A Segmentation based Classification Model for Primary User Detection Using Deep Learning Techniques

Abstract

Designing wireless communication devices involves the novel idea of cognitive radio (CR) to mitigate the spectrum scarcity problems in the available frequency spectrum. CR has the ability to learn and adapt to their environment. CR allows secondary users (SU) to share the licensed spectral band of primary users (PU) if and only if the PU is not subject to harmful interference. Spectrum sensing is a fundamental function of CR that helps it to gain opportunistic spectrum access to its users. To further expand the learning ability of CRs and to provide an efficient PU spectrum sensing, machine learning or deep learning algorithms can be applied. This paper proposes an efficient and well performing segmentation cum classification algorithm based on deep learning techniques for PU detection. The spectrogram of the PU's transmission signal pattern for different scenarios was classified using Res-Net 50 model. To further improve the accuracy, a region proposal based Res-Net50 model is proposed. The performance evaluations validate the effectiveness of the proposed model.