Abstract
The need to allocate a finite number of buffers among several customer classes or stations in a queueing system arises in many applications. We consider both static and dynamic buffer allocation problems, with the objective of minimizing the (weighted) blocking probability. We obtain asymptotically optimal allocation policies in light traffic, that is, for an arrival rate close to zero. Light traffic optimal policies have the feature of maximizing the number of customers who can arrive to the system without any possibility of a blocking event.

This publication has 0 references indexed in Scilit: