Search

The software R-TRANSFORM converts a conventional simulation model into a simulation model of the same system including one or more computers that apply the simulation model during the system operation in order to anticipate the near future and - according to the anticipation - to enhance the system efficiency. Simulation programming tools enable their users to describe the simulated system but not to program what should happen in the computer: the description is then automatically converted into the computer code. R-TRANSFORM enables an automatic nesting of one description of the simulated system into the other one, so that the model nesting is safe against a possible programming error that mixes both the descriptions. Simple applications of nesting of models are mentioned.

This tutorial introduces the MODSIM III language, showing how its simulation "world view" together with its object-oriented architecture and built in graphics contribute to successful simulation model building. 1 WHAT IS MODSIM III? Discrete systems simulation is one of the few truly new capabilities offered by the computing revolution. Computer simulation models provides us with a fascinating means to develop insight into the behaviors of the complex non-deterministic systems which surround us in communications networks, transportation logistics, business and manufacturing processes, to name a few. These systems are costly to develop and modify. Feasibility analysis and performance prediction through simulation can greatly reduce the risk of failure or wasteful expense. Simulation is an exploratory technique. We develop a model which represents our best understanding of a proposed system or modification. We run the model in the hope that it will confirm our design intuition. If it does not we endeavor to understand why, and use this knowledge to refine the model. Ultimately, we will use the model to explain and justify decisions affecting the real-world system. So, simulation serves three principal roles: to help us articulate a coherent description of a system; to validate its dynamic behavior to our satisfaction; and, then to communicate this to colleagues and decision makers. These steps are commonly iterative. It is usually wise to begin with a high-level abstraction of the system, allowing the model to evolve as we gain greater understanding of system behavior. Furthermore, the system itself may change in this rapidly changing environment, requiring changes to the model. For the simulation study to be effective, the model behavior should be presented in an understandable way, typically using animation. Ideally, domain experts should be involved in construction and validation of the model. Finally, the whole exercise must be completed within a useful time-frame, or it will serve little purpose. Before embarking on model development, it may be prudent to look for a suitable, off-the-shelf, domainspecific simulator. Frequently, however, unique characteristics or fidelity requirements dictate the need for a custom model. The characteristics of simulation modeling, then, make some special demands of the model development environment. The modeler needs help in the conceptualization of systems with dynamic, interacting, probabilistic behaviors. The model should be understandable, maintainable, and should lend itself to incremental addition of detail. The environment should support graphical interaction for scenario development and animated model execution. This tutorial will introduce the benefits of using an object oriented language specially developed for simulation, and enhanced with comprehensive graphics support. MODSIM III combines CACI's experience with simulation programming over three decades with advances in software engineering to offer the most productive environment for the development of large, complex, evolutionary, custom models. Examples of MODSIM III's simulation features, and the benefits of object-oriented architecture, are demonstrated below using code fragments from a hypothesized airport/airspace planning model. Such a model might be concerned with the representation of aircraft, flight duration, air traffic controllers, runway allocation procedures, and so on. To be of any interest, such a model must represent multiple aircraft in flight concurrently, and delays due to contending requests for resources, such as runways. 2 DEFINITION BLOCK-AN INTERFACE DESCRIPTION In support of modular program construction, objects in MODSIM III are described in two separate blocks of code. The Definition block describes the object type by declaring its variables and methods. This is the object description as it will be referred to by other objects in

Computer-based simulation modelling is one of the domains that is particularly demanding in terms of user interfaces. Issues that influence the 'usability' of such systems are examined. Several representative systems were investigated in order to generate some general assumptions with respect to those characteristics of user interfaces employed in simulation systems. There is a need for simulation systems that can support the developments of simulation models in many domains, which are not supported by contemporary simulation software. Many user interface deficiencies are discovered and reported. Proposals are made on how user interfaces for simulation systems can be enhanced to match better the needs specific to the domain of simulation modelling, and on how better to support users in simulation model developments.

