Modeling and optimization of buffers and servers in finite queueing networks.

Abstract

The joint buffer and server optimization problem (BCAP) is a non-linear optimization problem with integer decision variables that optimizes the numbers of buffers and servers such that the resulting throughput is greater than a pre-defined threshold throughput. This work presents a detailed review of the current literature that addresses allocation problems, particularly the BCAP, and a quite effective methodology for solving this problem, which consists of a combination of approximate methods and the Powell algorithm, a derivative-free optimization algorithm. The methodology was applied to networks of queues in the basic topologies series, split, and merge, producing very encouraging results that pointed at robust and homogeneous solutions.