Cyclic Feature-Based Modulation Recognition Using Compressive Sensing

Abstract

Higher-order cyclic cumulants (CCs) have been widely adopted as the features for automatic modulation recognition (AMR). However, the requirement of high-rate sampling has greatly restricted the application of CCs. In order to relieve this sampling burden, we attempt to bring AMR into the framework of compressive sensing. By exploiting the sparsity of CCs, the proposed AMR algorithm is implemented with a significantly small number of nonuniform samples of the observed signal. Simulation results demonstrate the availability and robustness of the proposed approach.