BRSR-OpGAN: Blind radar signal restoration using operational generative adversarial network

Area of Science:

• Electrical Engineering
• Signal Processing
• Artificial Intelligence

Background:

• Existing radar signal restoration methods often address isolated noise types and fixed signal-to-noise ratio (SNR) levels.
• Real-world radar signals face complex corruptions from multiple sources like jamming, interference, and sensor noise, varying in type and intensity.

Purpose of the Study:

• To develop a robust radar signal restoration technique capable of handling diverse and severe artifacts in real-world scenarios.
• To introduce the Blind Radar Signal Restoration using an Operational Generative Adversarial Network (BRSR-OpGAN) for enhanced radar signal quality.

Main Methods:

• Utilized a novel Operational Generative Adversarial Network (OpGAN) with a dual domain loss (temporal and spectral).
• Employed 1D Operational GANs with a generative neuron model optimized for blind restoration of corrupted radar signals.
• Evaluated the approach on a baseline and a newly curated Blind Radar Signal Restoration (BRSR) dataset simulating real-world conditions.

Main Results:

• Achieved an average signal-to-noise ratio (SNR) improvement of over 15.1 dB on the baseline dataset and 14.3 dB on the BRSR dataset.
• Demonstrated robust performance across various SNR values and artifact types, significantly outperforming existing methods.
• Confirmed the approach's capability for real-time application, even on resource-constrained platforms.

Conclusions:

• The BRSR-OpGAN method offers effective and computationally efficient blind radar signal restoration for real-world applications.
• The proposed technique significantly enhances radar signal quality by adapting dynamically to a wide range of artifact characteristics.
• This pilot study highlights the potential of advanced AI techniques for improving the reliability and performance of radar systems.