Implementation of Hidden Markov Model spectrum prediction algorithm

Abstract

The demand for spectrum is at an all time high due to the increasing popularity of wireless devices. As such it is imperative that technological and regulatory mechanisms are developed to maximise spectral efficiency. This paper documents a method for increasing spectral efficiency through the prediction of spectrum "holes" for use with cognitive radio technologies. An algorithm is developed based upon a Hidden Markov Model of the spectral environment. The Baum-Welch algorithm is employed to dynamically calculate the transition parameters of the model. The algorithm is tested upon data collected from the 450-470 MHz band in Australia with a reward function implemented to analyse the performance of the algorithm.