Blind Passive Beamforming for MIMO System

2025-06-01

AI Summary

Passive beamforming for RIS-assisted MIMO systems under discrete phase constraints suffers from heavy reliance on accurate channel state information (CSI), limiting robustness in practical deployments. Method: This paper proposes the first fully blind passive beamforming framework—requiring no CSI whatsoever—that autonomously estimates dominant wireless environmental features and optimizes discrete RIS phase configurations solely via randomized received-power sampling and statistical learning. The method integrates randomized sampling analysis, high-dimensional statistical modeling, and discrete optimization to achieve low overhead and strong adaptability. Contribution/Results: Evaluated on a live 5G commercial network, the proposed scheme significantly outperforms state-of-the-art CSI-dependent approaches in terms of throughput gain, robustness to channel variations, and deployment flexibility. It validates both theoretical efficacy and engineering feasibility, establishing a new paradigm for “perception-free” RIS deployment in real-world wireless systems.

Technology Category

• Search and Optimization: Sampling/Simulation-based Search
• Intelligent Robots: State Estimation
• Reasoning under Uncertainty: Stochastic Optimization

Application Category

• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation
• Systems and Infrastructure for Web, Mobile and WoT: Energy management for devices in mobile Web and WoT environments
• Security and Privacy: Large-scale security measurements

Abstract

Passive beamforming for the intelligent surface (IS)-aided multiple-input multiple-output (MIMO) communication is a difficult nonconvex problem. It becomes even more challenging under the practical discrete constraints on phase shifts. Unlike most of the existing approaches that rely on the channel state information (CSI), this work advocates a blind beamforming strategy without any CSI. Simply put, we propose a statistical method that learns the main feature of the wireless environment from the random samples of received signal power. Field tests in the 5G commercial network demonstrate the superiority of the proposed blind passive beamforming method.

Problem

• Blind passive beamforming without CSI for MIMO systems
• Overcoming nonconvex discrete phase shift constraints
• Statistical learning from signal power samples

Innovation

• Blind beamforming without channel state information
• Statistical learning from random signal samples
• Discrete phase shift constraints in MIMO