Importance sampling for stochastic simulations
Management Science
Optimization by mean field annealing
Advances in neural information processing systems 1
Measure specific dynamic importance sampling for availability simulations
WSC '87 Proceedings of the 19th conference on Winter simulation
Simulation and the Monte Carlo Method
Simulation and the Monte Carlo Method
An Introduction to the Regenerative Method for Simulation Analysis
An Introduction to the Regenerative Method for Simulation Analysis
IEEE/ACM Transactions on Networking (TON)
| Hi-index | 0.00 |
Importance sampling (IS) is recognised as a potentially powerful method for reducing simulation runtimes when estimating the probabilities of rare events in communication systems using Monte Carlo simulation. A dynamic application of IS combined with regenerative techniques has been shown to provide excellent simulation performance.To obtain large improvement factors in simulation runtime using IS, the modification, or bias of the underlying probability measures must be carefully chosen. We present in this paper a methodology which optimizes IS parameter settings using the mean field annealing (MFA) optimization algorithm in conjunction with statistical estimates of the IS estimator variance.We demonstrate the effectiveness of this methodology by estimating blocking probabilities for the Geo/Geo/1/K, IBP/Geo/1/K queues and a 16 × 16 synchronous Clos ATM switch. Improvement factors of three to fifteen orders of magnitude are obtained for these examples.