A modulation classification method in cognitive radios system using stacked denoising sparse autoencoder

This paper proposes a modulation classification method based on Stacked Denoising Sparse Autoencoder (SDAE). This method can extract modulation features automatically, and classify input signals based on the features it extracts. The scenarios of rapid classification and high accuracy classification are considered. In the rapid classification scenario, a long symbols sequence is not attainable for this scenario. Moreover, expert features are not necessary for this scenario, simplifying the modulation classification procedure and rendering rapid classification more achievable. In addition, in the high accuracy classification scenario, the higher cumulants are used as expert features due to its advantage over other tries at noise resistance. Moreover, we use complex symbols rather than pulse shaped complex signals as network input, which simplifies the network topology and saves the calculation overhead. The results of the average classification accuracy and the execution time are presented, indicating significant performance advantages over the other methods.