Fast simulation of blocking probabilities in loss networks

This paper is concerned with the determination of blocking probabilities in loss networks. In fact, our study consists of two parts. Primarily, we scale both arrival rates and link capacities, in order to derive rough asymptotical expressions. These expressions arise as the result of mathematical programming problems. Secondly, we develop a fast simulation technique to estimate the blocking probabilities. This technique is based on importance sampling, where the choice of the alternative probability model is closely related to the optimizing arguments of the above mentioned mathematical programming problem. Some examples show that huge gain or simulation effort can be achieved.