Academy & Industry Research Collaboration Center (AIRCC)

Volume 9, Number 4, March 2019

Implementation of Machine Learning Spectrum Sensing for Cognitive Radio Applications

  Authors

Mohamed El-Tarhuni1, Khaled Assaleh2, and Firas Kiftaro1, 1American University of Sharjah, UAE and 2Ajman University, UAE

  Abstract

In this paper, a cognitive radio system is implemented using National Instruments (NI) Universal Software Radio Peripheral (USRP) devices. The implemented system provides a working prototype based on real data generated and collected by an experimental laboratory setup to compare the performance of spectrum sensing algorithms based on energy detection and polynomial classifier channel sensing techniques. For a sensing time interval ranging from 0.05 ms to 5ms, the experimental results show that the polynomial classifier has a better performance compared to the conventional energy detector in terms of the misclassification rate, especially at lower SNR values

  Keywords

Cognitive Radio, Spectrum Dynamic Access, Spectrum Sensing, Polynomial Classifier, Energy Detection, USRP