Intelligent Reflecting Surface-Assisted Adaptive Beamforming for Blind Interference Suppression

In this paper, we consider the problem of adaptive beamforming (ABF) for intelligent reflecting surface (IRS)-assisted systems, where a single antenna receiver, aided by a close-by IRS, tries to decode signals from a legitimate transmitter in the presence of multiple unknown interference signals. Such a problem is formulated as an ABF problem with the objective of minimizing the average received signal power subject to certain constraints. Unlike canonical ABF in array signal processing, we do not have direct access to the covariance matrix that is needed for solving the ABF problem. Instead, for our problem, we only have some quadratic compressive measurements of the covariance matrix. To address this challenge, we propose a sample-efficient method that directly solves the ABF problem without explicitly inferring the covariance matrix. Compared with the methods which explicitly recover the covariance matrix from its quadratic compressive measurements, our proposed method achieves a substantial improvement in terms of sample efficiency. Simulation results show that our method, using a small number of measurements, can effectively nullify the interference signals and enhance the signal-to-interference-plus-noise ratio (SINR).