Bridging Digital Deception and Physical Fidelity: A Spatio-Temporal Attack on Cognitive Radars

Key points

• To overcome these bottlenecks, we propose a physical prior-driven Segmented Spatio-Temporal Attack (SSTA).1
• First, a global-local feature decoupling segmentation mechanism interrupts the adversarial gradients' dependence on long-range sequence coherence, bridging the cross-architecture feature gap.1
• Second, moving beyond static norm constraints devoid of physical semantics, we abstract channel distortions into differentiable spatio-temporal operators, integrating them into momentum iterations via an Expectation over Transformation (EoT) strategy.1
• Comprehensive cross-evaluations across 9 heterogeneous models spanning CNNs, RNNs, and Transformers demonstrate that SSTA outperforms the state-of-the-art baseline by 10.36% in Average Black-Box Attack Success Rate.1

Sources (1)

• [1]Bridging Digital Deception and Physical Fidelity: A Spatio-Temporal Attack on Cognitive Radars
  arXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 20, 03:52 AM

  As cognitive radars evolve towards the end-to-end processing of one-dimensional complex baseband sequences, deep networks face severe adversarial threats.

  To overcome these bottlenecks, we propose a physical prior-driven Segmented Spatio-Temporal Attack (SSTA).

Extractive summary: sentences quoted from the sources.