C45. Classification of OFDM signals using higher order statistics and clustering techniques

In the context of cognitive radio or military applications, it is a crucial task to distinguish various OFDM based systems such as fixed WiMAX and Wi-Fi from each others. This paper presents a novel technique that deals with the classification of OFDM signals using higher order moments and cumulants with different types of classifiers and clustering techniques is proposed. Four classification techniques were considered, namely, Support Vector Machines, K-nearest neighbors, Maximum Likelihood and neural network (NN) classifiers. Two clustering techniques were utilized, namely, Fuzzy K-Means and Fuzzy C-means. Simulation results show that the proposed technique is able to classify different types of OFDM signals in Rayleigh fading and additive white Gaussian noise (AWGN) channels with high accuracy. It is also shown that the NN classifier outperforms the other three considered classifiers.