Jamming Detection and Classification on Physical Layer with a Minimized CNN Model

Fink L, Reißland T, Maiwald T, Franchi N (2025)

Publication Type: Conference contribution, Conference Contribution

Publication year: 2025

Event location: Rabat MA

Abstract

Wireless communication networks are vulnerable to attacks via the air interface. So-called jamming attacks can negatively affect a wireless network and their participating wireless nodes, or even lead to a denial of service. In this work a strategy for jamming detection and classification is proposed. It combines targeting an embedded implementation of a convolutional neural network (CNN) that is used for the classification task and aiming high classification accuracy. Unlike other approaches that are mainly about detecting a jammer, this work investigates three scenarios i) classifying the strength of the jammer as well as its signal type; ii) classification of the signal type of the jammer regardless of its signal power; iii) detection of a jammer. Using 2D histograms of demodulated IQ samples in combination with a CNN in a non-line-of-sight (NLOS) scenario, an accuracy of 91.0% is achieved with a parameter reduction factor of 50 compared to the base line CNN architecture. Transferring the reduced model to a line-of-sight (LOS) dominant scenario leads to an accuracy of 99.1 %. For jamming detection, an accuracy of 99.5% is achieved for the NLOS scenario and only correct predictions for the LOS scenario. The results demonstrate that a complex jamming detection as well as jamming classification can be executed with a minimized CNN architecture using IQ data directly on the physical layer of a communication system.

Authors with CRIS profile

Lucas Fink Lehrstuhl für Intelligente Technische Elektronik und Systeme Torsten Reißland Lehrstuhl für Intelligente Technische Elektronik und Systeme Timo Maiwald Lehrstuhl für Intelligente Technische Elektronik und Systeme Norman Franchi Lehrstuhl für Intelligente Technische Elektronik und Systeme

How to cite

APA:

Fink, L., Reißland, T., Maiwald, T., & Franchi, N. (2025). Jamming Detection and Classification on Physical Layer with a Minimized CNN Model. In Proceedings of the 8th International Conference on Advanced Communication Technologies and Networking (CommNet). Rabat, MA.

MLA:

Fink, Lucas, et al. "Jamming Detection and Classification on Physical Layer with a Minimized CNN Model." Proceedings of the 8th International Conference on Advanced Communication Technologies and Networking (CommNet), Rabat 2025.

BibTeX: Download