Cumulant-based channel estimation algorithm for modulation classification in frequency-selective fading channels

Due to the lack of sufficient channel state information, modulation classification in frequency-selective fading channels is a challenging task. This is mainly because the complexity of the pre-processing stage, where the required channel state information for (optimal) likelihood-based classification is estimated, can be relatively high. For this reason, we propose in this paper a low-complexity cumulant-based channel estimation algorithm that enables the reliable classification of digital amplitude-phase modulated signals in frequency-selective fading channels. Numerical results are presented which show that the proposed algorithm outperforms a commonly used channel estimation algorithm for modulation classification, especially at low signal-to-noise ratio values. Using the proposed algorithm, it is also shown that the performance of a practical modulation classification method can approach that of a clairvoyant classifier assumed to have perfect channel knowledge.