• DocumentCode
    759826
  • Title

    Adaptive call admission control in circuit-switched networks

  • Author

    Gokbayrak, Kagan ; Cassandras, Christos G.

  • Author_Institution
    Dept. of Manuf. Eng., Boston Univ., Brookline, MA, USA
  • Volume
    47
  • Issue
    6
  • fYear
    2002
  • fDate
    6/1/2002 12:00:00 AM
  • Firstpage
    1004
  • Lastpage
    1015
  • Abstract
    We consider threshold-based admission control policies for traffic in fixed-route circuit-switched networks, and develop a scheme for adjusting the threshold parameters online so that, as operating conditions in the network change, the thresholds "adapt" with the objective of minimizing a weighted sum of call blocking probabilities. An algorithm for estimating online the sensitivity of the call blocking metric with respect to thresholds is presented. The formal optimization problem over the set of discrete threshold parameters is solved by means of a conversion to an optimization problem over a set of auxiliary real-valued parameters. Such threshold-based policies, though conservative at low traffic rates, have the advantage of being simple to implement, distributed in nature, adaptive, and not requiring explicit distributional modeling assumptions. Numerical results included in the paper indicate that at higher traffic rates these simple policies yield the same performance as more complex and less flexible call admission schemes
  • Keywords
    adaptive control; optimisation; perturbation techniques; sensitivity analysis; telecommunication congestion control; telecommunication networks; call admission control; call blocking; circuit-switched networks; perturbation analysis; sensitivity estimation; stochastic optimization; threshold parameters; Adaptive control; Admission control; Call admission control; Circuits; Communication system traffic control; Contracts; Intelligent networks; Programmable control; Routing; Traffic control;
  • fLanguage
    English
  • Journal_Title
    Automatic Control, IEEE Transactions on
  • Publisher
    ieee
  • ISSN
    0018-9286
  • Type

    jour

  • DOI
    10.1109/TAC.2002.1008364
  • Filename
    1008364