bstract—Spectrum sensing is a key aspect of Cognitive<br />Radio (CR). The main requirement in CR systems is the<br />ability to sense the primary signal accurately and rapidly.<br />In this paper, a novel hybrid spectrum sensing scheme in<br />CR is proposed which considers the hypothesis problem as<br />a binary classification problem. The proposed scheme is a<br />combination of classical energy detection, Likelihood Ratio<br />Test statistic (LRS-G2<br />) and Artificial Neural Network<br />(ANN). The scheme utilises energy from energy detection<br />and Zhang test statistic from LRS-G2<br />as features to train<br />the ANN while ANN provides the adaptive learning and<br />stable performance to the scheme. The performance of<br />proposed sensing scheme is evaluated on several real-<br />world primary signals of various radio technologies and<br />it has been found out that for all those radio technologies<br />the proposed scheme outperforms the classical energy<br />detection and the improved energy detection.