NEW PUBLICATION

DNNs for F0 Extraction in Ocean Meters on uC Edge

This poster presents an edge computing method for extracting the fundamental frequency of tidal currents using an ultra-low power ARM Cortex-M0+. A DNN-based algorithm narrows filtering bands to reduce power and improve efficiency, achieving a 1.81× speedup over a 32-point FFT. Time/frequency domain methods (FFT, Kalman), curve fitting, GAs, and NNs are compared. Only [9] targets uLP Cortex-M0+ directly.