Simulation run lengths to estimate blocking probabilities

We derive formulas approximating the asymptotic variance of four estimators for steady-state blocking probability in a multiserver loss system, exploiting diffusion process limits. These formulas can be used to predict simulation run lengths required to obtain desired statistical precision before the simulation has been run, which can aid in the design of simulation experiments. They also indicate that one estimator can be much better than another, depending on the loading. An indirect estimator based on estimating the mean occupancy is significantly more (less) efficient than a direct estimator for heavy (light) loads. A major concern is the way computational effort scales with system size. For all the estimators, the asymptotic variance tends to be inversely proportional to the system size, so that the computational effort (regarded as proportional to the product of the asymptotic variance and the arrival rate) does not grow as system size increases. Indeed, holding the blocking probability fixed, the computational effort with a good estimator decreases to zero as the system size increases. The asymptotic variance formulas also reveal the impact of the arrival-process and service-time variability on the statistical precision. We validate these formulas by comparing them to exact numerical results for the special case of the classical Erlang M/M/s/0 model and simulation estimates for more general G/GI/s/0 models. It is natural to delete an initial portion of the simulation run to allow the system to approach steady state when it starts out empty. For small to moderately size systems, the time to approach steady state tends to be negligible compared to the time required to obtain good estimates in steady state. However, as the system size increases, the time to approach steady state remains approximately unchanged, or even increases slightly, so that the computational effort associated with letting the system approach steady state becomes a greater portion of the overall computational effort as system size increases.

[1]  Ward Whitt,et al.  The Brownian approximation for rate-control throttles and the G/G/1/C queue , 1992, Discret. Event Dyn. Syst..

[2]  Ward Whitt,et al.  The Physics of the Mt/G/∞ Queue , 1993, Oper. Res..

[3]  Ward Whitt,et al.  On Arrivals That See Time Averages , 1990, Oper. Res..

[4]  P. Billingsley,et al.  Convergence of Probability Measures , 1969 .

[5]  W. Whitt.  Comparing counting processes and queues , 1981, Advances in Applied Probability.

[6]  W. Whitt,et al.  Heavy-traffic approximations for service systems with blocking , 1984, AT&T Bell Laboratories Technical Journal.

[7]  S⊘ren Asmussen,et al.  Validating the heavy traffic performance of regenerative simulation , 1989 .

[8]  Ronald W. Wolff,et al.  Poisson Arrivals See Time Averages , 1982, Oper. Res..

[9]  G. M. Jenkins,et al.  Stochastic Service Systems. , 1964 .

[10]  G. F. Newell,et al.  Introduction to the Theory of Queues. , 1963 .

[11]  Ruth J. Williams.  Asymptotic variance parameters for the boundary local times of reflected Brownian motion on a compact interval , 1992 .

[12]  Marcel F. Neuts,et al.  Structured Stochastic Matrices of M/G/1 Type and Their Applications , 1989 .

[13]  Sally Floyd,et al.  Wide-area traffic: the failure of Poisson modeling , 1994 .

[14]  Stephen S. Lavenberg,et al.  Statistical Results on Control Variables with Application to Queueing Network Simulation , 1982, Oper. Res..

[15]  Keith W. Ross,et al.  Multiservice Loss Models for Broadband Telecommunication Networks , 1997 .

[16]  A. Borovkov.  On limit laws for service processes in multi-channel systems , 1967 .

[17]  R. R. P. Jackson,et al.  Introduction to the Theory of Queues , 1963 .

[18]  David M. Lucantoni,et al.  The BMAP/G/1 QUEUE: A Tutorial , 1993, Performance/SIGMETRICS Tutorials.

[19]  W. Whitt.  On the heavy-traffic limit theorem for GI/G/∞ queues , 1982 .

[20]  D. Jagerman.  Some properties of the erlang loss function , 1974 .

[21]  Aleksandr Alekseevich Borovkov,et al.  Stochastic processes in queueing theory , 1976 .

[22]  Ward Whitt,et al.  A review ofL=λW and extensions , 1991, Queueing Syst. Theory Appl..

[23]  Ward Whitt,et al.  Indirect Estimation Via L = λW , 1989, Oper. Res..

[24]  P. Glynn.  On the Markov property of the GI/G/∞ Gaussian limit , 1982, Advances in Applied Probability.

[25]  W. Whitt.  Planning queueing simulations , 1989 .

[26]  Winfried K. Grassmann,et al.  The asymptotic variance of a time average in a birth-death process , 1987 .

[27]  M. David.  The BMAP / G / l QUEUE : A Tutorial * , 2022 .

[28]  Ward Whitt,et al.  Estimating Customer and Time Averages , 1993, Oper. Res..

[29]  Matthew J. Sobel,et al.  Note-Simple Inequalities for Multiserver Queues , 1980 .

[30]  Debasis Mitra,et al.  The transient behavior in erlang's model for large trunk groups and various traffic conditions , 1987 .

[31]  Ward Whitt,et al.  Asymptotic Formulas for Markov Processes with Applications to Simulation , 1992, Oper. Res..

[32]  Averill M. Law.  Efficient estimators for simulated queueing systems , 1974 .

[33]  Sheldon M. Ross.  Introduction to Probability Models. , 1995 .

[34]  Sheldon M. Ross,et al.  Introduction to Probability Models, Eighth Edition , 1972 .

[35]  David R. Cox,et al.  The Theory of Stochastic Processes , 1967, The Mathematical Gazette.

[36]  Ward Whitt,et al.  Networks of infinite-server queues with nonstationary Poisson input , 1993, Queueing Syst. Theory Appl..

[37]  Ward Whitt,et al.  Measurements and approximations to describe the offered traffic and predict the average workload in a single-server queue , 1989, Proc. IEEE.

[38]  Walter Willinger,et al.  Applications of Fractals in Engineering for Realistic Traffic Processes , 1994 .

[39]  Averill M. Law,et al.  Conservation equations and variance reduction in queueing simulations , 1977, WSC '77.

[40]  Ward Whitt,et al.  Heavy-Traffic Limits for Queues with Many Exponential Servers , 1981, Oper. Res..

[41]  François Baccelli,et al.  Elements Of Queueing Theory , 1994 .

[42]  Søren Asmussen,et al.  Queueing Simulation in Heavy Traffic , 1992, Math. Oper. Res..

[43]  Ward Whitt,et al.  The Asymptotic Efficiency of Simulation Estimators , 1992, Oper. Res..

[44]  Lajos Takcas.  Introduction to the Theory of Queues , 1962 .

[45]  S. Zachary,et al.  Loss networks , 2009, 0903.0640.

[46]  Aleksandr Alekseevich Borovkov,et al.  Asymptotic methods in queuing theory , 1984 .

[47]  P. Glynn,et al.  A new view of the heavy-traffic limit theorem for infinite-server queues , 1991, Advances in Applied Probability.

[48]  V. Benes.  The covariance function of a simple trunk group, with applications to traffic measurement , 1961 .

[49]  Ward Whitt,et al.  Sensitivity to the Service-Time Distribution in the Nonstationary Erlang Loss Model , 1995 .

[50]  Keith W. Ross,et al.  Monte Carlo Summation Applied to Product-Form Loss Networks , 1992, Probability in the Engineering and Informational Sciences.

[51]  Walter Willinger,et al.  Traffic modeling for high-speed networks: theory versus practice , 1995 .

[52]  W. Whitt.  The efficiency of one long run versus independent replication in steady-state simulation , 1991 .

[53]  Sheldon M. Ross,et al.  Introduction to probability models , 1975 .