Authors
Vaasu Taneja, Andrew Okoro, Clark Mattoon, Xin Li , and Mariya Zheleva
Abstract
This poster presents M-SHarC, a novel approach for characterizing faint harmonic activity and reconciling it with its fundamental frequency. M-SHarC operates on a 3D time-frequency-band tensor that captures a transmitter’s fundamental and harmonic activity across multiple spectra. It enhances low SNR inputs using a diffusion model, applies sparse dictionary learning to characterize transmitter and harmonic activity, and finally reconciles harmonics with their respective fundamental frequencies. Through simulation and over-the-air experimentation, we demonstrate M-SHarC’s high performance in extremely low SNR regimes, successfully detecting harmonics only 3dB above the noise floor. To our knowledge, this work is the first to conceptualize multi-spectral transmitter characterization, with direct implications for scalable radio frequency interference surveys and spectrum enforcement.