A robust algorithm based on wavelet transform for recognition of binary digital modulations

Automatic Modulation Recognition (AMR) of communication signals has been an important theme in wireless communication systems. Recently, digital modulation recognition has been considered remarkably, due to the vast application of this kind of AMR technique in spectrum sensing and cognitive radio. In this paper, we propose a wavelet-based algorithm for recognition of binary digital modulation schemes including BASK, BFSK and BPSK in presence of Additive White Gaussian Noise (AWGN). In this algorithm, Haar Wavelet Transform (HWT) of the received signal is compared to templates and the similarity of them are measured. Experimental results show a perfect classification of binary digital modulations in Signal to Noise Ratio (SNR) above -1 dB. For perfect recognition of BFSK, minimum SNR of -7 dB is required. The average accuracy of 99.97, 99.71 and 97.34 are obtained for classification of these three modulations in -5dB, -7dB and -10dB SNR values, respectively.