AI-Based Radio Frequency Interference Signals Detection and Classification
Fuente:
Zenodo
Gespeichert in:
| Hauptverfasser: | , , , , , |
|---|---|
| Format: | Recurso digital |
| Veröffentlicht: |
Zenodo
2026
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866901314232909824 |
|---|---|
| author | Mrs. I. Kamalamma Sk. Rizwana Y. Sri Poojitha S. Tanmai S. Bhagya Sri S. Raj Praveen |
| author_facet | Mrs. I. Kamalamma Sk. Rizwana Y. Sri Poojitha S. Tanmai S. Bhagya Sri S. Raj Praveen |
| contents | Abstract - Wireless spectrum congestion and radio astronomy observations are really hurt by Radio Frequency Interference. This is because Radio Frequency Interference degrades the quality of the signal we receive and the reliability of the system. We cannot use the way of finding Radio Frequency Interference by hand because it is not good enough for the crowded spectrum environments we have today. This paper is about a way to automatically find and classify Radio Frequency Interference using machine learning that is supervised. We look at the features of the signal that're important. This includes the frequency range, how strong the signal is at that moment, how the signal changes over time, how much bandwidth is being used and the shape of the spectrum. We take these features from recordings that are segmented and use them to train five classifiers: Logistic Regression, k-Nearest Neighbours, Decision Tree, Support Vector Machine and Random Forest. When we test these classifiers we find that Random Forest is the best at finding Radio Frequency Interference. It is very good at this. It has an accuracy of 96.8 percent, precision of 96.5 percent, recall of 96.9 percent and F1-score of 96.7 percent across four types of interference. We then use the trained model in a time streaming pipeline and it works very fast. It can make a decision in under 50 milliseconds per segment. This shows that the model is good enough to be used in situations both for communication and for radio astronomy and it can be used with Radio Frequency Interference. |
| format | Recurso digital |
| id | zenodo_https___doi_org_10_5281_zenodo_19631119 |
| institution | Zenodo |
| language | |
| publishDate | 2026 |
| publisher | Zenodo |
| record_format | zenodo |
| spellingShingle | AI-Based Radio Frequency Interference Signals Detection and Classification Mrs. I. Kamalamma Sk. Rizwana Y. Sri Poojitha S. Tanmai S. Bhagya Sri S. Raj Praveen communication feature extraction spectrum monitoring Radio Frequency Interference machine learning Random Forest signal classification wireless Abstract - Wireless spectrum congestion and radio astronomy observations are really hurt by Radio Frequency Interference. This is because Radio Frequency Interference degrades the quality of the signal we receive and the reliability of the system. We cannot use the way of finding Radio Frequency Interference by hand because it is not good enough for the crowded spectrum environments we have today. This paper is about a way to automatically find and classify Radio Frequency Interference using machine learning that is supervised. We look at the features of the signal that're important. This includes the frequency range, how strong the signal is at that moment, how the signal changes over time, how much bandwidth is being used and the shape of the spectrum. We take these features from recordings that are segmented and use them to train five classifiers: Logistic Regression, k-Nearest Neighbours, Decision Tree, Support Vector Machine and Random Forest. When we test these classifiers we find that Random Forest is the best at finding Radio Frequency Interference. It is very good at this. It has an accuracy of 96.8 percent, precision of 96.5 percent, recall of 96.9 percent and F1-score of 96.7 percent across four types of interference. We then use the trained model in a time streaming pipeline and it works very fast. It can make a decision in under 50 milliseconds per segment. This shows that the model is good enough to be used in situations both for communication and for radio astronomy and it can be used with Radio Frequency Interference. |
| title | AI-Based Radio Frequency Interference Signals Detection and Classification |
| topic | communication feature extraction spectrum monitoring Radio Frequency Interference machine learning Random Forest signal classification wireless |
| url | https://doi.org/10.5281/zenodo.19631119 |