A Machine Learning Approach to Improve Spectrum Utilization in Cognitive Radio Networks.

Abstract

In regards to wireless communications, due to the explosion in demand for higher capacity networks, availability of free spectrum resources have become increasingly scarce. The UHF spectrum band in particular, due to its excellent electromagnetic properties, has been reported as inefficiently used and congested by many spectrum regulators of the world. This spectrum resource scarcity issue combined with the ongoing research and development for more intelligent, autonomous and self-aware radio communication led to a vast amount of research on the concept of Cognitive Radio. Spectrum sensing plays a significant role in enabling utilization of spectrum holes by unlicensed secondary users in cognitive radio networks. Most of the related work concerning spectrum sensing has focused on sensing carried out by incoming secondary users aiming at locating spectrum opportunities. In this project, we analyze how to improve spectrum utilization in cognitive radio networks (CRNs) using different supervised machine learning techniques Logistic Regression, K-Nearest Neighbor and Naives Bayes which are studied to find the best technique with the highest classification accuracy (CA). Ten (10) different locations around Makerere University our case study and we manually collected one thousand (1000) data sets from each location making a complete data set of 10,000 samples using an RF explorer coupled with the RF explorer software that is installed on the computer. With the complete data set, we cleaned the data set, the data set was preprocessed and then divided it into \Test" and \Train" data. Numerical results show that techniques Logistic Regression and K-Nearest Neighbor are the best algorithm among all the supervised classifiers. We later on designed a web application which can be accessed by the user through a web browser with an active internet connection to easily understand all the aspects about our project.