is used in an entirely different context, referring to applications which are available in the world wide web. The term on-line simulation has been used in various ways, usually referring to a simulation which is initialised and driven by real-time sensor data. However, the definition of on-line simulation is still ambiguous. In addition, the term on-line simulation system does not emphasise the close relationship between the simulation system and the physical system. Because of these reasons, the paradigm of symbiotic simulation systems was defined at the Dagstuhl seminar. The term symbiotic simulation systems was chosen because it reflects the close relationship between a simulation system and a physical system. A paradigm which is closely related to symbiotic simulation systems, is that of dynamic data driven applications systems (DDDAS). As described in [30], “DDDAS is a paradigm whereby application/simulations and measurements become a symbiotic feedback control system. DDDAS entails the ability to dynamically incorporate additional data into an executing application, and in reverse, the ability of an application to dynamically steer the measurement process”. DDDAS emphasises the ability of the application or simulation to control and guide the measurement process. Although the definition of symbiotic simulation does not explicitly mention control or guidance of the measurement process, it is not in conflict with it either. It is reasonable

3

Motivation for an Extended Definition of Symbiotic Simulation Systems

The original meaning of symbiotic simulation systems is narrow as it refers to mutualism only. In mutualism, all partners involved in the symbiosis benefit from each other. Similarly, a simulation system benefits from a physical system and vice versa. While the simulation system benefits from real-time measurements, the physical system benefits from control feedback created by the simulation system. In the original definition of symbiotic simulation systems it is implicitly assumed that the control feedback is always beneficial for the physical system. However, this is not necessarily the case as further illustrated by the following two examples. Example I: Let us assume that a utility function is used to determine a rating for a simulated scenario. Each scenario reflects an alternative decision and the rating of a scenario depends on how beneficial a decision is for the physical system. Hence, the most beneficial decision can be determined by identifying the scenario with the highest rating. A mutualistic symbiosis between a simulation system and a physical system can be turned into the opposite by simply using the inverse value of the utility function. As a consequence, the simulation system will make decisions which are either less beneficial or even harmful for the physical system. This can be useful in the context of military applications, for instance, where the aim of the simulation system is to make decisions which actively damage the physical system (e.g., hostile fighting unit). In this example, the simulation system intentionally harms the physical system while it still benefits from real-time measurements. Example II: Decision making in a symbiotic simulation system depends entirely on the analysis of simulated scenarios. Therefore, the quality of decision making depends on the quality (i.e., accuracy) of the simulation. A simulation

110

is performed based on a model which is always only an approximation to the real system. It can therefore be assumed that there is always a certain discrepancy between the simulated behaviour and the actual behaviour of the physical system. A simulation model can gradually become less accurate as the physical system evolves over time, in which case the decision making will be negatively affected. Depending on the application context, it is reasonable to assume that poor decision making can actually harm the physical system to a certain degree. As a consequence, what has been mutualism before has been gradually turned into the opposite. The simulation system still benefits from the measurements (e.g., a simulation is initialised with real-time information about the physical system provided by corresponding sensors), while the physical system is suffering. A simulation system can damage, or at least harm, the physical system. Regardless whether this happens intentionally or unintentionally, such a symbiosis is not mutually beneficial anymore. The original definition of symbiotic simulation systems, which considers mutualism as the only form of symbiosis, is therefore too narrow. Other forms of symbiosis have to be considered as well.

4

tion system to the physical system. If the control feedback is beneficial for the physical system, the symbiotic relationship is of mutualistic nature. For several reasons, the control feedback can be harmful for the physical system. If this is the case, the symbiotic relationship is of parasitic nature. If there is no control feedback at all, the symbiotic relationship is of commensalistic nature, i.e., the physical system neither benefits nor suffers from its relationship with the simulation system. This is in contrast to the original definition of symbiotic simulation which considers control feedback as an obligatory feature. However, even without a control feedback there can be still a close association between the simulation system and the physical system because the simulation system depends on the real-time measurements from the physical system. We define a symbiotic simulation system as a close association between a simulation system and a physical system, which is beneficial to at least one of them. In this symbiosis, the simulation system benefits from measurements which can be used to initialise and drive simulations about the physical system. The simulation system employs an arbitrary number of scenarios which are concerned with the physical system. However, control feedback to the physical system is optional. Thus, conclusions derived from simulation results are used to affect the physical system in some symbiotic simulation systems, but not in all of them. Compared to the original definition, we do not restrict the symbiosis between a simulation system and a physical system to be mutually beneficial. Furthermore, our definition does not require a control feedback to the physical system. Therefore, our extended definition is less restrictive and can be used in various kinds of applications and is not limited to control of a physical system.

