New method fuses 2D/1D signals for robust underwater acoustic modulation recognition

Researchers have developed a novel method for recognizing underwater acoustic modulations, particularly addressing challenges posed by distribution shifts. Their approach, SCP-TriCA, fuses heterogeneous 2D and 1D signal representations, including time-frequency maps, cyclostationary maps, and statistical descriptors. This fusion is achieved through bidirectional cross-attention for 2D modalities and a sample-adaptive selective gate for incorporating 1D statistical information. The method was evaluated on UAMR-ShiftBench, a new benchmark designed to systematically test performance under various shift types, including real-world sea trials. AI

IMPACT This research could improve the reliability of underwater communication systems by enhancing signal recognition capabilities under challenging environmental conditions.

RANK_REASON The item is an academic paper detailing a new method and benchmark for a specific technical problem. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.AI →

• bidirectional cross-attention
• cyclostationary maps
• higher-order power spectra
• sample-adaptive selective gate
• SCP-TriCA
• short-time Fourier transform
• South China Sea
• UAMR-ShiftBench

• paper
• other

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →