HMM Models and Estimation Algorithms for Real-Time Predictive Spectrum Sensing and Cognitive Usage

This work investigates the use of Non-Stationary Hidden Markov and Hidden Bivariate Markov Models through simulations and real-time application to predict RF channel occupancy in cognitive radio systems. Real data collected in public safety frequency band during the Rio 2016 Olympic and Paralympic Games were used to test a simple cognitive spectrum sharing scheme here proposed with the main goal of maximizing secondary usage of available spectrum. Several algorithms for parameter estimation are compared in this context. Remarkably good performance is obtained with a windowed version of the Baum algorithm and Non-Stationary Hidden Markov Model.