Neural network metamodeling for cycle time-throughput profiles in manufacturing

This paper proposed a neural network (NN) metamodeling method to generate the cycle time (CT)–throughput (TH) profiles for single/multi-product manufacturing environments. Such CT–TH profiles illustrate the trade-off relationship between CT and TH, the two critical performance measures, and hence provide a comprehensive performance evaluation of a manufacturing system. The proposed methods distinct from the existing NN metamodeling work in three major aspects: First, instead of treating an NN as a black box, the geometry of NN is examined and utilized; second, a progressive model-fitting strategy is developed to obtain the simplest-structured NN that is adequate to capture the CT–TH relationship; third, an experiment design method, particularly suitable to NN modeling, is developed to sequentially collect simulation data for the efficient estimation of the NN models.     

Massively time-parallel,approximate simulation of loss queueing systems

A time-parallel simulation obtains parallelism by partitioning the time domain of the simulation. An approximate time-parallel simulation algorithm named GG1K is developed for acyclic networks of loss FCFSG/G/1/K queues. The GG1K algorithm requires two phases. In the first phase, a similar system (i.e. aG/G/1/ queue) is simulated using the GLM algorithm. Then the resultant trajectory is transformed into an approximateG/G/1/K trajectory in the second phase. The closeness of the approximation is investigated theoretically and experimentally. Our results show that the approximation is highly accurate except whenK is very small (e.g. 5) in certain models. The algorithm exploits unbounded parallelism and can achieve near-linear speedup when the number of arrivals simulated is sufficiently large.     

Robustness of perturbation analysis estimators for queueing systems with unknown distributions

Sample-path-based stochastic gradient estimators for performance measures of queueing systems rely on the assumption that a probability distribution of the random vector of interest (e.g., a service or interarrival time sequence) is given. In this paper, we address the issue of dealing with unknown probability distributions and investigate the robustness of such estimators with respect to possibly erroneous distribution choices. We show that infinitesimal perturbation analysis (IPA) can be robust in this sense and, in some cases, provides distribution-independent estimates. Comparisons with other gradient estimators are provided, including experimental results. We also show that finite perturbation analysis (FPA), though only providing gradient approximations, possesses some attractive robustness properties with respect to unknown distribution parameters. An application of FPA estimation is included for a queueing system performance optimization problem involving customers with real-time constraints.This work was supported in part by the National Science Foundation Grant ECS-88-01912 and by the Office of Naval Research Contract N00014-87-K-0304.The authors wish to thank Dr. Jack Holtzman for several useful comments and suggestions.     

Analysis of a queueing system with a general service scheduling function,with applications to telecommunication network traffic control

In this paper we analyze a queueing system with a general service scheduling function. There are two types of customer with different service requirements. The service order for customers of each type is determined by the service scheduling function α_k(i, j) where α_k(i, j) is the probability for type-k customer to be selected when there are i type-1 and j type-2 customers. This model is motivated by traffic control to support traffic streams with different traffic characteristics in telecommunication networks (in particular, ATM networks). By using the embedded Markov chain and supplementary variable methods, we obtain the queue-length distribution as well as the loss probability and the mean waiting time for each type of customer. We also apply our model to traffic control to support diverse traffics in telecommunication networks. Finally, the performance measures of the existing diverse scheduling policies are compared. We expect to help the system designers select appropriate scheduling policy for their systems.     

Performance evaluation for general queueing networks in manufacturing systems: Characterizing the trade-off between queue time and utilization

Performance evaluation plays a key role in manufacturing system design and productivity improvement. Characterizing performance objectively is the first step. Inspired by the underlying structure of tandem queues, we have derived an approximate model to characterize the system performance. The model decomposes system queue time and variability into bottleneck and non-bottleneck parts while capturing the dependence among workstations. Compared the new model with prior approaches, the new model not only is more accurate but also requires less information. The property of manufacturing system performance is given based on the insight from the model.     

Simulation methodology — An introduction for queueing theorists

Simulation is a widely used methodology for queueing systems. Its superficial simplicity hides a number of pitfalls which are not all as well known as they should be. In particular simulation experiments need careful design and analysis as well as good presentations of the results. Even the elements of simulation such as the generation of arrival and service times have a chequered history with major problems lying undiscovered for 20 years. On the other hand, good simulation practice can offer much more than is commonly realized.     

An application of interactive analysis and computer graphics in stochastic modeling

This paper will describe the application of an interactive queueing network analyzer and an interactive graphics system to the analysis of a multiple processor computer system. The application of these tools greatly increased the productivity of the modelers and resulted in insights which would have otherwise been difficult, if not impossible, to obtain. With this experience as background, we discuss how the increasing availability of computing resources, especially high resolution interactive computer graphics and sophisticated modeling packages, is likely to have a profound influence on the applied stochastic modeler.     

Relative priority policies for minimizing the cost of queueing systems with service discrimination

This paper considers several single-server two-class queueing systems with different cost functions. Customers in the two classes are discriminated by service rates and relative priorities. Most attention is focused on the ones with general quadratic bivariable and exponential cost functions that are usually applied in the relatively complicated systems. To the best of the authors’ knowledge, there is no literature analyzing these two kinds of cost functions on the subject of relative priority. We explicitly present the conditions under which relative priority outperforms absolute priority for reducing system cost and further provide the method to find the optimal DPS policy. Moreover, we also discuss variations where service rates of the two classes are decision variables under service equalization and service discrimination disciplines, respectively.     

