Cooperative spectrum sensing in cognitive radio networks using multi-class support vector machine algorithms

This paper addresses the problem of spectrum sensing in cognitive radio networks under multiple primary users condition using multi-class support vector machine (SVM) algorithms. First, we formulated the spectrum sensing problem under multiple primary users scenario as a multiple class signal detection problem where each class is comprised of one or more sub-classes and generalized expressions for the possible classes are provided. Next, we investigate the performance of energy based features and the error correcting output codes (ECOC) based multi-class SVM algorithms for solving the multi-class spectrum sensing problem using two different coding strategies. The performance of the proposed detector is quantified in terms of receiver operating characteristics curves and classification accuracy. Simulation results show that the proposed detector is robust to joint spatio-temporal detection of spectrum holes in cognitive radio networks.