Blind Beamforming for Interference Suppression Using Pre-trained Diffusion Models

Blind adaptive beamforming plays a crucial role from both the multi-antenna wireless communications and radar target detection perspectives when unknown mutual coupling effects and miscalibrated array elements are involved. We propose a new blind beamformer technique, called Score-based Weight Estimation for Eliminating Perturbations (SWEEP), which utilizes a pre-trained diffusion model for the signal of interest in order to optimize array weights for interference suppression in blind source separation settings. The score estimated by the diffusion model in combination with the least squares beamformer is used to guide the weights to form the optimal beam pattern. Extensive simulations show that SWEEP can achieve optimal interference suppression performance in some cases and improve the array signal-to-interference-plus-noise ratio by 3.5 dB compared to the state-of-the-art technique in high SNR conditions.