Symbiosis in Biology and Symbiotic Simulation Systems

In biology a symbiosis refers to two or more organisms, living in close association with each other. The partners in such a symbiotic relationship are called symbionts and in regard to their outcome, they are highly dependant upon each other. A symbiont may benefit, suffer, or may not be affected at all from its symbiotic relationship with another symbiont. This results in the three subcategories mutualism, parasitism, and commensalism, respectively [8]. Many symbiotic relationships, which are found in biology, are not static and there may be frequent transitions from one type to another [8]. Analogies to this can also be found in the context of symbiotic simulation systems. The second example described in the previous section, for instance, explains how mutualism can gradually turn into parasitism because the quality of decision making becomes poor due to an inaccurate simulation model. In the context of symbiotic simulation we assume that the simulation system always benefits from measurements of the physical system. We further assume that the measurement process itself is not significantly harmful to the physical system. This might not always be the case. For example, in a medical application it could be the case that the measurement process (e.g., x-rays) can be potentially harmful for the physical system (e.g., the patient). Since the simulation system always benefits from its symbiosis with the physical system, the specific form of this symbiosis depends on the control feedback of the simula-

5

Different Types of Symbiotic Simulation Systems

Since control feedback is optional, we can distinguish between closed-loop and open-loop symbiotic simulation systems similarly to closed-loop and open-loop DDDAS [19]. However, in the context of symbiotic simulation we can further distinguish whether feedback in a closed-loop system is either beneficial or harmful. The symbiotic relationship is therefore either mutualistic or parasitic, respectively. In contrast, in an open-loop symbiotic simulation system, the symbiosis between the simulation system and the physical system is always of a commensalistic nature because the physical system is not affected. With closed-loop and open-loop we strictly refer to the relationship between the simulation system and the physical system. Although there might not be any feedback from the simulation system to the physical system, information from the simulation system can always be used by an ex-

111

ternal system or human operator. For example, the weather cannot be influenced and it is therefore not possible to build a ‘weather control system’. However, symbiotic simulation can still be used to forecast the weather. These forecasts can then be used to make decisions regarding the evacuation of a region in case of a hurricane, for instance. In this case, the symbiotic simulation system is an open-loop system because the weather cannot be influenced. However, there is still feedback from the simulations which can be of use for an external decision maker.

5.1

Closed-loop Systems

Symbiotic

Simulation Figure 2. Overview of symbiotic simulation control system (SSCS).

In a closed-loop symbiotic simulation system, a control feedback is created which affects the physical system. The simulation system uses what-if experiments to investigate alternative decision making scenarios and the best decision is determined by analysing the simulation results. This decision is either proposed to an external decision maker or directly implemented by means of actuators. We therefore distinguish between a symbiotic simulation decision support system (SSDSS) and a symbiotic simulation control system (SSCS). An SSDSS supports an external decision maker rather than implementing decision directly. Control of the physical system is entirely up to the external process which may or may not consider input from the SSDSS. Therefore, an SSDSS only indirectly influences the physical system. An SSCS is the extension of an SSDSS which is capable of directly implementing decisions by means of actuators. An overview of an SSDSS and an SSCS is illustrated in Figure 1 and 2, respectively.

by fire fighters for situation assessment [28], simulation of alternative threat management scenarios upon an incident in a water distribution network [22], and dynamic path planning of a UAV using what-if scenarios [16]. In these examples, various scenarios are simulated and analysed to make appropriate decisions. There are also examples in which what-if scenarios are not explicitly mentioned. These include traffic control and management [14], and electric power transmission systems [26, 27]. Examples for possible application of an SSCS, which uses actuators to directly implement decisions, include the use of controlling agents to make necessary modifications to a semiconductor manufacturing system [20], on-line planning and control in manufacturing [6], implementation of simulated situational escalation scenarios by activating actuation mechanisms [29], and control of hydraulic gates in the context of floodwater diversion [36].

