Blind equalization and automatic modulation classification of underwater acoustic signals

Automatic characterization of underwater acoustic signals enables better use of the acoustic spectrum through activities such as interference avoidance and marine mammal protections. Determining the modulation of a received waveform can permit sonar and communications within the same bandwidth with minimal collisions, and it can identify systems operating outside their permitted regime. The characterization system determines the signal modulation in the presence of an unknown, time-varying channel impulse response. The work presented here demonstrates the use of blind equalization along with Convolutional Neural Networks (CNNs) for automatic classification of underwater signals. The current research focuses on classification of constant modulus signals since these signals are widely used for acoustic communications. The proposed approach provides an approximate 24 percent improvement in modulation classification compared to approaches without equalization and requires less training data than previous appr...