Multichannel Software Defined Radio with Spectral Decision via Centralized Artificial Intelligence

Abstract

The present paper aims to bring Artificial Intelligence (AI) in Software Defined Radio (SDR). A multichannel spectrum sensing problem, extended to a long-term spectral occupancy observation, enabled the authors to derive a “vertical” per-channel machine learning model that was tested in an integrated National Instruments (NI) environment – Ettus/NI USRP (Universal Software Radio Peripherals) service-driven, top-down, by LabVIEW. The proof-of-concept was based on a simple 8 channels PMR (Private Mobile Radio) use-case.