5.2

Open-loop Symbiotic Simulation Systems

In an open-loop symbiotic simulation system no feedback is created to the physical system. Simulation of scenarios can be used for several purposes. A symbiotic simulation forecasting system (SSFS) can be used to forecast the behaviour of the physical system. Several scenarios are considered in such a forecast, each reflecting slightly different assumptions which are used to drive the simulation. The output of these what-if simulations can be used by an external process for visualisation purposes or further analysis. An SSFS is similar to an SSDSS or SSCS as it predicts future states of the physical system depending on a number of what-if scenarios. However, in contrast to SSDSS and SSCS, an SSFS does not interpret simulation results in order to draw any conclusions. An overview of an

Figure 1. Overview of symbiotic simulation decision support system (SSDSS).

An SSDSS can be applied in various applications. For example, this includes answering what-if questions asked

112

SSFS is illustrated in Figure 3.

[19], prediction of material properties of geological structures [1], and testing of hypotheses about emergency events [21]. A symbiotic simulation anomaly detection system (SSADS) can be used to detect anomalies either in the underlying simulation model or in the physical system. It uses the reference model and compares the simulated behaviour of the physical system continuously with the actual behaviour of the physical system. An anomaly is detected if the discrepancy between the measured behaviour and the actual behaviour of the physical system is beyond a certain threshold. Although an SSADS can be used to detect anomalies, it does not distinguish whether an anomaly is due to an abnormal behaviour of the physical system or due to model inaccuracy. However, it is possible to determine the source of the anomaly with support from an external operator or process.

Figure 3. Overview of symbiotic simulation forecasting system (SSFS).

For example, an SSFS can be applied to short-term wildland fire prediction [24, 9, 25] and image guided neurosurgery [23]. A symbiotic simulation model validation system (SSMVS) can be used for model validation purposes. It aims to determine a model which describes the current behaviour of the physical system with sufficient accuracy. This particular model is further referred to as reference model mR . Each what-if scenario in an SSMVS uses a different model or slightly modified version of the current reference model. Unlike SSFS and SSDSS/SSCS which aim to predict the future behaviour of the physical system, an SSMVS aims to emulate the current behaviour of the physical system. A model which describes the behaviour of the physical system better than the current reference model is used as a reference model for all subsequent simulations. An overview of an SSMVS is illustrated in Figure 4.

Compared with other symbiotic simulation systems, which simulate many scenarios concurrently, an SSADS is different because it uses a single scenario only. This scenario uses the reference model mR and reflects the current state of the physical system. An overview of an SSADS is illustrated in Figure 5.

Figure 5. Overview of symbiotic simulation anomaly detection system (SSADS).

For example, an SSADS can be used to detect structural damage by comparing simulated and measured values, as described in [5]. An example of detecting an inaccurate model in the context of a social sciences application is described in [18]. Discrepancies between expected and actual behaviour are detected in the context of exception management on a shop floor in [17].

Figure 4. Overview of symbiotic simulation model validation system (SSMVS).

For example, an SSMVS can be used for determination of unknown boundary conditions in fluid-thermal systems

113

6

A Hybrid Symbiotic Simulation System and its Application

Each of the different types of symbiotic simulation systems can be applied independently. For example, a single SSCS is used in [20] to control semiconductor manufacturing backend operations. However, depending on the application context it makes sense to combine various symbiotic simulation systems. For example, the WIPER application [21] consists of several subsystems which are responsible for anomaly detection, simulation-assisted hypothesis testing (model validation), and decision support. Each of these subsystems could be realised with a symbiotic simulation system. A system which consists of a number of symbiotic simulation subsystems is further referred to as a hybrid symbiotic simulation system.

6.1

Figure 6. Overview of hybrid symbiotic simulation system for semiconductor manufacturing equipment control.

Application in Semiconductor Manufacturing

Our current research efforts are concerned with the application of symbiotic simulation in the context of semiconductor manufacturing. Semiconductor manufacturing is a complex and asset intensive process which turns pure silicon wafers into integrated circuits with thousands of components. This process requires up to several hundred steps and up to three months for production [33]. A variety of tools are used in this manufacturing process and a single tool can cost up to US$ 2 million [34]. Improving tool efficiency is important to reduce the need for additional equipment. In addition, various other performance metrics, such as cycle time, have to be considered. The performance of a semiconductor fabrication plant (fab) highly relies on operational decision making for equipment control. Current efforts of the industry focus on automation solutions which help to improve the performance of a fab. Although simulations have already been used for off-line analysis and optimisation, their use in on-line solutions is still in their infancy. To demonstrate the applicability of symbiotic simulation in semiconductor manufacturing, we propose a hybrid symbiotic simulation system, consisting of four different symbiotic simulation subsystems. An overview of the proposed system is illustrated in Figure 6. In this hybrid symbiotic simulation system, a SSCS is used to conduct what-if experiments regarding the configuration of the various equipment of the fab. Decisions made by the SSCS are directly implemented by means of actuators. For example, if demand for a certain product has increased, various tools can be reconfigured in order to maintain performance targets. In addition, the SSCS can be used to find an immediate solution in order to minimise the negative effects on performance due to the breakdown of a specific tool, for instance.

