A new public synthetic dataset of 6000 radar pulse trains with up to 110 overlapping emitters enables standardized benchmarking and model development for pulse deinterleaving.
Over-the-air deep learning based radio signal classification
7 Pith papers cite this work. Polarity classification is still indexing.
7
Pith papers citing it
years
2026 7representative citing papers
Selecting IQ, amplitude-phase, and autocorrelation representations as complementary signal priors, then fusing them with a lightweight attention mechanism and adversarial domain alignment, improves cross-domain modulation classification by 8–15 percentage points over source-only baselines.
Dualformer applies a parameter-sharing DualNN via Transformer patches to complex signals, claiming better results on AMR, SSR, and SSP tasks than baselines.
VLMs trained on synthetic RF spectrograms generalize to real signals for physical attribute extraction but lack reliable semantic grounding without additional priors.
A sparse coding plus hierarchical tree pipeline for automatic modulation classification cuts model parameters by 41% and FLOPs to 10^{-4} of lightweight deep learning baselines.
A compact SDR platform using HackRF One and Raspberry Pi captures 20 Msps IQ data with GNSS metadata and shows distinct RF propagation in foliage, urban, and indoor settings.
GAMC is a four-stage interpretable ML pipeline for AMC that transforms I/Q signals into constellation and graph representations, extracts features, learns discriminative projections, and uses SNR soft routing to achieve higher accuracy with 50% fewer parameters and 3-42% of the compute of comparable
citing papers explorer
-
The Turing Synthetic Radar Dataset: A dataset for pulse deinterleaving
A new public synthetic dataset of 6000 radar pulse trains with up to 110 overlapping emitters enables standardized benchmarking and model development for pulse deinterleaving.
-
DKDNet: Dual Knowledge and Data-Driven Network for Cross-Domain Automatic Modulation Classification
Selecting IQ, amplitude-phase, and autocorrelation representations as complementary signal priors, then fusing them with a lightweight attention mechanism and adversarial domain alignment, improves cross-domain modulation classification by 8–15 percentage points over source-only baselines.
-
Dualformer: Efficient Feature Extractor for Complex-valued Blind Communication Signal Analysis
Dualformer applies a parameter-sharing DualNN via Transformer patches to complex signals, claiming better results on AMR, SSR, and SSP tasks than baselines.
-
RF-Analyzer: Can Vision-Language Models Learn RF Understanding from Synthetic Data?
VLMs trained on synthetic RF spectrograms generalize to real signals for physical attribute extraction but lack reliable semantic grounding without additional priors.
-
G-AMC: A Green Automatic Modulation Classification Method
A sparse coding plus hierarchical tree pipeline for automatic modulation classification cuts model parameters by 41% and FLOPs to 10^{-4} of lightweight deep learning baselines.
-
Field-Deployable RF Capture System for Indoor, Outdoor, and Foliage Environments
A compact SDR platform using HackRF One and Raspberry Pi captures 20 Msps IQ data with GNSS metadata and shows distinct RF propagation in foliage, urban, and indoor settings.
-
Automatic Modulation Classification via Green Machine Learning
GAMC is a four-stage interpretable ML pipeline for AMC that transforms I/Q signals into constellation and graph representations, extracts features, learns discriminative projections, and uses SNR soft routing to achieve higher accuracy with 50% fewer parameters and 3-42% of the compute of comparable