Comparative Study of SVM and KNN Machine Learning Algorithm for Spectrum Sensing in Cognitive Radio

The fast growth of wireless technology in today’s scenario has paved huge demand for licenced and unlicenced frequencies of the spectrum. Cognitive radio will be useful for this issue as it provides better spectrum utilisation. This paper deals with the study of machine learning algorithm for cognitive radio. Two supervised machine learning techniques namely SVM and KNN are chosen. The probability of detection is plotted using SVM and KNN algorithms with constant probability of false alarm. Comparison of the two machine learning methods is made based on performance with respect to false alarm rate, from which KNN algorithm gives better spectrum sensing than SVM. ROC curve is also plotted for inspecting the spectrum when secondary users are used.