The decision making process depends entirely on conducting what-if experiments. Using a simulation model that is not sufficiently accurate will therefore affect the quality of the decision making process. This may possibly lead to sub-optimal decisions. An SSMVS is responsible for updating the reference model for subsequent use by the SSCS. The importance of continuous model validation has already been explained in [6]. An SSADS is used to detect anomalies either in the reference model or in the physical system. An SSADS can be used to trigger other symbiotic simulation subsystems. If an anomaly is detected which is caused by a problem in the physical system (e.g., wrong machine settings), the SSCS is triggered to find a solution which reduces the negative effects of the anomaly. If no appropriate solution is found, external operators can be notified about the anomaly. The SSADS can also be used to trigger the SSMVS if the cause of the anomaly is model inaccuracy. In addition, an SSFS provides forecasts of the physical system. Unlike the SSCS which uses forecasts for decision making, the SSFS aims to provide forecasts to external operators who monitor the manufacturing system and the control system. This information can then be used for highlevel analysis purposes and long-term decision making.

6.2

Realisation and Evaluation

We aim to realise a hybrid symbiotic simulation system, as described above, using an agent-based framework for symbiotic simulation systems [3]. This framework supports different types of symbiotic simulation systems and

114

can be used to design symbiotic simulation systems which are tailored according to application specific requirements. Depending on the particular problem, the use of some types of symbiotic simulation systems might not be necessary or possible. For example, a weather forecasting application may use symbiotic simulation for forecasting and model validation but not a control system. In the context of operational control of semiconductor manufacturing equipment, we will use four different types of symbiotic simulation systems. So far, we have realised the SSCS [4] which is the most important part of the hybrid symbiotic simulation system as it makes decisions and controls the physical system. An emulator of the physical system was used rather than a real semiconductor manufacturing system for practical reasons. The physical system, i.e., the emulator, is connected to the symbiotic simulation system using corresponding sensors and actuators. The sensors provide information regarding the performance and the current state of the physical system. An actuator is used to allow the SSCS to exercise control over the settings of the various tools of the physical system. Our experiments have shown that using symbiotic simulation control yields higher performance compared to the common practise approach which is currently used in the semiconductor manufacturing industry. The results of these experiments suggest that significantly higher load can be handled when using symbiotic simulation [4]. Both, the emulator simulation and the what-if simulations of the SSCS use the same model of the physical system. Therefore, a case as described in Example II in Section 3 cannot occur. The SSMVS, which continuously calibrates the reference model, and the SSADS, which can be used to detect anomalies, are both not required at this stage. A SSFS has been realised and used for visualisation purposes.

7

work, we will further investigate the application of the different types of symbiotic simulation systems in the context of semiconductor manufacturing and other applications. In further research we will also investigate generic solutions for the various functional components of a symbiotic simulation system.

