Automatic Modulation Recognition techniques based on cyclostationary and multifractal features for distinguishing LFM, PWM and PPM waveforms used in radar systems as example of artificial intelligence implementation in test

Abstract

Automatic Modulation Recognition (AMR) is an example of implementation of Artificial Intelligence to cognitive radio received signal software testing. This article proposes two fairly simple and computationally feasible AMR algorithms, based on the principles of cyclostationarity and multi-fractals, suitable for practical real-time software radio communications applications for distinguishing Linear Frequency Modulation (LFM or Chirp), Pulse Width and Pulse Position Modulations (PWM/PPM) waveforms used in Radar systems, both commercial and military, from other commonly employed modulations such as, for example, BPSK, BFSK, GMSK. In these techniques, the incoming received signal is processed to determine the cyclostationary and multifractal features of the waveforms which are later matched by a neural network classifier with corresponding feature patterns of stored modulated waveforms, declaring the appropriate modulation present for whichever waveform produces the highest matching output. A spreadsheet of classification probabilities for both techniques is generated which compares their performance for the six studied waveforms.