Ensemble Autoregressive Neural Network (ARNN) Model for Spectrum Occupancy Prediction

Abstract

A secondary spectrum user cannot transmit in a channel before sensing and knowing the spectrum occupancy state as this may cause interference. This poses a major challenge because these operations ought to be performed in each time slot and thereby causing a substantial delay before the user gains access to the spectrum, leading to inefficient utilization. Therefore, a channel predictive system will mitigate this problem. In this work, an ensemble machine learning model for spectrum occupancy prediction was developed. The developed model was trained using a sample of Power Spectrum Density (PSD) data collected from the field for a period of twenty-four hours within a frequency range of 30-300 MHz. The frequency range was grouped into sub bands. Based on the training data and the corresponding output data, the neural network model trains itself to come up with the best weights which can generally be used by the AR model for unseen data. After computing the weights, the performance is first tested on the entire training data, on the validation dataset and on the test dataset. Prediction results reveal an overall accuracy of 98.32% with band 4 (74.85-87.45 MHz) having the highest accuracy of 99.01% and the lowest accuracy of 89.39% in band 2 (47.05-68 MHz).