Multi-User Beamforming and Transmission Based on Intelligent Reflecting Surface

Yihong Liu, Lei Zhang, Muhammad Ali ImranIEEE Trans. Wireless Communications 2022 · DOI

Abstract

Intelligent Reflecting Surfaces (IRS) enable programmable wireless environments by shifting the phase of electromagnetic waves via low-cost electronic elements. This paper explores the multi-user beamforming challenge, where a single IRS serves multiple transceiver pairs simultaneously. The authors identify a critical "redundant beam" phenomenon that inevitably occurs during multi-user multiplexing, leading to a power efficiency reduction of up to $1/N$ for $N$ users. Using a Multi-User Linearly Constrained Minimum Variance (MU-LCMV) framework, the study demonstrates effective interference nulling and provides a robust optimization algorithm for practical phase-only control.

Introduction

In traditional wireless communication, the environment is a "black box" where signals bounce off obstacles in unpredictable ways, leading to fading and interference. While MIMO (Multiple-Input Multiple-Output) technology mitigates these effects, it remains limited by the natural physics of the channel. Intelligent Reflecting Surfaces (IRS) change this paradigm by making the environment programmable. By precisely controlling the phase shifts of tiny elements on a surface, reflections can be "steered" toward specific receivers.

While single-user IRS applications are well-documented, this paper tackles the multi-user challenge: configuring a single IRS to serve multiple transceiver pairs without mutual interference. This requires managing complex beam steering while simultaneously nulling interference in a shared physical space.

Methodology

The authors model a system comprising $N$ pairs of transmitters and receivers and one IRS with $M$ elements. To manage the task of steering multiple beams while nulling interference, they adapt the Multi-User Linearly Constrained Minimum Variance (MU-LCMV) technique from radar applications.

System model of multi-user beamforming with an Intelligent Reflecting Surface, illustrating the generation of intended and redundant beams.

The optimization problem is approached through two primary lenses:

  1. Amplitude-Unconstrained Phase-Continuous (AUPC): A theoretical closed-form solution where elements can adjust both phase and amplitude, serving as a performance benchmark.
  2. Amplitude-Constrained Phase-Continuous (ACPC): A realistic scenario where elements have a fixed reflection amplitude and only the phase is tunable. This non-convex problem is solved using Sequential Quadratic Programming (SQP), which proves more robust than standard genetic algorithms for this task.

To verify these models, the authors conducted extensive numerical simulations using a simulated IRS of $16 \times 16$ (256) elements with half-wavelength spacing.

Key Insights and Results

The Redundant Beam Phenomenon

The most striking insight is that satisfying constraints for $N$ independent user pairs inevitably creates "ghost" or redundant beams. For $N$ pairs, up to $N-1$ redundant beams are generated. These beams carry signal power away from intended receivers, resulting in a "power efficiency tax" where efficiency can drop to $1/N$ in worst-case scenarios.

Geometric configuration of a Uniform Rectangular Array (URA) IRS defining the incident and reflected angles used for beam steering derivations.

Performance and Geometric Limits

  • Mathematical Accuracy: Simulated beampatterns perfectly matched the authors' derived formulas for redundant beam directions, such as predicting a redundant beam at specific angular coordinates for a two-user setup.
  • Interference Suppression: The MU-LCMV approach successfully suppressed interference to levels as low as -40 dB, effectively creating interference-free zones.
  • Angular Sensitivity: Performance is highly dependent on the angular separation of users. If transmitters or receivers are too close in angular direction, the IRS cannot effectively distinguish between them, leading to a "self-contradicted" optimization problem where interference cannot be nulled.
  • Mutual Coupling: Smaller element spacing was found to increase side lobe levels, potentially increasing interference in unmanaged directions.

Conclusion

This research provides a rigorous framework for multi-user IRS deployments, highlighting the fundamental trade-off between multiplexing gain and power efficiency due to redundant beams. While the SQP-based solution offers a computationally viable path for real-time operations, practical hurdles remain. Future work must address the requirement for Global Channel State Information (CSI) and explore distributed IRS deployments to overcome the geometric limitations of a single surface. Additionally, the potential to leverage redundant beams for multicast scenarios offers a promising avenue for improving overall system utility.

Summary updated

BibTeX
@article{liu2022multi,
  author = {Yihong Liu and Lei Zhang and Muhammad Ali Imran},
  title = {{Multi-User Beamforming and Transmission Based on Intelligent Reflecting Surface}},
  journal = {IEEE Trans. Wireless Communications},
  year = {2022},
  doi = {10.1109/twc.2022.3157808},
}
Figure