How kalit31/drone-detection-and-identification-using-rf-signals works
This repository is an experimental RF-drone detection and identification framework. Its documented main workflow extracts power-spectral-density (PSD) features from DroneRF data and trains classifiers for 2-, 4-, and 10-class tasks. It also explores feature fusion, 1D-CNN classification, and a hybrid CNN-feature-extractor/XGBoost approach. The tree contains additional MFCC, DFT, and time-domain experiments. Source excerpts were unavailable, so links beyond relationships stated in the README are omitted.
Main entry points: CNN_Based_Feature_Extractor_XGBoost.ipynb XGBoost classifier · psd_feature_extraction.m PSD extraction · feature_fusion.m Feature fusion
Components 13
Feature Engineering
- PSD extraction
code/matlab code/psd_feature_extraction.m - MFCC extraction
code/matlab code/mfcc_extraction.m - Time-domain features
code/matlab code/time domain features - Feature fusion
code/matlab code/feature_fusion.m
Traditional ML
- PSD experiments
code/traditional-ml/Complete_analysis_of_PSD_features.ipynb - DFT experiments
code/traditional-ml/Complete_analysis_on_DFT_extracted_features.ipynb - MFCC experiments
code/traditional-ml/Complete_analysis_on_MFCC_features.ipynb - Time-domain experiments
code/traditional-ml/Time_domain_features.ipynb - Fusion experiments
code/traditional-ml/Feature_Fusion_experiments.ipynb - XGBoost classifier
code/feature extractor + classifier/CNN_Based_Feature_Extractor_XGBoost.ipynb
Deep Learning
- 1D-CNN
code/feature extractor + classifier/CNN_Based_Feature_Extractor_XGBoost.ipynb
Other components
- DroneRF data
- Detection and identification
Connections 6
- DroneRF data → PSD extraction: provides raw RF
README.md - PSD extraction → XGBoost classifier: trains on features
README.md - Feature fusion → XGBoost classifier: trains on fused features
README.md - 1D-CNN → XGBoost classifier: provides features
README.md - XGBoost classifier → Detection and identification: classifies signals
README.md - PSD extraction → Detection and identification: supports CNN classification
README.md