Analyzing Power Correlation Distributions for Dynamic Spectrum Awareness in 5G Networks

With continued advancements in the cellular domain, the number of network devices is expanding rapidly. Furthermore, as devices become smarter and more advanced with the deployment of dynamic spectrum awareness techniques that facilitate increased spectral efficiency, developing robust methodologies to analyze limited spectrum resources will be crucial to the success of FutureG wireless communications. Central to the idea of dynamic spectrum awareness is the need for robust techniques and metrics that provide insight into the spectral utilization and can be harnessed by dynamic control loops to maximize spectral usage and eliminate wasteful inefficiencies. While a plethora of dynamic spectrum awareness tactics have been proposed, existing strategies do not analyze the power spectral density probability distribution across time, which we demonstrate is a relevant indicator of spectral usage and signal quality. Specifically, we harness the commercial-grade, standalone, 5G network architecture located at the Air Force Research Laboratory (AFRL) to perform an experimental measurement study and show the effectiveness of representing the localized spectral utilization as a distribution of power correlations w.r.t. the bandwidth’s frequency bins. We analyze the distributional structure of power spectral density correlations for a variety of wireless channel environments (i.e., LOS, NLOS, and Edge scenarios) and locations, and our key finding is that the shape of the frequency-bin correlation distribution changes significantly depending upon the particular wireless channel situation. As such, our approach - which is not only blind (i.e., it doesn’t require knowledge about the signal prior to sensing the spectrum), but also highly practical from an implementation standpoint - provides fresh insight into the spectral utilization, signal structure, and channel quality conditions.