MobileNetV3-based Automatic Modulation Recognition for Low-Latency Spectrum Sensing
Myke D. M. Valadão, Lucas F. M. da Silva, Matheus Serrao, Willian Guerreiro, V Viana Furtado, Natalia Freire, Gelson Monteiro, Carlos Craveiro
Abstract
The cognitive radio senses and tracking the spectrum of frequency and allocate dynamically secondary users in spectral holes. Automatic modulation recognition provides preliminary information to assist in spectrum sensing. This letter proposes a MobileNetV3 for the task of modulation recognition. The MobileNetV3 is a recent network that achieved higher accuracy rates with less computation cost. Experiments showed that the proposed framework achieved higher accuracy than other state-of-art techniques in a comparable latency, which 5G and 6G required.
Topics & Concepts
Cognitive radioComputer scienceLatency (audio)ComputationModulation (music)Radio spectrumSpeech recognitionArtificial intelligenceWirelessTelecommunicationsAlgorithmAestheticsPhilosophyWireless Signal Modulation ClassificationRadar Systems and Signal ProcessingCognitive Radio Networks and Spectrum Sensing