Richter, R. and J. C. Walrand. 1989. Distributed simulation of discrete event systems. Proceedings of the IEEE 77(1):99-113.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....reducing the Operating System intervention. It is a network standard which provides 1.25 Gb s bi directional communication speed. LSDS decomposes the computation for simulation into loosely coupled processes, and assign the simulation of each process or several processes to one or more processors [5], as shown in Figure 3. The messages exchanged between the processes are the events that are simulated using LSDS here. Processor1 Processor2 DES e4 06 07 I DTS FI FI FL, me n H Scaled Time l tt e2 e3 e5 j VIA SAN P1, P2, P3 : Parts M1, M2 : Machines Figure 3. ....

Rhonda Righter and Jean C. Waltand "Distributed simulation of discrete event systems", Proceedings of the IEEE, 77, 99-113, 1989.

....to that point in the continuum between TW and CMB protocols, that is most appropriate for the parallelism inherent in the simulation model. 5 Sources of Literature and Further Reading Overviews on the field of parallel and distributed simulation are the surveys by Richter and Walrand [66], Fujimoto [35] and most recently Nicol and Fujimoto [59] and Ferscha [28] The primary reading for Time Warp is [44] for conservative protocols it is [52] The most relevant literature in the context of parallel discrete event simulations of Petri nets is [7] 6] 67] 8] while the ....

R. Richter and J. C. Walrand. Distributed Simulation of Discrete Event Systems. Proceedings of the IEEE, 77(1):99 -- 113, Jan 1989.

....order. To satisfy this causality constraint, a synchronization mechanism is required. Two of the most common synchronization protocols for parallel simulation are the Chandy Misra protocol [2] and the Time Warp protocol [7] Di#erent approaches for parallel simulation are discussed elsewhere [5, 10, 13, 16, 22, 26, 27]. An introduction to the Chandy Misra protocol and the Time Warp protocol can be found in [5] Recently there has been a great deal of interest in performance evaluation of parallel simulation. Most work [3, 4, 13, 14, 15, 18, 19, 20, 21, 23, 24, 29] is devoted to the time complexity and assumes ....

Righter, R., and Walrand, J.C. Distributed Simulation of Discrete Event Systems. Proceedings of the IEEE, 77(1), January 1989.

....of time, and if it will not generate an event during this period, it should send a null message to all its output links. In [46] it is proved that there is no deadlock with this null message algorithm. Variations of the conservative algorithm have been developed, see [43] for more details and [47] for an excellent overview and comparisons of different conservative algorithms. 74 In the optimistic approach, or Time Warp, as it is called by Jefferson and Sowizral [48] every processor checks its input links and executes the event with the lowest time stamp. Contrary to the conservative ....

R. Righter and J. C. Walrand, "Distributed simulation of discrete-event systems," Proceedings of the IEEE, vol. 77, pp. 99--113, January 1989.

....performance of the throttling scheme with the conventional TW scheme. Section 6 contains our concluding remarks. 2 Parallel and Distributed Simulation Research on PDES during the last two decades is documented in various landmark survey and tutorial articles at different development stages [11, 12, 13, 23, 26]. Briefly, PDES adopts the process oriented modeling methodology to partition an application system into physical processes that interact at simulated time. The simulator consists of a set of logical processes (LP) one per physical process. LPs are mapped onto a machine for execution based on its ....

R. Righter and J.C. Walrand, "Distributed Simulation of Discrete Event Systems", Proc. of the IEEE, Vol. 77, No. 1, pp. 99-113, January 1989.

....of events that need to be processed. 5 A third alternative is to apply statistical methods in order to reduce the number of runs required to make decisions regarding certain characteristics of the simulated system [Ripl87] these techniques however are at still at an early stage of development [Righ89]. 6 Righter [Righ89] mentions one more approach, namely applying a parallelizing compiler to a sequential simulation program. Although this technique is transparent to the user, it exploits only a small portion of the available parallelism and thus is very rarely used. CHAPTER 3. MODELLING AND ....

....to be processed. 5 A third alternative is to apply statistical methods in order to reduce the number of runs required to make decisions regarding certain characteristics of the simulated system [Ripl87] these techniques however are at still at an early stage of development [Righ89] 6 Righter [Righ89] mentions one more approach, namely applying a parallelizing compiler to a sequential simulation program. Although this technique is transparent to the user, it exploits only a small portion of the available parallelism and thus is very rarely used. CHAPTER 3. MODELLING AND SIMULATION 60 onto ....

