Sparse decomposition framework for maximum likelihood classification under alpha-stable noise

Recently, automatic modulation classification has gained a lot of attention in the area of cognitive radio (CR), signal detection, electronic warfare and surveillance etc. Most of the existing modulation classification algorithms are developed based on the assumption that the received signal to be identified is corrupted by only additive white Gaussian noise. The performances of these conventional algorithms degrade significantly by addition of impulse noise. In this paper, we propose a robust algorithm using sparse signal decomposition which comprises of an overcomplete dictionary for detection and classification of modulated signals. In this work, an overcomplete dictionary is constructed using the identity basis, cosine and sine elementary waveforms to capture morphological components of the impulse noise and deterministic modulated signals effectively. The proposed method of modulation classification consists of the three major steps: sparse signal decomposition (SSD) on hybrid dictionaries, modulated signal extraction, and maximum likelihood (ML) based classification. The testing and validation of both direct ML and SSD-based ML classification methods are carried out under different Gaussian and impulse noise conditions for modulation classification. Our proposed method achieves a classification accuracy of 85% at 5 dB SNR and outperforms the conventional classification methods.