Simulation is used for many purposes: for example, to analyse a complex system, to visualise the functioning of a system, and to optimise or tune a system. While there is no limitation on the use of simulation, the general consensus is that an analytical solution, if one is possible, is always to be preferred to simulation as a methodology. In the field of information systems, client-server models exhibit a degree of complexity and richness not amenable to easy analytical solutions, except for some specific algorithms useful in limited contexts. Simulation could, therefore, be a good strategy to analyse the client-server systems and help in better implementation of feasible solutions. This paper examines the current state of client-server models and use of simulation in dealing with the problems encountered. The paper then compares the seven-layer OSI model for communications architecture and recommends that a similarly layered approach is likely to prove useful in simulating client-server systems. In the process, the paper also points out that the simulation models bring into a sharp focus the importance of software metrics, an area of vital importance in software development.

Due to the high complexity of semiconductor manufacturing systems simulation has become a useful tool for the performance analysis of these systems. However, there are some serious problems in analyzing simulation output data. For example, the quality of confidence intervals for certain performance measures is affected by correlations among the data. Furthermore, analyzing the correlation of simulation output data might give valuable hints for understanding the behavior of complex systems such as semiconductor manufacturing systems. In this paper we present examples where we have detected remarkable correlation phenomena, in the simulation output of semiconductor fabrication facilities caused by exo- or endogenous factors. Based on real specifications we examine several semiconductor manufacturing systems in steady state. First we calculate the autocorrelation function of the sequence of successive cycle times and then compute the corresponding frequency spectrum. By this means, periodicities in the simulation output data can efficiently be determined.

The fail-stop failure model appears frequently in the distributed systems literature. However, in an asynchronous distributed system, the fail-stop model cannot be implemented. In particular, it is impossible to reliably detect crash failures in an asynchronous system. In this paper, we show that it is possible to specify and implement a failure model that is indistinguishable from the fail-stop model from the point of view of any process within an asynchronous system. We give necessary conditions for a failure model to be indistinguishable from the fail-stop model, and derive lower bounds on the amount of process replication needed to implement such a failure model. We present a simple one-round protocol for implementing one such failure model, which we call simulated fail-stop. >

This paper provides a language-independent introduction to the major simulation modeling world views for discrete-event systems simulation. A world view is the modeling framework that a modeler uses to represent a system and its behavior. The main terminology and concepts include systems and models, system state variables, entities and their attributes, lists, resources, events, activities and delays. These concepts are covered from the perspective of the modeler. Finally, in the tutorial, we will attempt to make these ideas concrete through the use of a number of examples and exercises.

There is increasing focus on the model development aspect of systems simulation, termed "Model Engineering". Current software development tools consist primarily of CASE tools with simulation extensions or simulation tools with model-building extensions. The need for a Computer-Aided Simulation Model Engineering (CASME) environment is identified and discussed. In particular, a CASME environment should support a model building process based on the interconnection of components which are either primitives or collections of primitives (model modules) to form other model modules. The features and tools comprising a CASME environment which support such a process and provide model specifications for use in simulation execution environments are also identified and discussed. The conclusion is that such a tool would be useful, but only insofar as simulation environments provide for implementation of model and measurement (experimental frame) specifications.

The availability of low-cost digital communication services has made it productive for computer users to remotely access multiple computer resources. This is a particularly sound strategy for accessing complex resources that are infeasible to relocate. At the same time, the availability of low-cost and reliable minicomputer systems which can accommodate sizable numbers of terminals has enabled a shift of some interactive programming and information processing to these small, standalone systems. Although these small systems are powerful and can be configured flexibly, there are a number of opportunities to expand the scope of an individual user by relating the small system directly to other host computers via a network. In accomplishing this feat, however, there are a bewildering number of technical and economic alternatives.This paper addresses one of the issues surrounding the development of such systems namely the relative performance and cost characteristics of a minicomputer system in the role of both a standalone processor and a network interface. The methodology for the evaluation is stochastic simulation. The basic measurements for the base line simulations and model validation is a DEC PDP-11/45 computer system operating under the UNIX operating system in the Science Center at Harvard University. The paper describes the simulation methodology and the individual experiments in some detail. Several conclusions are drawn about the performance and cost characteristics of such systems and also about the use of simulation to evaluate these characteristics.

This paper describes Nest, a graphical environment for distributed networked systems simulation and rapid-prototyping. Nest users can develop and test distributed systems and protocols (from crude models to actual system code) within simulated network scenarios. Nest represents an environment-based approach to simulation. Users view Nest as an extension of their standard UnixTM environment. Nest offers the generality of language-based simulation techniques and the efficiencies of model-based techniques. Users interact with Nest through standardized graphical interfaces. Nest permits the users to modify and reconfigure the simulation during execution. Thus, it is possible to study the dynamic response of a distributed system to failures or burst-loads. Nest is organized as a simulation server, responsible for execution of complex simulation scenarios, and client monitors responsible for simulation control. The client/server model permits distribution of Nest over a network environment. This permits migration of simulations to powerful remote computational servers as well as development of a shared multi-site simulation/integration testbed. Nest is portable and extensible. It has been ported to virtually all UnixTM variants and distributed since 1987 to over 150 sites worldwide. It has been used in scores of studies ranging from communication protocols, to distributed databases and operating systems as well as distributed manufacturing systems.

This paper describes the significant positive impact that the Ada® programming language can have on the development and application of simulation models. The Ada language has specific features which are shown to be ideally suited for implementing the unique requirements of discrete-event simulation, including such simulation functions as list processing, dynamic memory management, and event calendar management. Experiences and conclusions are described based upon the development of a general purpose discrete-event simulation package. The central issues of discrete-event simulation are addressed by describing the implementation of key simulation requirements in the package and indicating how Ada language capabilities drove the design of these features. Examples of design alternatives of general interest to both simulation modelers and Ada developers are presented. In addition, issues related to very large simulation models such as long model running times and verification of complex models are specifically addressed.

Parallel discrete event simulation (PDES), sometimes called distributed simulation, refers to the execution of a single discrete event simulation program on a parallel computer. PDES has attracted a considerable amount of interest in recent years. From a pragmatic standpoint, this interest arises from the fact that large simulations in engineering, computer science, economics, and military applications, to mention a few, consume enormous amounts of time on sequential machines. From an academic point of view, parallel simulation is interesting because it represents a problem domain that often contains substantial amounts of parallelism (e.g., see [59]), yet paradoxically, is surprisingly difficult to parallelize in practice. A sufficiently general solution to the PDES problem may lead to new insights in parallel computation as a whole. Historically, the irregular, data-dependent nature of PDES programs has identified it as an application where vectorization techniques using supercomputer hardware provide little benefit [14]. A discrete event simulation model assumes the system being simulated only changes state at discrete points in simulated time. The simulation model jumps from one state to another upon the occurrence of an event. For example, a simulator of a store-and-forward communication network might include state variables to indicate the length of message queues, the status of communication links (busy or idle), etc. Typical events might include arrival of a message at some node in the network, forwarding a message to another network node, component failures, etc. We are especially concerned with the simulation of asynchronous systems where events are not synchronized by a global clock, but rather, occur at irregular time intervals. For these systems, few simulator events occur at any single point in simulated time; therefore parallelization techniques based on lock-step execution using a global simulation clock perform poorly or require assumptions in the timing model that may compromise the fidelity of the simulation. Concurrent execution of events at different points in simulated time is required, but as we shall soon see, this introduces interesting synchronization problems that are at the heart of the PDES problem. This article deals with the execution of a simulation program on a parallel computer by decomposing the simulation application into a set of concurrently executing processes. For completeness, we conclude this section by mentioning other approaches to exploiting parallelism in simulation problems. Comfort and Shepard et al. have proposed using dedicated functional units to implement specific sequential simulation functions, (e.g., event list manipulation and random number generation [20, 23, 47]). This method can provide only a limited amount of speedup, however. Zhang, Zeigler, and Concepcion use the hierarchical decomposition of the simulation model to allow an event consisting of several subevents to be processed concurrently [21, 98]. A third alternative is to execute independent, sequential simulation programs on different processors [11, 39]. This replicated trials approach is useful if the simulation is largely stochastic and one is performing long simulation runs to reduce variance, or if one is attempting to simulate a specific simulation problem across a large number of different parameter settings. However, one drawback with this approach is that each processor must contain sufficient memory to hold the entire simulation. Furthermore, this approach is less suitable in a design environment where results of one experiment are used to determine the experiment that should be performed next because one must wait for a sequential execution to be completed before results are obtained.

