Manifold learning-based automatic signal identification in cognitive radio networks

Abstract

Adaptive signal identification has been an important issue in cognitive radio networks (CRNs). Most existing techniques require high-level signal-to-noise ratio (SNR) for signal identification. This study presents an intelligent technique that focuses on a theoretical and experimental study of the signal identification by using manifold learning algorithm in CRNs. The authors pose the problem of signal identification in CRNs as signal classification by using manifold learning on high dimensions, and a novel manifold learning algorithm named as SIEMAP is proposed, which is able to identify signals in a low-dimensional space. Simulation results indicate that SIEMAP outperforms classical methods in low dimensions and is capable of identifying signal types from the received signals.