A New Fuzzy Adaptive Approach for Fault Identification in Computer Networks

A.A. Mohamed and O. Basir (Canada)


Fault management systems, computer networks, constraint satisfaction problem, fuzzy inference


To examine the current status of the network entities, widely available software testing tools (also called probes) are employed by network fault management systems. A single probe can be used to test the working conditions of only a subset of the network nodes. Therefore, to perform the fault identification task, a fault management system often utilizes a collection of such probes such that they should completely cover all the nodes in the managed network. However, large number of these probes may in fact exacerbate some of the problems (such as congestion) currently experienced by the network and introduce extra management traffic into it. In this paper, we propose an adaptive fuzzy constraint satisfaction problem-based approach to significantly reduce management traffic (probes) while maintaining its fault diagnostic power. In this new approach, the most informative probes among the available ones are selected based on some criteria that are well-formulated in the definitions of the problem constraints. The efficiency of the new algorithm has been examined by extensive experiments and its results are reported.

Important Links:

Go Back