This tutorial paper provides information relevant to the selection and generation of stochastic inputs to simulation studies. The primary area considered is multivariate but much of the philosophy at least is relevant to univariate inputs as well.

Owing to the complexity and enormous cost involved in Automated Material Handling Systems, there is a growing need to use computer simulation in both the physical system and control software design of such systems. Simulation models can be developed to test not only the final system configuration, but also each installation phase. This paper describes the development of an interactive and flexible simulation software for Automated Storage/Retrieval Systems( AS/RS) of the Miniload variety. A modular approach is taken in the development of the simulation software so that the user, through an interactive capability to model an menu, has the AS/RS by selecting a combination of modules that define the AS/RS. The user then enters the values of the system variables specified for each module. This user-defined simulation model is translated into a simulation language source code and then executed.

The AUTASIM System (Automated Assembly of Simulation Models) is described. The system is designed to rapidly assemble discrete event, stochastic simulation models of node network systems. Systems of this type are distribution systems, plant operations, systems for the provision of services, transportation systems, and combinations of these. Models can be used for comparative analysis of new concepts and proposed systems with other systems.The main AUTASIM components are a Module Library of computer programs, a program called the Model Assembler and a Model Description Language. The Library contains functional and simulation service modules which are the “building blocks” for the simulation models created. The Model Assembler program reads a coded model description, selects the required modules from the Library, and creates the necessary linkage routines for a complete model program.The Model Description Language, AUTASCRIPT, is defined. The operation of the FORTRAN coded Model Assembler and linkage control programs is described, as is the GASP based model simulation control.

One of the very first applications of digital computers was that of simulation. Perhaps more computer time has been used over the years in this area than any other. Many programs are responsible for the largest computers in existence grinding away, day in and day out, in this general area. This paper will cover simulation as it applies to the design and development of Very Large Scale Integrated (VLSI) circuits. These techniques cover broadly the areas of process and circuit simulation, logic and timing simulation (to include faulted machine performance), and system simulation or simulation at the major block level, sometimes referred to the register transfer level or RTL simulation. This paper will present numerous references so that the reader can pursue further study as needed. Naturally only a summary of each of the major techniques can be presented. It is again noted that this paper covers only the restricted area of simulation as it applies to the development of VLSI although many topics have application far beyond this.

The use of digital computers as simulation devices for analysis and synthesis of discrete flow systems is an increasingly important area of computer applications technology. Discrete flow systems are characterized by channeled movement of items according to basic patterns or constraints which are modified by events and elapsed time. In general, mathematical models are not developed for such systems in order that computer simulation be effected; it is more efficient to formulate the simulation flow chart directly from knowledge of the physical system. In this sense, and also in the programming techniques employed, simulation of discrete flow systems is accomplished by a different methodology than is commonly used in the simulation of engineering and scientific problems of the continuous flow type.Two discrete flow problems were programmed for a medium-speed digital computer (IBM 650). One of the models was based on a straightforward time-sequenced, grid-representation of the physical system, and the other, while also time sequenced, used an event-oriented representation of the system.

The analytical complexity of most automobile traffic intersections strongly directs the researcher to simulation as his tool of analysis. This tool, through the General Purpose Systems Simulation language (GPSS), was used in a recent investigation of two different stop sign controlled intersections. On the basis of a comprehensive literature search, the authors feel that the simulation models reflect, as nearly as is presently possible, the “real-life” situations. This paper, an out-growth of that investigation, includes discussions of:(1) basic assumptions of the models,(2) empirical probability distributions employed,(3) comparisons for a selected objective function,(4) comparisons of the models with existing literature, and(5) conclusions and recommendations for further research.

Three models are presented in which stochastic simulation models are used to construct simulations of systems composed of unreliable components. The releability of the resultant systems is inferred from behavior of these simulation models.

The conceptual and construction procedure planning process for modern major resource development projects represents a challenging problem in scheduling, logistics and resource allocation. The large capital investment, the magnitude of many of the projects, the impact of minor decisions, and the fact that many resource extractions are located in remote regions of the world combine to present a potentially high degree of risk to management. This paper discusses why discrete systems simulation can be a valuable planning tool to aid management in preparing for the real world contingencies which inevitably occur when major projects move into the field.

