RadVIEW: Robust radar detection and characterization in high-noise regimes

Accurate and objective mutual awareness between coexisting technologies is the root of trust for future spectrum sharing. Despite this, our ability for unsupervised detection of narrow-band and short-lived signals in spectrum traces is limited. This poses practical challenges in sharing spectrum with entities such as naval or weather radars, whose transmissions are inherently short-lived and often narrow-band. Ultimately, this creates a reluctance within licensed spectrum users to share their resources. Thus, the detection of short-lived and narrowband emitters, especially in high noise scenarios, will ensure the practical applicability of dynamic spectrum sharing. To this end, we develop RadVIEW, a lightweight and fullyunsupervised wavelet-based spectrum characterization method for detection of narrow-band and short-lived transmitters in highnoise regimes. We demonstrate RadVIEW’s performance and applicability on realistically emulated naval radar traces across a wide range of signal-to-noise regimes. We show near-ideal characterization performance in comparison with counterparts from the literature. We demonstrate RadVIEW’s applicability to CBRS systems and beyond as it canreduce the required Environmental Sensing Capabilities protection zones by a factor of three and detect the presence of incumbent radars in less than 0.05 seconds.