Automatic modulation classification via instantaneous features

Abstract

Automatic modulation classification has attracted a lot of interests in the research community in recent years due to the advances in cognitive RF signal processing such as cognitive radio, cognitive radar and cognitive electronic warfare. There are two major approaches in automatic modulation classification, namely the feature based approach and the decision theoretic approach. In our previous work, we have demonstrated the feasibility of using cyclostationary statistical features such as spectrum correlation function to perform modulation detection and classification for both RF signals and underwater acoustic signals. In this paper, we try to develop automatic modulation classification algorithms employing instantaneous features such as instantaneous amplitude, phase and frequency parameters. By extending previously developed features and evaluating appropriate decision metrics, we have been able to expand our modulation classification capability to 9 popular modulations including 2ASK, 4ASK, 8ASK, 2FSK, 4FSK, 8FSK and 2PSK, 4PSK, 8PSK. Thorough simulation results confirm the effectiveness of our proposed algorithm and threshold choices. The success of this approach also suggests a future research direction to combine statistical features with instantaneous features to provide a more accurate and more robust modulation identification algorithm.