International Journal of Engineering and Technology

In order to improve the reliability and availability in distributed networks, the occurrence of failures are to be predicted. Fault Prediction aims at predicting failure before they occur in Distributed Systems. This is possible by closely observing the behavior of the network. Network packets get latest information of network for every instant. The proposed work uses standard method to analyze network packets and simple mathematical calculations and timed automata in order to estimate Time-To-Failure (TTF). This method is able to detect only observable faults.