Robust and fast failure detection and prediction for fault-tolerant communication network

Fault detection performance in fault-tolerant communication networks can be greatly improved by making use of statistical filtering. In the Letter, the channel is modelled as a finite-state system and a maximum-likelihood sequential filtering approach is proposed which uses hidden Markov modelling techniques to improve the model based on empirical field data.