Home // International Journal On Advances in Networks and Services, volume 5, numbers 1 and 2, 2012 // View article
Authors:
Nawel Gharbi
Leila Charabi
Keywords: Wireless networks; Retrial phenomenon; Heterogeneous servers; Performance indices; Service disciplines.
Abstract:
This paper proposes an algorithmic approach based on Generalized Stochastic Petri Nets, for modeling and analyzing finite-source wireless networks with retrial phenomenon and two servers classes. The particularity of this approach is the direct computing of the infinitesimal generator of the proposed Generalized Stochastic Petri Net without generating neither the reachability graph nor the underlying Markov chain. Furthermore, we assume in this model that servers of one class are faster than those of the second one. In Random Server policy, customers requests are assigned randomly to free servers of both classes. The disadvantage of this policy is the increased response time when fast servers are free and requests are assigned (randomly) to slow ones. Hence, this paper aims at presenting another service policy, where priority is given to faster free servers. This policy is called the Fastest Free Server policy. Moreover, we compare through numerical examples, Random Service policy to Fastest Free Server one, by developing formulas of the main stationary performance indices of the network. We compare also these two policies to Averaged Random case, where the same global number of servers is assumed, but all homogeneous with the average service rate. We show that Fastest Free Server discipline gives better results than both Averaged Random case and Random Server discipline.
Pages: 102 to 115
Copyright: Copyright (c) to authors, 2012. Used with permission.
Publication date: June 30, 2012
Published in: journal
ISSN: 1942-2644