Naïve Bayes classification of adaptive broadband wireless modulation schemes with higher order cumulants

Adaptive modulation schemes have been proposed to optimize Shannon's channel capacity in recent orthogonal frequency division multiplexing (OFDM) based broadband wireless standard proposals. By adapting the modulation type (effectively changing the number of bits per symbol) at the transmitter end one can improve the bit error rate (BER) during transmission at designated SNR. Blind detection of the transmitted modulation type is desirable to optimise the bandwidth available at the receivers. Hence, there is a need for an intelligent modulation classification engine at the receiver end. In this work, we evaluate some higher order statistical measures coupled with a classical Naive Bayes classifier for fast identification of adaptive modulation schemes. We also benchmark the experimental results with the optimal Maximum Likelihood Classifier, and Support Vector Machine based Classifier using the same feature set.