A BB network faults simulator based on neural networks

The aim of this paper is to explore possibilities of applying neural networks to create a simulator of user interference on broadband (BB) networks. The model is made on the basis of one-year monitoring of broadband interferences, divided by service, the reporting day of the week and the exact time of occurrence. The basic idea is to create a neural network capable of accepting the three groups of input data (time, date and a service), and calculate the number of disturbances in the time unit for a given service as output data. Dispersion is also added in form of white noise based on actual values. Accuracy check of the tested model is made by comparing the simulated and the real data.