Stochastic modelling and analysis: a computational approach
Stochastic modelling and analysis: a computational approach
A guide to simulation (2nd ed.)
A guide to simulation (2nd ed.)
Parallel simulation of real-time systems via the Standard Clock approach
Mathematics and Computers in Simulation
Some guidelines and guarantees for common random numbers
Management Science
Effect of correlated estimation errors in ordinal optimization
WSC '92 Proceedings of the 24th conference on Winter simulation
Convergence properties of ordinal comparison in the simulation of discrete event dynamic systems
Journal of Optimization Theory and Applications
Universal alignment probabilities and subset selection for ordinal optimization
Journal of Optimization Theory and Applications
Rates of convergence of ordinal comparison for dependent discrete event dynamic systems
Journal of Optimization Theory and Applications
Admission-control policies for multihop wireless networks
Wireless Networks
Hi-index | 0.00 |
In the design and optimizationof discrete event dynamic systems, it is often necessary to orderalternative designs based on their relative performance, i.e.,to rank them from best to worst. In this paper, alignment ofobserved performance orders with true orders is considered andproperties of the alignment are investigated. Spearman‘s rankcorrelation coefficient is a measure of agreement between theobserved performance orders and the true ones. It is shown thatSpearman‘s coefficient converges exponentially in the simulationtime or observation time, which gives a strong evidence of theefficiency of order comparison for discrete event dynamic systems.In the context of simulation, the effect of simulation dependenceon the alignment is also discussed. It is found that neitherindependent simulation nor the scheme of common random numbers(CRN), a popular scheme for variance reduction, can yield dominantperformance. Finally, numerical examples based on a networkingoptimization problem are provided to illustrate the convergenceof Spearman‘s coefficient. In these examples, the standard clock(SC) simulation technique provides much faster convergence thaneither independent simulations or CRN simulations. Both the SCand CRN methods use the same random number sequence to drivemany events in parallel; however, under SC the events drivingthe parallel experiments are all identical, whereas under CRNthey may be different.