Performance Analysis of Systems with Coupled and Decoupled RISs

Dominik Semmler, Josef A. Nossek, Michael Joham, Wolfgang Utschick

TL;DR

This paper introduces a more efficient $O(N^3)$ element-wise algorithm to optimize Reconfigurable Intelligent Surfaces (RIS) while accounting for mutual coupling between elements. Furthermore, the authors propose using hardware-based "decoupling networks" that mathematically eliminate coupling effects, enabling closed-form global optimization and revealing a "super-quadratic" array gain that scales with $N^4$.

What Problem Does This Paper Address?

Most research on Reconfigurable Intelligent Surfaces (RIS) assumes a simplified model where each reflecting element acts independently. In reality, because RIS elements are packed tightly together to save space, they exhibit mutual coupling—the electromagnetic interaction where one element's state affects its neighbors.

Ignoring mutual coupling leads to inaccurate performance predictions and suboptimal beamforming. However, accounting for it is difficult: existing "coupling-aware" algorithms are either based on approximations (like Neumann series) or are computationally expensive ($O(N^4)$ complexity). Moreover, the optimization landscape for coupled RIS is non-convex, meaning standard gradient-based algorithms often get stuck in local optima, failing to find the best possible configuration.

Methods

The authors tackle the problem from two angles: a new numerical algorithm and a hardware-based architectural change.

  1. Efficient Element-Wise Algorithm: The authors derive a new algorithm that updates the reactance of each RIS element successively using a closed-form solution. By applying the matrix inversion lemma, they avoid the need for full matrix inversions at every step. This reduces the computational complexity from $O(N^4)$ to $O(N^3)$, making it as efficient as existing approximation-based methods while remaining exact.
  2. Decoupling Networks: The paper introduces the concept of a "decoupling network" (a lossless, reciprocal multiport network) placed between the RIS antennas and the adjustable loads. Mathematically, this network "diagonalizes" the system's impedance matrix.
  3. Closed-Form Optimization: By using the decoupling network, the complex coupled system is transformed back into a structure identical to a non-coupled system. This allows the authors to solve for the optimal RIS configuration (phase shifts) in closed-form, guaranteed to reach the global optimum.

Datasets

This paper does not use external datasets. It relies on theoretical derivations and numerical simulations based on standard electromagnetic models (impedance matrices) for Uniform Linear Arrays (ULA) in Line-of-Sight (LOS) scenarios.

Key Insights and Ideas

  • Coupling as an Opportunity: While usually viewed as a nuisance, the authors show that if mutual coupling is handled correctly via decoupling networks, it can be leveraged. In an "end-fire" scenario (where the signal travels along the plane of the array), the array gain can scale as $N^4$ (super-quadratic) rather than the usual $N^2$.
  • Circumventing Local Optima: Iterative algorithms (gradient descent, etc.) frequently fail in coupled RIS scenarios because they get trapped in suboptimal local maxima. The decoupling network approach bypasses this entirely by providing a direct, closed-form path to the global optimum.
  • The "Loss" Caveat: The $N^4$ gain is theoretically possible for lossless antennas. However, the authors demonstrate that even small amounts of Ohmic loss (heat) significantly dampen this effect as the spacing between elements approaches zero.

Key Results

  • Complexity Reduction: The proposed element-wise algorithm achieves a 10x speedup in scaling ($O(N^4) \to O(N^3)$) compared to previous exact methods.
  • Performance Superiority: In simulations with $N=4$ elements at small spacings ($< 0.25\lambda$), the "Decoupled" approach significantly outperformed all other iterative mutual-coupling-aware algorithms (like Z-OPT or Gradient methods), which all got stuck in the same local optimum.
  • Array Gain: The paper analytically proves and numerically confirms that for front-fire directions, the RIS array gain is exactly the square of a conventional transmit array gain. For end-fire directions with $d \to 0$, the gain approaches $N^4$.

Limitations and Future Work

  • Hardware Complexity: The primary limitation is the physical implementation of the decoupling network. A fully connected decoupling network requires a number of components that scales quadratically ($N^2$) with the number of RIS elements. For a surface with hundreds of elements, this becomes prohibitively complex and bulky.
  • Future Direction: The authors suggest that future research should focus on partially connected decoupling networks. These would only account for coupling between immediate neighbors, allowing the hardware complexity to scale linearly ($O(N)$) while still capturing the most significant coupling effects.

Summary updated

Abstract

We analyze and compare different methods for handling the mutual coupling in RIS-aided communication systems. A new mutual coupling aware algorithm is derived where the reactance of each element is updated successively with a closed-form solution. In comparison to existing element-wise methods, this approach leads to a considerably reduced computational complexity. Furthermore, we introduce decoupling networks for the RIS array as a potential solution for handling mutual coupling. With these networks, the system model reduces to the same structure as when no mutual coupling were present. Including decoupling networks, we can optimize the channel gain of a RIS-aided SISO system in closed-form which allows to analyze the scenario under mutual coupling analytically and to draw connections to the conventional transmit array gain. In particular, a super-quadratic channel gain can be achieved which scales as N^4 where N is the number of RIS elements.
BibTeX
@inproceedings{semmler2023performance,
  author = {Dominik Semmler and Josef A. Nossek and Michael Joham and Wolfgang Utschick},
  title = {{Performance Analysis of Systems with Coupled and Decoupled RISs}},
  booktitle = {ISWCS},
  year = {2023},
  doi = {10.1109/iswcs61526.2024.10639116},
}

Loading related content...

Figure