[Article contains additional citation context not shown here]

Righter, R., Warland, J. C., "Distributed Simulation of Discrete Event Systems", Proceedings of the IEEE, 77, 1, January 1989, pp. 99-113.

.... simulation of event based models (so called discrete event models) and to tie these generic concepts to a specific application: personal communications services (PCS) Several introductory articles have been presented in the literature such as Fujimoto [1] Nicol et al. 2] and Richter et al. [3]. These papers have helped to disseminate the asynchronous parallel discrete event simulation (PDES) methodology for a wide readership. Our approach is similar but stresses a single real world application for discussing the methodology of PDES. By defining the methodology and all PDES terminology ....

Richter, R. and Walrand, J.C., "Distributed Simulation of Discrete Event Systems", Proceedings of the IEEE, vol. 77, no. 1, pp. 99--113, January 1989.

....work of DES in parallel dates back roughly 20 years. Hundreds of papers have been published since then, the majority appearing within this decade (see References and Bibliography) Various ways of decomposing a simulation for processing on multiple processors have been proposed, as outlined below [RW89,Fer95, Cal96] 1. Parallelizing Compilers. In this approach, parallelizing compilers are used to exploit the parallelism available in a given sequential program. This approach requires no changes in the code for sequential simulation, and thus is readily applicable to many existing sequential ....

....way to accurately predict the initial state of each interval. Nevertheless, the idea is elegant and deserves future research. Among these classes of approaches, the space parallel decomposition shows the greatest potential and is considered the most promising approach to perform DES in parallel [RW89,Win92] Since the event list is also decomposed into individual ones, the event lists would not become the bottleneck as with the fourth approach (distributed events with centralized event list) A higher degree of parallelism is expected since this class of approaches encourages concurrent ....

Rhonda Righter and Jean C. Walrand. Distributed simulation of discrete event systems. Proceedings of the IEEE, 77(1):99--113, January 1989.

....of this many researchers are interested in parallelizing their execution. One of the key issues is synchronization between processors, as the synchronization demands are highly variable, depending dynamically on the simulation model s state. Comprehensive surveys on the topic are found in [2] [21], and [15] Parallel simulation is hard because synchronization between processors is very dynamic and often unpredictable. Each processor maintains its own simulation clock, and we require that the end result of the simulation be consistent with a scenario in which every processor executes its ....

R. Righter and J.V. Walrand, "Distributed Simulation of Discrete Event Systems," Proceedings of the IEEE, vol. 77, no. 1, pp. 99-113, January 1989.

....carried out, the results of one simulation execution determines the performance indices for the next experiment. Towards this end, researchers have been looking at improving the performance of simulation through concurrent execution by exploiting the parallelism inherent in the simulation model [32, 34, 38, 40, 74]. The concurrent execution of a single simulation program on multiple processors is referred to as Parallel Discrete Event Simulation (PDES) Several paradigms for accelerating a simulation via parallel execution exist [13, 38, 56, 58, 74] The five main policies are: 5 Parallelizing Compilers ....

.... exploiting the parallelism inherent in the simulation model [32, 34, 38, 40, 74] The concurrent execution of a single simulation program on multiple processors is referred to as Parallel Discrete Event Simulation (PDES) Several paradigms for accelerating a simulation via parallel execution exist [13, 38, 56, 58, 74]. The five main policies are: 5 Parallelizing Compilers Use of a parallelizing compiler to determine code segments that may be executed in parallel. Replicated Experiments Parallel executions of a set of entire independent simulation programs, each on its own individual processor. Functional ....

R. Righter and J. Walrand. Distributed simulation of discrete event systems. In Proceedings of the IEEE, volume 77(1), pages 99--113, January 1989.

....currently processed events may affect the next event on the list. Therefore event dependencies must be known prior to scheduling. The distributed events approach is particularly appropriate for shared memory systems because in such an environment the event list can be accessed by all processors [Righter 89] The global event list is composed of all events sorted in chronological order. Safe events, that is those events whose execution do not depend upon previously unexecuted events, must be identified and scheduled. A master processor is required to maintain the global event list. Each available ....

Righter, R., Walrand, J. C., "Distributed Simulation of Discrete Event Systems," Proceedings of the IEEE, vol. 77, no. 1, Jan. 89, pp 99 - 113.

