论文信息 - Towards Adversarial and Unintentional Collisions Detection Using Deep Learning

Towards Adversarial and Unintentional Collisions Detection Using Deep Learning

We introduce a set of techniques to achieve transfer learning from computer vision to RF spectrum analysis. In this paper, we demonstrate the usefulness of this approach to scale the learning, accuracy, and efficiency of detection of adversarial and unintentional communications collisions using VGG-16. We achieve high accuracy (94% collisions detected) on a DARPA Spectrum Collaboration Challenge (SC2) dataset.

[1] Hamed Haddadi,et al. Deep Learning in Mobile and Wireless Networking: A Survey , 2018, IEEE Communications Surveys & Tutorials.

[2] Tien Dang Vo-Huu,et al. Interleaving Jamming in Wi-Fi Networks , 2016, WISEC.

[3] T. Charles Clancy,et al. Convolutional Radio Modulation Recognition Networks , 2016, EANN.

[4] Koorosh Firouzbakht,et al. On the Performance of Adaptive Packetized Wireless Communication Links Under Jamming , 2013, IEEE Transactions on Wireless Communications.

[5] Li Fei-Fei,et al. ImageNet: A large-scale hierarchical image database , 2009, CVPR.

[6] Andrew Zisserman,et al. Very Deep Convolutional Networks for Large-Scale Image Recognition , 2014, ICLR.

[7] Tien Dang Vo-Huu,et al. Fingerprinting Wi-Fi Devices Using Software Defined Radios , 2016, WISEC.

[8] Koorosh Firouzbakht,et al. Packetized wireless communication under jamming, a constrained bimatrix game , 2014, 2014 IEEE Global Communications Conference.

[9] Alex Krizhevsky,et al. Learning Multiple Layers of Features from Tiny Images , 2009 .