Stochastic Control and Numerical Methods with Applications to Communications. Game Theoretic/Subsolution to Importance Sampling for Rare Event Simulation
Abstract
Game theoretic methods were developed to greatly improve importance sampling for rare event simulation for stochastic systems and queueing networks that arise in communications. With the same goal in mind, large deviations approach to design and analysis was developed to facilitate rare event simulations for systems with discontinuities in the dynamics and analyze escape events for queueing networks. Results for heavy tailed distributions were obtained. Numerical methods were developed for the solution of non-zero sum stochastic game models. Effective stability based methods for scheduling and routing in networks of mobiles with randomly varying channels were produced and they are also useful with long term memory system.
Document Details
- Document Type
- Technical Report
- Publication Date
- Nov 01, 2008
- Accession Number
- ADA499989
Entities
People
- Harold Kushner
- Paul Dupuis
Organizations
- Brown University