Beyond Diagonal RIS: Passive Maximum Ratio Transmission and Interference Nulling Enabler

Hamad Yahya, Hongyu Li, Matteo Nerini, Bruno Clerckx, Mérouane DebbahIEEE Open J. Commun. Soc 2024

TL;DR

This paper introduces a two-stage design for Beyond Diagonal Reconfigurable Intelligent Surfaces (BD-RIS) that allows these passive surfaces to perform complex beamforming tasks—like maximizing signal strength or nulling interference—traditionally handled by power-hungry base stations. By interconnecting reflecting elements, BD-RIS achieves significantly higher data rates and better interference management with far fewer elements than conventional RIS technologies.

What Problem Does This Paper Address?

Reconfigurable Intelligent Surfaces (RIS) are a cornerstone of 6G research, acting as "smart mirrors" that redirect wireless signals to improve coverage. However, conventional Diagonal RIS (D-RIS) has a major limitation: each element operates independently, meaning the surface can only manipulate the phase of a signal. This often results in a "double fading" effect that limits performance.

Furthermore, in multi-user scenarios, the Base Station (BS) usually bears the entire burden of "precoding"—mathematically tweaking signals to ensure they don't interfere with other users. This requires significant computational power and complex hardware at the BS. The authors address these gaps by moving some of this "intelligence" to the RIS itself using a Beyond Diagonal architecture, where elements are interconnected to allow control over both phase and amplitude.

Methods

The authors propose a two-stage design to decouple the complex optimization of the Base Station and the RIS:

  1. Stage 1: BD-RIS Scattering Matrix Design
    • Passive MRT (Maximum Ratio Transmission): A closed-form solution designed to maximize the received signal power at users.
  2. Passive Interference Nulling: An Alternating Optimization (AO) algorithm that treats the RIS as a passive "Zero Forcing" (ZF) enabler, cancelling out inter-user interference before the signal even reaches the users.
  3. They evaluate three hardware architectures: Single-connected (standard D-RIS), Group-connected, and Fully-connected (where every element is linked).
  4. Stage 2: Base Station Precoding
    • Once the RIS is set, the BS applies simpler power allocation (Uniform or Optimized) or standard ZF precoding.

To ensure the RIS remains passive and lossless, the authors use a SVD-based symmetric unitary projection, a mathematical technique that maps the ideal "beyond diagonal" matrix onto a set of constraints that physical, reactive circuits can actually implement.

Datasets

As a theoretical communication theory paper, no external datasets were used. Instead, the authors evaluated their algorithms using Monte-Carlo simulations with:

  • i.i.d. Rayleigh fading channels to model the wireless environment.
  • Standard path-loss models (distance-dependent) at a 2.4 GHz operating frequency.
  • Comparisons against established benchmarks like "Joint Design" (highly complex) and "Max-F" (Frobenius norm-based) designs.

Key Insights and Ideas

  • Interconnection is Key: By connecting RIS elements, the scattering matrix is no longer restricted to being diagonal. This allows the RIS to mimic the behavior of a multi-antenna transmitter, effectively "strengthening" the channel.
  • Passive Interference Management: The most striking insight is that a passive surface can perform interference nulling. This relaxes the requirements for the Base Station, allowing it to use simpler, more energy-efficient hardware.
  • Efficiency in Numbers: BD-RIS requires significantly fewer reflecting elements ($N$) to null interference for $K$ users. For a fully-connected architecture, you only need $N \geq 2K - 1$, whereas conventional D-RIS requires $N \geq 2K(K-1)$.
  • Complexity Trade-off: While Fully-connected BD-RIS offers the best performance and fastest algorithmic convergence, Group-connected architectures provide a "sweet spot," offering most of the gains with much lower circuit complexity.

Key Results

  • Sum Rate Gains: In a scenario with 5 users and 64 reflecting elements, the proposed BD-RIS design achieved a sum rate of ~27.7 bps/Hz, nearly reaching the performance of the computationally "optimal" joint design (~28.3 bps/Hz) but with much lower complexity.
  • Convergence Speed: The Fully-connected BD-RIS algorithm converges to an interference-nulling state ($10^{-8}$ error) in significantly fewer iterations than the single-connected (D-RIS) version.
  • SNR Robustness: At high Signal-to-Noise Ratios (SNR), passive interference nulling at the RIS was found to be more robust than passive MRT, as it inherently eliminates the interference bottleneck that usually saturates data rates.

Limitations and Future Work

  • Hardware Complexity: The paper acknowledges that the "Fully-connected" architecture requires a complex network of reconfigurable impedances ($N(N+1)/2$ connections), which may be difficult to fabricate at scale.
  • CSI Requirements: While the authors simplified the channel estimation process, the system still relies on accurate Channel State Information (CSI). Future work aims to generalize these designs to account for imperfect CSI.
  • Multi-Antenna Users: This study focused on MU-MISO (single-antenna users). Extending the "Beyond Diagonal" logic to users with multiple antennas (MU-MIMO) remains an open challenge.

Summary updated

Abstract

Beyond diagonal reconfigurable intelligent surfaces (BD-RIS) generalizes and goes beyond conventional diagonal reconfigurable intelligent surfaces (D-RIS) by interconnecting elements to generate beyond diagonal scattering matrices, which significantly strengthen the wireless channels. In this work, we use BD-RIS for passive multiuser beamforming in multiuser multiple-input-single-output (MU-MISO) systems. Specifically, we design the scattering matrix of BD-RIS to either maximize the sum received signal power at the users following maximum ratio transmission (MRT), or to nullify the interference at the users following zero forcing (ZF). Furthermore, we investigate uniform/optimized power allocation and ZF precoding at the base station (BS). Numerical results show that BD-RIS improves the interference nulling capability and sum rate with fewer reflecting elements (REs) compared to D-RIS. In addition, at moderate to high signal to noise ratios (SNRs), passive interference nulling reduces the complexity at the BS by relaxing the need for precoding or water-filling power allocation design. Furthermore, the passive MRT with ZF precoding achieves a tight sum rate performance to the joint design considering MU-MISO scenarios with many REs while maintaining low computational complexity and simplifying the channel estimation.
BibTeX
@article{yahya2024beyond,
  author = {Hamad Yahya and Hongyu Li and Matteo Nerini and Bruno Clerckx and Mérouane Debbah},
  title = {{Beyond Diagonal RIS: Passive Maximum Ratio Transmission and Interference Nulling Enabler}},
  journal = {IEEE Open J. Commun. Soc.},
  year = {2024},
}
Figure