SVM-Based Spectrum Sensing in Cognitive Radio

Spectrum sensing is a fundamental process in cognitive radio. In order to sense primary user (PU) information, this paper applied support vector machines (SVM), which is a data mining method, to develop a real-time approach for detecting. The sample data could be classified as PU or not by training and testing on proposed SVM classification model in time domain. For linear classification, kernel function is proposed to map the input low dimensional vector into a high dimensional feature space. The paper exploits thoroughly two parameters: the parameter in RBF kernel function and the sampling data dimension. The simulation shows that the SVM model possesses excellent recognized ability in the low SNR compared with energy detection. The probability of correct detection will achieve an accuracy of 100% when false rate is a little.