In recent years, computer simulation has come to play an important role in the design of communication systems. Such systems frequently cannot be classified as either purely continuous or as purely discrete but are, instead, a combination of both continuous and discrete subsystems. The simulation of such hybrid systems cannot readily be carried out either in a discrete simulation language such as GPSS or in a continuous simulation language such as CSMP. As a consequence, most simulation models of communication systems are written in some general purpose higher level language, e.g., FORTRAN, ALGOL or PL-1. A notable exception to this general rule is BLODI and BLODIB described in Reference 4. The simulation of even moderately complicated communication systems in a general purpose language poses several major problems. First of all, the simulation model tends to be time consuming to write and debug, difficult to modify once written, and, unless considerable care is taken, requires an inordinate amount of time to execute. However, the most important weakness of a communication system simulation model written in a higher level language is that the model usually is written by someone with a limited knowledge of the actual system being simulated. As a consequence, there frequently exists the possibility that the simulation model differs in some significant, but unnoticed, respect from the actual system. Primarily to avoid this problem, a new simulation language, COMSL, is described which facilitates the simulation of a wide variety of communication systems.

Recent years have seen a very rapid growth in the use of digital computers for simulation work, particularly in the field of system studies. The need for such simulation has been generated by the ever-increasing complexity of systems that are being designed, while the speed and capacity of modern digital computers have provided the means by which to expand simulation efforts. The amount of simulation and the complexity of the studies that are being conducted have presented several difficulties in organizing the use of simulation.

A computer simulation model was used as an analysis “tool” for computer system design trade-offs for an on-line interactive command and control system preliminary design study project.Three basic hardware configurations were modelled at the hardware interrupt/byte flow level:a. A Centralized Dual Multiprocessorb. Dual Computersc. A Distributed System of Central and Remote ComputersThe software of the system was modelled in several modules:a. The Operating System Routinesb. The Data Base Management Routinesc. Interactive File Maintenance and Query Routinesd. Object-Coded Functional Applications Programse. Support and Control SoftwareEach module consists of parametric descriptions for each corresponding loadable software module (e.g., load module size, re-enterability) and procedural descriptions (e.g., read/write statements, processing statements, calls to system macros and other subroutines). The simulation is conducted in two modes: A fast-forward mode to load-up the system, and an observation mode to collect detailed data. A history tape of the environmental stimuli (interactive operator loads) on the system, produced by a separate “loading model”, enters tasks into the system. The system model is implemented on the IBM 360, Model 65 Computer using a high level computer systems simulation language. The technique developed was so successful that it is now being actively used as a design tool and system performance evaluator, providing a means to minimize the technical risk in the development of all of Litton DSD's major software projects.

A major problem in the area of discrete time-step simulation is the precise, rapid, and fully automatic scheduling and simulation of hundreds or thousands of serial and parallel events. A partial solution to this problem has been found in the “next event” simulation technique 1 which, based upon inter-related concepts of list processing and dynamic storage allocation, has replaced “fixed time step” methods 1 of the past. However, this solution is not satisfactory for large systems because the next event technique, particularly useful for handling a moderate number of serial events, decreases rapidly in scheduling efficiency for large numbers of events and in simulation efficiency for parallel events. This paper extends and generalizes work reported previously and describes a technique originally developed for digital simulation 2 of digital systems which is applicable to general-purpose simulation. The technique was specifically developed for simulating large numbers of parallel events. It is based on “time mapping,&rdquo i.e., a simple, time-to-address mapping method, and falls in the category of fixed time step simulation. On a second level, the advantages of next-event simulation are exploited by using this technique in auxiliary roles. The result of combining these techniques leads to an optimization of simulation speed, which becomes significant when the average event queue exceeds 20 events. Also, the simulation of any number of parallel events is possible without the usual amount of additional complexity and bookkeeping.

Until recently, systems simulation problems could generally be split into two classes: problems requiring only an analog computer for solution and problems requiring only a digital computer for solution. Any general problem has a number of characteristics which adapt it to either one or the other method of solution.