Showing papers by "Richard M. Fujimoto published in 2017"
01 Jan 2017
TL;DR: The model framework layer provides interface code to allow independently created submodels to communicate, synchronize, and interoperate in various ways to become a single federated model.
Abstract: ion layer Function Model layer The code of a particular model (or component) Model framework layer Collection of model classes, components, and libraries for a single application area, such as network simulation or PDE solution Simulator layer Provides a single paradigm for simulation time, space, naming, parallelism, and synchronization for use in one component of a (possibly) federated simulation Component federation layer Provides interface code to allow independently created submodels, possibly written in different languages, to communicate, synchronize, and interoperate in various ways to become a single federated model Load management layer Within one parallel model execution, measures resource utilization (time, energy, bandwidth, memory) at runtime and dynamically manages or migrates loads to optimize some performance metric Ensemble layer Runs many instances of the same model as an ensemble in a single large job, for such purposes as parameter sensitivity studies, parameter optimization, variance estimation. Handles scheduling, failures, accounting, and time estimates, allocates file directories, decides on ensemble termination, etc. Operating system /job scheduler layer Runs independent jobs in parallel. Provides processes, interprocess communication, I/O, files systems, etc. 62 C. Carothers et al.
67 citations
01 Jan 2017
TL;DR: The results suggest that this approach offers a robust and effective method of calibrating simulation models where disaggregate-level vehicle data are available—which is becoming more prevalent with further advancements in mobile sensor and connected vehicle technologies.
Abstract: Efforts to address operational issues in transportation have been the focus of many research efforts. A number of these efforts were geared toward developing microscopic traffic simulation models to accurately represent the complex and dynamic operation of a transportation network. One of the challenges with such models is that they do not always adequately reflect field conditions-particularly when representing traffic operations across different time periods. This paper presents a robust calibration procedure that aims to increase the accuracy of calibrated microscopic traffic simulation models. This procedure is based on a Monte Carlo approach to generate candidate parameter sets, which are aimed to produce calibrated simulation models. Model runs of these parameter sets are evaluated against robust calibration criteria, including startup and saturation flow characteristics and travel time distributions. The parameter sets that satisfy these criteria are considered as adequately calibrated to accurately reflect field performance measures. In applying this procedure, the results suggest that this approach offers a robust and effective method of calibrating simulation models where disaggregate-level vehicle data are available-which is becoming more prevalent with further advancements in mobile sensor and connected vehicle technologies.
20 citations
03 Dec 2017
TL;DR: The origins and development of the parallel discrete event simulation field are described in the words of many who were deeply involved, providing a unique characterization of how the field was created and developed.
Abstract: Originating in the 1970's, the parallel discrete event simulation (PDES) field grew from a group of researchers focused on determining how to execute a discrete event simulation program on a parallel computer while still obtaining the same results as a sequential execution. Over the decades that followed the field expanded, grew, and flourishes to this day. This paper describes the origins and development of the field in the words of many who were deeply involved. Unlike other published work focusing on technical issues, the emphasis here is on historical aspects that are not recorded elsewhere, providing a unique characterization of how the field was created and developed.
17 citations
03 Dec 2017
TL;DR: Position statements by leading researchers are presented concerning important computational challenges and opportunities facing the M&S community.
Abstract: Modeling and simulation faces many new computational challenges in the design of complex engineered systems. The systems that need to be modeled are increasingly interconnected and interdependent, achieving unprecedented levels of complexity. The computational platforms upon which simulations execute have undergone dramatic changes in recent years. Position statements by leading researchers are presented concerning important computational challenges and opportunities facing the M&S community.
12 citations
TL;DR: A real-time data driven arterial simulation methodology is proposed to assist such computations that are performed over a testbed in the midtown area of Atlanta, Georgia and field results provide evidence to validate the proposed approach.
Abstract: Congestion-induced delays and pollution in modern transportation systems remain formidable impediments to the sustainable growth of our cities. Next generation Intelligent Transportation Systems (ITS) will attack these problems by relying on extensive in-vehicle sensing, crowd-sourced data, ubiquitous computing, and communications to augment existing infrastructure-based deployments. Advances in wireless networking and mobile computing have made it possible to create dynamic, data driven application systems (DDDAS) that address many challenges in modern transportation systems. We outline a vision for future dynamic data-driven transportation systems, and focus on the effectiveness of an approach to real-time management based on online simulations. The online simulations are embedded in the traffic network where distributed simulators perform the modeling task individually but project the future states collectively. A real-time data driven arterial simulation methodology is proposed to assist such computations that are performed over a testbed in the midtown area of Atlanta, Georgia. Field results are presented that provide evidence to validate the proposed approach.
12 citations
16 May 2017
TL;DR: Work to characterize the power and energy consumed by different elements of parallel and distributed simulation systems are discussed and empirical measurements presented to quantify energy and power use are presented, suggestive of directions for future research in this area.
Abstract: Energy and power consumption have become important concerns for many computing systems ranging from embedded and mobile systems operating on battery-powered devices to high performance and cloud computing applications running on supercomputers and in data centers. To date, only a limited amount of work has considered power consumption in parallel and distributed simulations. A variety of options to analyze and explore power consumption in distributed simulations are discussed. These options range from design decisions in developing the simulation model to selection of algorithms in distributed simulation middleware to exploitation of hardware techniques. Work to characterize the power and energy consumed by different elements of parallel and distributed simulation systems are discussed and empirical measurements presented to quantify energy and power use, suggestive of directions for future research in this area.
11 citations
TL;DR: This study suggests that existing distributed simulation algorithms require a significant amount of additional energy compared to a sequential execution, and different synchronization algorithms can yield different energy consumption behaviors.
Abstract: Power and energy consumption are important concerns in the design of high performance and mobile computing systems, but have not been widely considered in the design of parallel and distributed simulations. The importance of these factors is discussed and metrics for power and energy overhead in parallel and distributed simulations are proposed. Factors affecting the energy consumed by synchronization algorithms and software architectures are examined. An experimental study is presented examining energy consumption of the well-known Chandy/Misra/Bryant and YAWNS synchronization algorithms. The effects of lookahead and event communication on energy use are examined. Initial results concerning queueing network simulations are also presented. The results of this study suggest that existing distributed simulation algorithms require a significant amount of additional energy compared to a sequential execution. Further, different synchronization algorithms can yield different energy consumption behaviors.
8 citations
01 Jan 2017
TL;DR: This chapter outlines key computational challenges associated with maximally exploiting emerging computing platforms ranging from handheld mobile computers to massively parallel supercomputers, use of models and simulations in interacting with the real world,challenges associated with managing a “plethora of models” that exists today, and synergies between M&S and “big data.”
Abstract: Computing and communication technologies have advanced rapidly in the last decade. M&S has not yet fully realized the potential and opportunities afforded by technologies such as mobile and ubiquitous computing, big data, the Internet of Things, cloud computing, and modern supercomputer architectures. This has kept M&S from achieving its fullest potential in modeling complex systems, or being widely deployed in new contexts such as online management of operational systems. Research advances are needed to enable M&S technologies to address issues such as the complexity and scale of the systems that need to be modeled today. This chapter outlines key computational challenges associated with maximally exploiting emerging computing platforms ranging from handheld mobile computers to massively parallel supercomputers, use of models and simulations in interacting with the real world, challenges associated with managing a “plethora of models” that exists today, and synergies between M&S and “big data.”
6 citations
03 Dec 2017
TL;DR: This paper addresses the following questions: what was the state of the art in M&S 50 years ago, what is it today, how much progress has been made, and what were the obstacles to innovation?
Abstract: Innovation in Modeling and Simulation (M&S) refers to exploiting new ideas, exploiting new technology, and employing out-of-the-box thinking, which lead to the creation of new methodologies, techniques, concepts, frameworks, and software. This paper addresses the following questions: (a) what was the state of the art in M&S 50 years ago, what is it today, how much progress has been made? (b) how much innovation in M&S has been accomplished over the last half a century? (c) what were the obstacles to innovation in M&S? (d) what are some recommendations to promote innovation in M&S? (e) what message should be sent to the funding agencies to encourage innovation in M&S?
3 citations
01 Jan 2017
TL;DR: This chapter is about the history of the Time Warp algorithm and optimistic approaches to parallel discrete event simulation and active developers of the ideas over several decades.
Abstract: This chapter is about the history of the Time Warp algorithm and optimistic approaches to parallel discrete event simulation. It concentrates on the early history from our personal perspective as active developers of the ideas over several decades.
3 citations
03 Dec 2017
TL;DR: Experimental data illustrate that region based approaches tend to consume more energy than grid based approaches for computations, but less for communications, and that the choice of grid cell size and grid cell constraints on publication regions can play an important role in the energy efficiency of grid based DDM approaches.
Abstract: Energy and power aware data distribution methods are essential for using these approaches in energy constrained devices and environments. Data Distribution Management (DDM) is a set of services defined in the High Level Architecture (HLA) that aims to efficiently propagate distributed simulation state information. This paper describes an empirical study of the energy consumption of computation and communication components of region based and grid based DDM approaches in mobile devices. Experimental data illustrate that region based approaches tend to consume more energy than grid based approaches for computations, but less for communications. These results also show that the choice of grid cell size and grid cell constraints on publication regions can play an important role in the energy efficiency of grid based approaches.
03 Dec 2017
TL;DR: A variety of options to reduce power consumption in parallel and distributed simulation systems are discussed, suggestive of directions for future research in this increasingly important area.
Abstract: The energy and power consumed by computing applications have long been important concerns in mobile systems and have recently become of great interest in high performance and cloud computing. To date, only a limited amount of work has considered power consumption in parallel and distributed simulation systems. A variety of options to reduce power consumption in these systems are discussed, suggestive of directions for future research in this increasingly important area.
TL;DR: This paper considers the use of simulation cloning for estimation of performance measures via transient simulation, and presents simulation cloning algorithms with induced negative correlation to reduce variance.
Abstract: Simulation cloning involves expediting a simulation by sharing computational results among different sample paths. It resembles the idea of splitting, which is widely researched in rare event simulation, as splitting techniques also produce clones when a sample path reaches a certain state. In this paper, we consider the use of simulation cloning for estimation of performance measures via transient simulation. In addition to the basic cloning approach, we present simulation cloning algorithms with induced negative correlation to reduce variance. We also consider the possibility of cloning the simulation at multiple (two) decision points. Numerical experiments are provided to illustrate the effectiveness of our algorithms. In real-life situations where simulating one replication is expensive and variability is larger in the replicated part of the simulation, simulation cloning can significantly improve the efficiency of the simulation.
03 Dec 2017
TL;DR: Experimental results utilizing real-world traffic data for the continental U.S. are presented demonstrating the speed ups achieved by a prototype simulator, illustrating that time-parallel simulation can be used to significantly accelerate certain air traffic simulations.
Abstract: Air traffic management is widely studied in several different fields because of its complexity and criticality to a variety of stakeholders. However, the exploding amount of air traffic in recent years has created new challenges to ensure effective management of the airspace. A fast time simulation capability is essential to effectively explore the consequences of decisions of air traffic management. A new algorithm for simulating air traffic networks using a time-parallel simulation approach is proposed that distributes time segments of the simulation scenarios across different processors. A simulation model for the National Airspace System (NAS) is described and validated. The components of the simulator are described as well as the parallel simulation algorithms. Experimental results utilizing real-world traffic data for the continental U.S. are presented demonstrating the speed ups achieved by a prototype simulator. These results illustrate that time-parallel simulation can be used to significantly accelerate certain air traffic simulations.
07 Jul 2017