A survey of the perturbation analysis of discrete event dynamic systems

This paper introduces an analysis and optimization technique for discrete event dynamic systems, such as flexible manufacturing systems (FMSs), and other discrete part production processes. It can also be used for enhancement of the simulation results of, or the monitoring of the operations of such systems in real time. Extensive references are given where readers may pursue futher details.     

Importance functions for restart simulation of general Jackson networks

RESTART is an accelerated simulation technique that allows the evaluation of extremely low probabilities. In this method a number of simulation retrials are performed when the process enters regions of the state space where the chance of occurrence of the rare event is higher. These regions are defined by means of a function of the system state called the importance function. Guidelines for obtaining suitable importance functions and formulas for the importance function of two-stage networks were provided in previous papers. In this paper, we obtain effective importance functions for RESTART simulation of Jackson networks where the rare set is defined as the number of customers in a particular (‘target’) node exceeding a predefined threshold. Although some rough approximations and assumptions are used to derive the formulas of the importance functions, they are good enough to estimate accurately very low probabilities for different network topologies within short computational time.     

n-Order linear state space systems: Computing the transfer function using the DFT

The discrete Fourier transform is used to determine the coefficients of a transfer function for n-order linear systems: . The algorithm is fast, straight forward and easily can be implemented. Two step-by-step examples, illustrating the application of the algorithm, are presented.     

Heavy traffic limit theorems for a sequence of shortest queueing systems

A sequence of shortest queueing systems is considered in this paper. We give weak convergence theorems for the queue length and waiting time processes when the sequence of traffic intensities associated with the sequence of shortest queueing systems approaches the critical value (=1) at appropriate rates.Research supported by the National Natural Science Foundation of China.     

Applied general equilibrium modeling for environmental policy analysis

Incorporating environmental regulation and its impact in general equilibrium models has by now a twenty years long tradition. In 1970, Leontief extended his input-output approach to quantify environmental repercussions on the economic structure. The shortcomings of this model, such as fixed coefficients in technological processes or zero price and income elasticities of demand, can be resolved by using a quantitative general equilibrium model. The objective of this paper is to employ such a model to quantify the economic effects of environmental policy and to present a cost-effectiveness analysis of different market instruments in environmental policy.     

Infinitesimal perturbation analysis for queueing networks with general service time distributions

We study infinitesimal perturbation analysis (IPA) for queueing networks with general service time distributions. By general we mean that the distributions may have discrete components. We show that in the presence of service time distributions with discrete components commuting condition (CC) is no longer sufficient for unbiasedness of IPA. To overcome this difficulty, we introduce the notion of separability of realvalued random variables, and show that separability of service times together with (CC) establishes unbiasedness of IPA for queueing systems with general service time distributions. It turns out that the piecewise analyticity of service times is a sufficient condition for separability.     

Boundary value methods for transient solutions of queueing networks with variant vacation policy

Boundary value methods are applied to find transient solutions of M/M/2$M/M/2$ queueing systems with two heterogeneous servers under a variant vacation policy. An iterative method is employed to solve the resulting large linear system and a Crank–Nicolson preconditioner is used to accelerate the convergence. Numerical results are presented to demonstrate the efficiency of the proposed method.     

Statistical analysis of queueing systems

This paper provides an overview of the literature on statistical analysis of queueing systems. Topics discussed include: model identification, estimation, hypothesis testing and other related aspects. Not all of these statistical problems are covered in books on queueing theory or stochastic processes. The bibliography is not exhaustive, but comprehensive enough to provide sources from the literature.     

A heuristic technique for the transient behavior of Markovian queueing systems

Through an examination of numerical solutions to Markovian queueing systems, it has been shown that the expected queue length eventually approaches its equilibrium value in an approximately exponential manner. Based on this observation a heuristic is proposed for approximating the transient expected queue length for Markovian systems by scaling the numerical solution of an M/M/1 system.     

Sample-path analysis of general arrival queueing systems with constant amount of work for all customers

We consider a discrete-time queueing system where the arrival process is general and each arriving customer brings in a constant amount of work which is processed at a deterministic rate. We carry out a sample-path analysis to derive an exact relation between the set of system size values and the set of waiting time values over a busy period of a given sample path. This sample-path relation is then applied to a discrete-time $G/D/c$ queue with constant service times of one slot, yielding a sample-path version of the steady-state distributional relation between system size and waiting time as derived earlier in the literature. The sample-path analysis of the discrete-time system is further extended to the continuous-time counterpart, resulting in a similar sample-path relation in continuous time.     

Efficient space-filling and non-collapsing sequential design strategies for simulation-based modeling

Simulated computer experiments have become a viable cost-effective alternative for controlled real-life experiments. However, the simulation of complex systems with multiple input and output parameters can be a very time-consuming process. Many of these high-fidelity simulators need minutes, hours or even days to perform one simulation. The goal of global surrogate modeling is to create an approximation model that mimics the original simulator, based on a limited number of expensive simulations, but can be evaluated much faster. The set of simulations performed to create this model is called the experimental design. Traditionally, one-shot designs such as the Latin hypercube and factorial design are used, and all simulations are performed before the first model is built. In order to reduce the number of simulations needed to achieve the desired accuracy, sequential design methods can be employed. These methods generate the samples for the experimental design one by one, without knowing the total number of samples in advance. In this paper, the authors perform an extensive study of new and state-of-the-art space-filling sequential design methods. It is shown that the new sequential methods proposed in this paper produce results comparable to the best one-shot experimental designs available right now.