Spectrum Occupancy Prediction Algorithm Using Artificial Neural Networks

The cognitive radio system is a promising solution to the problem of deficit of the radio frequency spectrum. Dynamic spectrum access technology is a key step in the development of these systems. An analysis of the existing methods for predicting the state of the spectrum has shown that it is necessary to solve the problem of long-term forecasting for the effective operation of the dynamic spectrum access technology. The solution of this problem makes it possible to provide this technology with information about the state of occupancy of frequency channels for a long period, thus reducing the probability of interference for the primary user. An algorithm for predicting the spectrum based on neural networks is proposed to solve the problem. The proposed algorithm consists of a technique for generating and preparing input data sets and a recurrent neural network model using a long short-term memory architecture.