Fault Diagnosis Method for Mobile Ad-hoc Network by Using Smart Neural Networks

MANETs are dynamic collection of autonomous nodes that communicate with each other via wireless connections. One of the events that the network should have expected it to be a fault, and the behavior is more important, in this network. So that fault diagnosis can effect on final performance of the network in such a way that it does not fall under the negative impact of the fault. A non-linear neural network is a statistical method for modeling data or the tools to make decisions. Artificial neural network is a method for pattern recognition and classification. Error detection is a problem of categorization or classification. The use of neural networks can be useful in fault diagnosis in MANETs because of fault diagnosis is a classification problem. But one problem with this method is placed in a local optimum. Here a method based on the combination of the back-propagation algorithm, a local search algorithm and learning automata as efficient global search, is proposed. In the proposed method, the algorithm of learning automata adjusting learning rate on neural network according to given formula. For training and testing the neural network of the mobile network parameters that were measured, were used as input and output. The results show that the proposed method in terms of repeatability, reliability and lack of placement in a local optimum is better.