References [1] V. Akcelik, J. Bielak, G. Biros, I. Epanomeritakis, O. Ghattas, L. F. Kallivokas, and E. J. Kim. A framework for online inversion-based 3D site characterization. In Proceedings of the International Conference on Computational Science, pages 717–724, 2004. [2] S. Annan and J. Banks. Design of a knowledge-based online simulation system to control a manufacturing shop floor. IIE Transactions, 24(3):72–83, 1992. [3] H. Aydt, S. J. Turner, W. Cai, and M. Y. H. Low. An agentbased generic framework for symbiotic simulation systems. In A. M. Uhrmacher and D. Weyns, editors, Agents, Simulation and Applications (to appear). Taylor & Francis, 2008. [4] H. Aydt, S. J. Turner, W. Cai, M. Y. H. Low, P. Lendermann, and B. P. Gan. Symbiotic simulation control in semiconductor manufacturing. In Proceedings of the International Conference on Computational Science (to appear), 2008. [5] J. Cortial, C. Farhat, L. J. Guibas, and M. Rajashekhar. Compressed sensing and time-parallel reduced-order modeling for structural health monitoring using a DDDAS. In Proceedings of the International Conference on Computational Science, pages 1171–1179, 2007. [6] W. Davis. Handbook of simulation. In J. Banks, editor, On-Line Simulation: Need and Evolving Research Requirements, pages 465–516. Wiley-Interscience, 1998. [7] A. de Bary. Die Erscheinung der Symbiose. Verlag von Karl J. Tr¨ubner, Strasbourg, 1879. [8] A. Douglas. Symbiotic Interactions. Oxford University Press, 1994. [9] C. C. Douglas, J. D. Beezley, J. L. Coen, D. Li, W. Li, A. K. Mandel, J. Mandel, G. Qin, and A. Vodacek. Demonstrating the validity of a wildfire DDDAS. In Proceedings of the International Conference on Computational Science, pages 522–529, 2006. [10] G. R. Drake and J. S. Smith. Simulation system for realtime planning, scheduling, and control. In Proceedings of the Winter Simulation Conference, pages 1083–1090, 1996. [11] H. ElMaraghy, I. Abdallah, and W. ElMaraghy. On-line simulation and control in manufacturing systems. CIRP AnnalsManufacturing Technology, 47(1):401–404, 1998. [12] R. Fujimoto, D. Lunceford, E. Page, and A. M. U. (editors). Grand challenges for modeling and simulation: Dagstuhl report. Technical Report 350, Schloss Dagstuhl. Seminar No 02351, August 2002. [13] R. M. Fujimoto. Parallel and Distributed Simulation Systems. Wiley Series on Parallel and Distributed Computing. John Wiley & Sons, Inc., New York, NY, USA, 2000.

Conclusions

In this paper, we have given an extended definition of symbiotic simulation in relation to on-line simulation and DDDAS. This definition is motivated by symbiosis in biology and as a result we have identified five different types of symbiotic simulation systems. Each of them can either be used independently or in combination with others. In this context, we have proposed a hybrid symbiotic simulation system which can be used in semiconductor manufacturing for real-time equipment control. Unlike related work in symbiotic simulation which is application specific, we aim to provide generic solutions which can be used in the context of various applications. Although we describe a showcase in the context of semiconductor manufacturing, symbiotic simulation can be applied in many other domains. Examples of such applications can be found in the literature, in particular in related work on DDDAS. In future

115

[14] R. M. Fujimoto, R. Guensler, M. Hunter, H. K. Kim, J. Lee, J. Leonard, M. Palekar, K. Schwan, and B. Seshasayee. Dynamic data driven application simulation of surface transportation systems. In Proceedings of the International Conference on Computational Science, pages 425–432, 2006. [15] F. Kamrani. Using on-line simulation in UAV path planning. Licentiate Thesis in Electronics and Computer Systems, KTH, Stockholm, Sweden, 2007. [16] F. Kamrani and R. Ayani. Using on-line simulation for adaptive path planning of UAVs. In Proceedings of the 11th IEEE International Symposium on Distributed Simulation and Real-time Applications, pages 167–174, Chania, Greece, October 2007. [17] D. Katz and S. Manivannan. Exception management on a shop floor using online simulation. In Proceedings of the Winter Simulation Conference, pages 888–896, 1993. [18] C. Kennedy, G. K. Theodoropoulos, V. Sorge, E. Ferrari, P. Lee, and C. Skelcher. AIMSS: An architecture for data driven simulations in the social sciences. In Proceedings of the International Conference on Computational Science, pages 1098–1105, 2007. [19] D. Knight, Q. Ma, T. Rossman, and Y. Jaluria. Evaluation of fluid-thermal systems by dynamic data driven application systems - part ii. In Proceedings of the International Conference on Computational Science, pages 1189–1196, 2007. [20] M. Y. H. Low, K. W. Lye, P. Lendermann, S. J. Turner, R. T. W. Chim, and S. H. Leo. An agent-based approach for managing symbiotic simulation of semiconductor assembly and test operation. In Proceedings of the 4th International Joint Conference on Autonomous Agents and Multiagent Systems, pages 85–92, New York, NY, USA, 2005. ACM Press. [21] G. R. Madey, A.-L. Barab´asi, N. V. Chawla, M. Gonzalez, D. Hachen, B. Lantz, A. Pawling, T. Schoenharl, G. Szab´o, P. Wang, and P. Yan. Enhanced situational awareness: Application of DDDAS concepts to emergency and disaster management. In Proceedings of the International Conference on Computational Science, pages 1090–1097, 2007. [22] K. Mahinthakumar, G. von Laszewski, S. R. Ranjithan, D. Brill, J. Uber, K. Harrison, S. Sreepathi, and E. M. Zechman. An adaptive cyberinfrastructure for threat management in urban water distribution systems. In Proceedings of the International Conference on Computational Science, pages 401–408, 2006. [23] A. Majumdar, A. Birnbaum, D. J. Choi, A. Trivedi, S. K. Warfield, K. Baldridge, and P. Krysl. A dynamic data driven grid system for intra-operative image guided neurosurgery. In Proceedings of the International Conference on Computational Science, pages 672–679, 2005. [24] J. Mandel, J. D. Beezley, L. S. Bennethum, S. Chakraborty, J. L. Coen, C. C. Douglas, J. Hatcher, M. Kim, and A. Vodacek. A dynamic data driven wildland fire model. In Proceedings of the International Conference on Computational Science, pages 1042–1049, 2007. [25] J. Mandel, M. Chen, L. P. Franca, C. J. Johns, A. Puhalskii, J. L. Coen, C. C. Douglas, R. Kremens, A. Vodacek, and W. Zhao. A note on dynamic data driven wildfire modeling. In Proceedings of the International Conference on Computational Science, pages 725–731, 2004.

