7th International Conference on Performance Evaluation Methodologies and Tools

Research Article

Network of Queues with Inert Customers and Signals

  • @INPROCEEDINGS{10.4108/icst.valuetools.2013.254369,
        author={Jean Michel Fourneau and Thu Ha Dao Thi and Minh Anh Tran},
        title={Network of Queues with Inert Customers and Signals},
        proceedings={7th International Conference on Performance Evaluation Methodologies and Tools},
        publisher={ICST},
        proceedings_a={VALUETOOLS},
        year={2014},
        month={1},
        keywords={queuing networks signals product-form},
        doi={10.4108/icst.valuetools.2013.254369}
    }
    
  • Jean Michel Fourneau
    Thu Ha Dao Thi
    Minh Anh Tran
    Year: 2014
    Network of Queues with Inert Customers and Signals
    VALUETOOLS
    ACM
    DOI: 10.4108/icst.valuetools.2013.254369
Jean Michel Fourneau1,*, Thu Ha Dao Thi2, Minh Anh Tran3
  • 1: PRiSM, Univ. Versailles St Quentin
  • 2: PRiSM, CNRS, Univ. Versailles St-Quentin
  • 3: LACL, Univ. Creteil, France
*Contact email: jean-michel.fourneau@prism.uvsq.fr

Abstract

We introduce the concept of inert customers in a generalized queue with signals. A signal interacts with the customers when it arrives in a non empty queue. Then, it disappears instantaneously. Ordinary customers wait for service and eventually receive service if they do not interact with a signal heforehand. An inert customer does not receive service. It stays in the queue until an interaction with a signal makes it disappear. As it is present in the queue, it has an effect on the service capacity. We consider multiple classes of ordinary customers and a symmetric queueing discipline for ordinary and inert customers. In this paper we consider two types of interaction between signals and inert customers: the deletion of one inert customer and the deletion of a group of inert customers. Despite this deep modification of the model, the queue is still quasi-reversible. Therefore we prove that such a network of queues have a steady-state product form solution.
We illustrate this approach with various examples consisting in PS and LIFO queues.