Novel RF Spectrum Characterization Using Information Measures

As the number of radio frequency (RF) systems in use continues to increase, the need to monitor and securely share limited spectrum continues to correspondingly grow. Tracking and analyzing spectrum usage over time is pivotal to secure dynamic spectrum sharing. This paper presents a novel unsupervised, information-based approach to identifying and characterizing the complexity and quality of an RF signal's time-frequency (TF) characteristics. The proposed method draws on tools from information geometry and utilizes the set of correlation matrices. In particular, the informativeness is a recently developed measure of the homogeneity of a data set. The informativeness provides a two-parameter characterization of multi-dimensional data that can be used to assess TF grids for homogeneity. This intrinsic consistency can be used to assess the quality or complexity of recorded data at a single sensor, and to assess consistency between pairs of sensor network nodes.