Modulation Recognition Based on Feature Extraction by AutoRegressive Model

Abstract

Through the research of the digital modulation signals, we find that the signal would produce a series of spectrum lines after a specific nonlinear transformation. Therefore, this characteristic has a certain potential in the signal modulation recognition. In this paper, first, we analyze the spectrum lines mechanism of modulation signals (4PAM、BPSK、QPSK、8PSK、2FSK、4FSK), then, we use AR model to extract the spectrum lines characteristics of modulation signals to identify different signals. In the end of the paper, the above-mentioned six signals are simulated in the MATLAB environment. Simulation results show that the recognition of the modulation signals through this method has good stability in non-ideal communication environment.

DOI : http://dx.doi.org/10.11591/telkomnika.v12i3.4156