Semi-blind algorithms for automatic classification of digital modulation schemes

The problem of automatic classification of digital communication modulation schemes is considered in this work. Firstly, the maximum likelihood (ML) classifier for classifying phase-amplitude modulated schemes in coherent environment is presented. It is well known that the ML classifier requires the knowledge of the signal-to-noise ratio (SNR) and has a higher computational complexity. To relax the first requirement, we introduce a novel idea to estimate the SNR and this gives rise to a novel estimated ML (EsML) classifier. After which, in an attempt to reduce the computational complexity of the EML and EsML classifiers, we propose a simplified minimum distance (MD) classifier. The performance of these classifiers are compared against each other's under the ideal channel condition as well as under a channel condition with an unknown carrier phase offset. In the second part of the paper, we adapt a closed form blind source separation (BSS) algorithm for rectifying the carrier phase offset prior to the actual classification procedures.