In this paper the authors consider a multiclass stochastic network with state-dependent service rates and arrival rates that describes a bandwidth sharing mechanism as well as admission control and/or load balancing schemes. The network model is constituted by a finite set of nodes I. Each node is a single-server queuing station. An arriving customer is served by one of the nodes and then leaves the network. More precisely, I is partitioned into finitely many nonempty subsets Ik with k ∈K. Each customer has a class, which is identified by an element k of K, and a customer of class k has to be served by one of the nodes in Ik. Class k customers arrive to the network according to a Poisson process of rate λk. The different arrival processes are mutually independent. An arriving customer of class k is either routed to a node in Ik or rejected. The routing/admission policy depends on the whole state of the system at the instant of arrival. The service requirements of the customers are independent and exponentially distributed with parameter 1. In this context the authors study how to choose the routing rates to optimize a performance criterion. Leonardo Pasini
Recensione dell'articolo: ( Jonckheere, Matthieu; Mairesse, Jean - " Towards an Erlang formula for multiclass network " - Queueing Syst. 66 (2010), no.1, 53–78. )
PASINI, Leonardo
2011-01-01
Abstract
In this paper the authors consider a multiclass stochastic network with state-dependent service rates and arrival rates that describes a bandwidth sharing mechanism as well as admission control and/or load balancing schemes. The network model is constituted by a finite set of nodes I. Each node is a single-server queuing station. An arriving customer is served by one of the nodes and then leaves the network. More precisely, I is partitioned into finitely many nonempty subsets Ik with k ∈K. Each customer has a class, which is identified by an element k of K, and a customer of class k has to be served by one of the nodes in Ik. Class k customers arrive to the network according to a Poisson process of rate λk. The different arrival processes are mutually independent. An arriving customer of class k is either routed to a node in Ik or rejected. The routing/admission policy depends on the whole state of the system at the instant of arrival. The service requirements of the customers are independent and exponentially distributed with parameter 1. In this context the authors study how to choose the routing rates to optimize a performance criterion. Leonardo PasiniI documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