....state variables that are modified during the simulation. The synchronization mechanism must ensure that each LP processes events in timestamp order in order to prevent events in the simulated future from affecting those in the past. The synchronization issue has been widely studied (e.g. see [4, 6, 13, 16]) The Time Warp mechanism uses a detection and recovery protocol to synchronize the computation. For a more detailed discussion of the Time Warp mechanism we refer the reader to [4, 10] 3.2 Implementing the PCS Models A natural way to map both PCS models to a general purpose, sequential ....

R. Richter and J. C. Walrand. Distributed simulation of discrete event systems. Proceedings of the IEEE, 77(1):99-- 113, January 1989.

....e.g. denoting the arrival of a new portable at that cell. The synchronization mechanism must ensure that each LP processes events in timestamp order in order to prevent events in the simulated future from affecting those in the past. The synchronization issue has been widely studied (e.g. see [8, 9, 3, 4]) The Time Warp mechanism uses a detection and recovery protocol to synchronize the computation [7] Any time an LP determines that it has processed events out of timestamp order, it rolls back those events, and reexecutes them. Rolling back an incorrect event computation requires one to (1) undo ....

R. Richter and J. C. Walrand. Distributed simulation of discrete event systems. Proceedings of the IEEE, 77(1):99--113, January 1989.

....it an event with smaller time stamp s t, or it must be prepared to rollback and begin recomputing at time s if such an event is later scheduled. There is a rich literature of solutions to this synchronization problem. The seminal articles are [4] and [22] Good introductory surveys are found in [19, 38], and a survey of the state of the art is found in [31] Conservative synchronization protocols prohibit a simulator from executing an event if there is any possibility of an earlier event being scheduled there later. Optimistic protocols allow out of sequence event processing. Each style has its ....

R. Righter and J.V. Walrand. Distributed simulation of discrete event systems. Proceedings of the IEEE, 77(1):99--113, January 1989.

....to that point in the continuum between TW and CMB protocols, that is most appropriate for the parallelism inherent in the simulation model. 5 Sources of Literature and Further Reading Overviews on the field of parallel and distributed simulation are the surveys by Richter and Walrand [66], Fujimoto [35] and most recently Nicol and Fujimoto [59] and Ferscha [28] The primary reading for Time Warp is [44] for conservative protocols it is [52] The most relevant literature in the context of parallel discrete event simulations of Petri nets is [7] 6] 67] 8] while the ....

R. Richter and J. C. Walrand. Distributed Simulation of Discrete Event Systems. Proceedings of the IEEE, 77(1):99 -- 113, Jan 1989.

....The threat of stragglers is dealt with by saving the simulation state periodically, and rolling back as appropriate when a straggler arrives . Messages sent at times ahead of the straggler s time stamp must be undone. Fundamental problems of PDES are reviewed in Misra (1986) Fujimoto (1990) Righter and Walrand (1989). Nicol and Fujimoto (1994) give a more current state of the art review. Most PDES synchronization protocols fall into two basic categories. Conservative protocols (e.g. Chandy and Misra 1979, Bryant 1979, Peacock, Wong and Manning 1979, Lubachevsky 1988, Chandy and Sherman 1989 and Nicol 1993) do ....

Righter, R and J. Walrand 1989. Distributed Simulation of Discrete Event Systems. Proceedings of the IEEE, Vol. 77, No. 1 Jan. 1989.

....system refers as any application that modelizes space time phenomena. Three usual examples are: numerical resolution of partial differential equations [1] describing continuously evolving systems; discrete event simulation where the system transitions occur when discrete events happen [2]; hybrid systems simulation which are dynamical systems whose phase space involves continuous and discrete components [3] 4] A language for the modelisation and simulation of such systems requires the following features: being high level: most of the simulation processes are developed using ....

R.Righter, J.Walrand, Distributed simulation of discrete event systems, Proc. of the IEEE, vol.77, n°1, Jan.1989.

....by message intensity, communications overhead, and rollback frequency. Key words: parallel and distributed simulation, discrete event simulation, synchronization 1 Introduction Theory and practice of parallel and distributed simulations (PADS) have made significant progresses in recent years [7, 5, 11, 6, 12, 9]. In distributed simulation, the simulation tasks are divided into a number of logical processes or LP, which then are distributed to one or more processors. These LPs can execute the simulation in parallel. Thus a significant speedup often can be achieved. The correctness of the simulation is ....

Rhonda Righter and Jean Walrand. Distributed simulation of discrete event systems. Proceedings of the IEEE, 77(1), January 1989.

