LAUSR

With the popularity of software-defined radio and cognitive radio-technologies in wireless communication, radio frequency devices have to adapt to changing conditions and adjust its transmitting parameters such as transmitting power, operating frequency, and modulation schemes. Thus, automatic modulation classification becomes an essential feature for such scenarios where the receiver has a little or no knowledge about the transmitter parameters. This paper presents kth nearest neighbor (KNN)-based classification of M-QAM and M-PSK modulation schemes using higher-order cumulants as input features set. Genetic programming is used to enhance the performance of the KNN classifier by creating super features from the data set. Simulation result shows improved accuracy at comparatively lower signal-to-noise ratio for all the considered modulations.