[26] J. D. McCalley, V. Honavar, S. M. Ryan, W. Q. Meeker, D. Qiao, R. A. Roberts, Y. Li, J. Pathak, M. Ye, and Y. Hong. Integrated decision algorithms for auto-steered electric transmission system asset management. In Proceedings of the International Conference on Computational Science, pages 1066–1073, 2007. [27] J. D. McCalley, V. Honavar, S. M. Ryan, W. Q. Meeker, R. A. Roberts, D. Qiao, and Y. Li. Auto-steered informationdecision processes for electric system asset management. In Proceedings of the International Conference on Computational Science, pages 440–447, 2006. [28] J. Michopoulos, P. Tsompanopoulou, E. N. Houstis, and A. Joshi. Agent-based simulation of data-driven fire propagation dynamics. In Proceedings of the International Conference on Computational Science, pages 732–739, 2004. [29] J. Michopoulos, P. Tsompanopoulou, E. N. Houstis, J. R. Rice, C. Farhat, M. Lesoinne, and F. Lechenault. DDEMA: A data driven environment for multiphysics applications. In Proceedings of the International Conference on Computational Science, pages 309–318, 2003. [30] National Science Foundation. DDDAS: Dynamic data driven applications systems. Program Solicitation 05-570, http://www.nsf.gov/pubs/2005/nsf05570/ nsf05570.htm, 2005. [31] S. Paracer and V. Ahmadjian. Symbiosis: An Introduction to Biological Associations. Oxford University Press US, 2000. [32] K. Perumalla, R. Fujimoto, T. McLean, and G. Riley. Experiences applying parallel and interoperable network simulation techniques in on-line simulations of military networks. In Proceedings of the 16th Workshop on Parallel and Distributed Simulation, pages 88–95, 2002. [33] J. Potoradi, O. Boon, J. Fowler, M. Pfund, and S. Mason. Using simulation-based scheduling to maximize demand fulfillment in a semiconductor assembly facility. In Proceedings of the Winter Simulation Conference, pages 1857–1861, 2002. [34] W. Scholl and J. Domaschke. Implementation of modeling and simulation in semiconductor wafer fabrication with time constraints between wet etch and furnace operations. IEEE Transactions on Semiconductor Manufacturing, 13(3):273– 277, August 2000. [35] D. Wilkinson. At cross purposes. Nature, 412(6846):485, 2001. [36] L. Xue, Z. Hao, D. Li, and X. Liu. Dongting lake floodwater diversion and storage modeling and control architecture based on the next generation network. In Proceedings of the International Conference on Computational Science, pages 841–848, 2007. [37] T. Ye, D. Harrison, B. Mo, B. Sikdar, H. Kaur, S. Kalyanaraman, B. Szymanski, and K. Vastola. Traffic management and network control using collaborative on-line simulation. In Proceedings of the IEEE International Conference on Communications, pages 204–209, 2001.

116

Suggest Documents