....military battlefield simulations, or performance analysis of computer systems or parallel programs. 2.2 Parallel Discrete Event Simulation DES of large models can be enormously time as well as memory consuming. As there seems to be quite a big amount of inherent parallelism in most DES models [Righ 89, Lin 92] it is a challenging task to find this parallelism and use it to efficiently implement DES on multiprocessor computer systems to increase the performance of DES runs. Methods to use multiprocessor architectures for DES are summarized as parallel discrete event simulation (PDES) ....

....systems to increase the performance of DES runs. Methods to use multiprocessor architectures for DES are summarized as parallel discrete event simulation (PDES) Figure 1. Logical processes for DDES. There are multiple approaches how to apply parallelization on DES implementations [Kaud 87, Righ 89] depending on which level the sequential simulation algorithm is parallelized: ffl A parallelizing compiler can be used. ffl Independent simulation runs can be performed on separate processors. ffl Subroutines and functions can be computed on separate processors. ffl A main processor can ....

R. Righter and J. C. Warland. "Distributed Simulation of Discrete Event Systems". Proceedings of the IEEE, Vol. 77, No. 1, pp. 99--113, January 1989.

....Simulation algorithms of single trajectories of DEDS which is required for performance and or performance gradient estimation are inherently difficult to parallelize. One must take great care so that one part of the distributed computation does not get too far ahead of another [Fujimoto 90, Righter and Walrand 89] The resultant overhead often negates much of the speed advantage. In short, two problems stand in the way of parallel implementation of traditional approaches : i) the optimization scheme is inherently sequential, ii) the simulation of a single trajectory of a DEDS is hard to parallelize. We ....

Righter, R. and Walrand, J.C., "Distributed Simulation of Discrete Event Systems", Proceedings of the IEEE, Vol. 77, No. 1, pp. 99-113,1989.

....and principles of LP simulation have been outlined, ending up in the presentation of the classical LP simulation protocols and the various optimizations that appeared. Further comprehensive readings on the field of parallel and distributed simulation are the surveys by Richter and Walrand [55], Fujimoto [26] and most recently Nicol and Fujimoto [49] The primary source for Time Warp is [35] for conservative protocols it is [46] A conclusive judgement on one of the protocols could not be drawn, since performance is always dependent on the respective simulation model and the ....

....a distributed DES protocol [61] The difficult problem of debugging parallel programs finds a high chance to be tackled by similar ideas. 6 Sources of Literature and Further Reading Comprehensive overviews on the field of parallel and distributed simulation are the surveys by Richter and Walrand [55], Fujimoto [26] and most recently Nicol and Fujimoto [49] The primary reading for Time Warp is [35] for conservative protocols it is [46] The most relevant literature appears in the frame of the ffl Workshop on Parallel and Distributed Simulation (PADS) formerly (while being held as part of ....

R. Richter and J. C. Walrand. Distributed Simulation of Discrete Event Systems. Proceedings of the IEEE, 77(1):99 -- 113, Jan 1989.

....other simulator can still schedule on it an event with smaller time stamp s t, or it must be prepared to rollback and begin recomputing at time s if such an event is later scheduled. There is a rich literature of solutions to this synchronization problem. Good introductory surveys are found in [10, 25], and a survey of the state of the art is found in [20] Conservative synchronization protocols prohibit a simulator from executing an event if there is any possibility of an earlier event being scheduled there later. Optimistic protocols allow out of sequence event processing. Each style has its ....

R. Righter and J.V. Walrand. Distributed simulation of discrete event systems. Proceedings of the IEEE, 77(1):99--113, January 1989.

....to DES although based on a very efficient algorithm [5] has been unable to provide but medium sized system simulations. In order to overcome this limitation, parallel approaches to DES have been considered since early 1980 s. Most parallel discrete event approaches reported so far fall [6 9] into one of the following classes. Automatic parallelism detection The attempts belonging to this category use a parallelizing compiler to detect parallelism in the sequential simulation code. As the structure of the problem is completely neglected, only a reduced speedup is attainable in this ....

Righter R., Walrand J.C., Distributed simulation of discrete event systems. In: Proceedings of the IEEE, vol. 77, no. 1, Jan. 1989, pp. 99--113.

No context found.

Richter, R. and J. C. Walrand. 1989. Distributed simulation of discrete event systems. Proceedings of the IEEE 77(1):99-113.

No context found.

R. Righter and J.C. Walrand, "Distributed Simulation of Discrete-event Systems", Proceedings of the IEEE, vol. 77, no. 1, pp. 99-113, 1989.

First 50 documents

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC