Conference proceeding
Generative AI for Wireless Interference Modeling: Text-Controlled Waveform Synthesis Using Stable Diffusion
2025 IEEE Wireless and Microwave Technology Conference (WAMICON), pp 1-4
14 Apr 2025
Abstract
Modern communication systems face challenges in replicating jammer behaviors when interference parameters are unknown. This study explores the use of Stable Diffusion, fine-tuned with LoRA and DreamBooth, to generate high-level, text-based abstractions that characterize unknown interference patterns. By utilizing a single jammer and receiver, and providing spectrogram images of the received signals, the model is able to generate reasonable representations of the interference. Unlike traditional methods that rely on predefined signal models, our approach enables the synthesis of interference spectrograms without prior knowledge of jammer parameters or transmitted signals. By leveraging diffusion models, we provide a scalable and flexible framework for approximating and representing unknown interference, offering a novel direction for abstracted interference modeling in software-defined radio (SDR) systems.
Metrics
3 Record Views
Details
- Title
- Generative AI for Wireless Interference Modeling: Text-Controlled Waveform Synthesis Using Stable Diffusion
- Creators
- Matthew Tylek - Drexel UniversityKeith Truongcao - Drexel UniversityMd Shakir Hossain - Drexel UniversityKapil R. Dandekar - Drexel University
- Publication Details
- 2025 IEEE Wireless and Microwave Technology Conference (WAMICON), pp 1-4
- Publisher
- IEEE
- Number of pages
- 4
- Grant note
- CNS-1816387 / National Science Foundation (NSF) (10.13039/100000001)
- Resource Type
- Conference proceeding
- Language
- English
- Academic Unit
- Electrical and Computer Engineering
- Web of Science ID
- WOS:001508542600039
- Scopus ID
- 2-s2.0-105007519469
- Other Identifier
- 991022054115304721