| Popis: |
Automatic modulation classification (AMC) is a technique to identify the modulation of signals. Traditionally, this has been done through feature extraction; however, recently machine learning has been used. Convolutional neural networks (CNNs) are a type of deep learning that have the ability to learn feature extraction. They are excellent for classification, and analyzing large amounts of data. In this research, CNNs were used to automatically classify the modulation type of underwater acoustic signals. The architecture of the CNN was iteratively improved, until a final accuracy of approximately 93% was achieved. The training and testing was done using a dataset collected by Applied Research Laboratories at the University of Texas Austin. |
