Dynamic Programming Based Link Cost Function and Load Estimation Method for ATM Links

Applying Routing Intelligence in ATM Networks
Part of the IFIP — The International Federation for Information Processing book series (IFIPAICT, volume 50)


In connection-oriented networks dynamic routing schemes are designed to improve the call blocking performance by introducing network intelligence. An accurate model describing a networks statistical behaviour is the network Markov model. Based on this model the theory of dynamic programming for Markov processes (also called the Markov Decision Process theory, MDP) has in the past successfully been applied in the field of dynamic routing. The Forward Looking Routing (FLR) scheme [1] defines a dynamic link cost function for single-rate circuit-switched networks. In ATM networks carrying multirate traffic a more complex traffic theory is encountered than for single-rate traffic. Applications of the MDP concept for routing multirate traffic are made for example in [2], [3], [4] and [5]. In this work, a state dependent link cost function for multirate links is derived, based on the scalar link state model introduced in [5]. It is shown that for the single rate traffic model, this link cost function becomes equal to the FLR link cost function. An approximation method is proposed for determining the offered link load in weak mesh networks. Furthermore, a scheme for estimating the temporal evolution of the carried link load is developed. It is shown by simulation how the ATM link cost function and load estimation may be used within the frame of a dynamic routing scheme called DR/ATM. The performance of the DR/ATM scheme in simulation runs involving different network topologies is presented. Moreover the performance of the ATM link cost function in combination with link state flooding is shown. The simulation runs show that the blocking performance is substantially improved as compared with standard reference routing schemes.

Key words

Dynamic Routing Dynamic Programming ATM 


  1. [1]
    K. R. Krishnan and T. J. Ott: „Forward-Looking Routing: A New State-Dependent Routing Scheme“, ITC-12, pp. 1026–1031, 1989.Google Scholar
  2. [2]
    R. Hwang, J. Kurose and D.Towsley: „State Dependent Routing for Multirate Loss Networks“, IEEE Globecom `92, vol.’, pp. 565–570, Orlando, FL, USA, Dec. 1992.Google Scholar
  3. [3]
    A. Kolarov, J. Hui: „On Computing Markov Decision Theory-Based Cost for Routing in Circuit-Switched Broadband Networks “, Journal of Network and Systems Management, Vol. 3, No. 4, 1995.Google Scholar
  4. [4]
    Z. Dziong and L. Mason: “An Analysis of Near Optimal Call Admission and Routing Model for Multi-Service Loss Networks”, IEEE Infocom ‘82, 1992, Vol. 2, pp. 142–152.Google Scholar
  5. [5]
    K.R. Krishnan and F. Huebner-Szabo de Bucs:“Admission Control for Multirate Circuit-Switched Traffic”, ITC-15, 1997.Google Scholar
  6. [6]
    L. Bella, F. Chummun, M. Conte, G. Fischer, and J. Rammer,.: “Performance Evaluation of Dynamic Routing Based on the Use of Satellites and Intelligent Networks”, in Wireless Networks 4, 167–180 (1998).Google Scholar
  7. [7]
    H. Akimaru and K. Kawashima, Teletraffic Theory and Applications, Telecommunication Networks and Computer Systems, Springer-Verlag, 1993.CrossRefGoogle Scholar
  8. [8]
    R. B. Cooper, Introduction to Queueing Theory, North Holland, 1972.Google Scholar
  9. [9]
    J. S. Kaufman, “Blocking in a Shared Resource Environment”, IEEE Transaction on Communications, COM-29, Nr. 10, pp. 1474–1481, October 1981.CrossRefGoogle Scholar
  10. [10]
    R. A. Howard: Dynamische Programmierung and Markov-Prozesse, (translation) 1965.Google Scholar
  11. [11]
    M. Conte, J. Rammer, G.Fischer, N. Mersch, F. Chummun, L. Bella: “Simulation Study of Dynamic Cost-Based Routing in PNNI Networks, to appear in IEEE ATM Workshop 2000, June 2000.Google Scholar
  12. [12]
    M. Conte: “Dynamic Routing Schemes for ATM Networks”, Dissertation, TU Wien, to be published.Google Scholar

Copyright information

© IFIP International Federation for Information Processing 2000

Authors and Affiliations

  1. 1.Siemens AG ÖsterreichWienAustria

Personalised recommendations