FHDN: a multimodal deep learning network for jamming detection in frequency hopping communications

To achieve efficient detection of complex jamming attacks in frequency hopping communication systems, this paper proposes an interference detection method based on a multimodal neural network. The method utilizes the FHDN network, which combines time-frequency images and temporal information, and introduces a temporal attention module to enhance model performance. Experimental results show that the proposed method achieves over 90% accuracy in detecting different types of jamming in multi-frequency hopping environments. Compared to traditional neural networks, the FHDN network demonstrates superior detection performance in complex jamming environments, with strong generalization ability to adapt to various types and intensities of jamming attacks, providing an efficient and reliable solution for jamming detection in frequency hopping communication systems.