Identification of ISM Band Signals Using Deep Learning

Abstract

Spectrum awareness is now becoming more and more important in recent years, which can be utilized in areas like spectrum resource allocation, spectrum management, inference control, and security protection. Deep learning (DL) models, including convolutional neural network models have been widely used for classification related tasks, such as modulation classification, medium access control protocol (MAC) classification, and spectrum sensing. In this paper, a pre-trained Inception V3 model (CNN-based) is used to classify industrial, scientific, and medical (ISM) radio band signals. Experimentation results demonstrate the effectiveness of deep learning in ISM band signal identification.