Premium
Design of an optimal BER monitoring scheme for preventive maintenance of an ATM network
Author(s) -
Tuchinda Chindakorn,
Zhu Hou Tao,
Lu Weiyi
Publication year - 1998
Publication title -
international journal of communication systems
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.344
H-Index - 49
eISSN - 1099-1131
pISSN - 1074-5351
DOI - 10.1002/(sici)1099-1131(199811/12)11:6<369::aid-dac378>3.0.co;2-5
Subject(s) - computer science , erlang (programming language) , computer network , asynchronous transfer mode , functional programming , theoretical computer science
In today's telecommunication networks it is well known that unexpected failures in the physical links may cause significant loss or interruption of services. It is desirable if the deterioration of links can be recognized early and maintenance is carried out before failure occurs. Asynchronous transfer mode (ATM) networks have recently received much attention and are considered to be the prime universal networks of the future. Unfortunately, there has been very little research in the areas of preventive maintenance and performance monitoring of the ATM network. In this paper we propose a preventive maintenance monitoring scheme for an ATM network where a performance index called the bit error rate (BER) index of each service of the ATM network is continuously monitored. The BER index of each path in the ATM network is assumed to be slowly deteriorating and is modelled as a Brownian motion process with positive drift. If any of the BER indices exceeds its action limit or threshold, a repair process is initiated. The optimal action limit is derived by minimizing the modified long‐run average maintenance cost. We found that the optimal action limit depends strongly on the multiservice characteristics of the ATM network and on the network topology, which relates to the virtual path (VP) bandwidth management and routing protocol used. To obtain the optimal action limit, an optimization model for VP bandwidth management is set up and the multidimensional Erlang loss model is used to determine the path selecting probabilities. Some numerical examples are given to illustrate the proposed model and the optimization results. © 1998 John Wiley & Sons, Ltd.