Clustering RF Signals with the Growing Self-Organizing Map for Dynamic Spectrum Access
Elliott Konink-Donner, Aaron Ruen, Rashmi Jha · 2023
The ability to dynamically navigate and analyze significant spectral features is a desirable characteristic of cognitive radio networks, commonly referred to as dynamic spectrum access (DSA). DSA technologies have primarily been applied to allow more efficient use of the radio frequency (RF) spectrum as more devices crowd the spectrum. DSA can enable cognitive radio networks to reconfigure themselves when a communication channel is shown to have active users, or simply provide more information about the signals transmitted over the channel. The goal of this work is to explore unsupervised neural network architectures that extract significant features for DSA with little a priori knowledge. We focus on well-established clustering algorithms for the unsupervised organization of spectral features.