Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Fast Iterative Configuration of Reconfigurable Intelligent Surfaces in mmWave Systems

Anna V. Guglielmi, Stefano Tomasin

TL;DR

This work tackles the RIS configuration overhead in mmWave single-user MIMO by introducing the Fast Iterative Configuration (FIC) protocol, which exploits the typically small number of mmWave paths. FIC partitions the RIS into $L$ subsets aligned with channel taps and, for each subset, performs an iterative search over a discretized 2D AoA/AoD grid to maximize the end-to-end rate $C(\mathbf{Q})$, refining the grid around the best angles and using discrete phase shifts. The method handles both single-path and multi-path scenarios, with a structured estimation-time cost $T_{\rm FIC} = T_0 M (L_1 + P\sum_{i=2}^I L_i)$ and the option of multiple starting points to avoid local maxima. Numerical results in a mmWave urban model show that FIC converges to near-optimal RIS configurations much faster than exhaustive approaches, reducing overhead while maintaining high achievable rates, making RIS-aided mmWave links more practical in dynamic deployments.

Abstract

Reconfigurable intelligent surfaces (RISs) are a promising solution to improve the coverage of cellular networks, thanks to their ability to steer impinging signals in desired directions. However, they introduce an overhead in the communication process since the optimal configuration of a RIS depends on the channels to and from the RIS, which must be estimated. In this paper, we propose a novel fast iterative configuration (FIC) protocol to determine the optimal RIS configuration that exploits the small number of paths of millimetre-wave (mmWave) channels and an adaptive choice of the explored RIS configurations. In particular, we split the elements of the RIS into a number of subsets equal to the number of channel taps. For each subset, then an iterative procedure finds at each iteration the optimal RIS configuration in a codebook exploring a two-dimensional grid of possible angles of arrival and departure of the path at the RIS. Over the iterations, the grid is made finer around the point identified in previous iterations. Numerical results obtained using an urban channel model confirm that the proposed solution is fast and provides a configuration close to the optimal in a shorter time than other existing approaches.

Fast Iterative Configuration of Reconfigurable Intelligent Surfaces in mmWave Systems

TL;DR

This work tackles the RIS configuration overhead in mmWave single-user MIMO by introducing the Fast Iterative Configuration (FIC) protocol, which exploits the typically small number of mmWave paths. FIC partitions the RIS into subsets aligned with channel taps and, for each subset, performs an iterative search over a discretized 2D AoA/AoD grid to maximize the end-to-end rate , refining the grid around the best angles and using discrete phase shifts. The method handles both single-path and multi-path scenarios, with a structured estimation-time cost and the option of multiple starting points to avoid local maxima. Numerical results in a mmWave urban model show that FIC converges to near-optimal RIS configurations much faster than exhaustive approaches, reducing overhead while maintaining high achievable rates, making RIS-aided mmWave links more practical in dynamic deployments.

Abstract

Reconfigurable intelligent surfaces (RISs) are a promising solution to improve the coverage of cellular networks, thanks to their ability to steer impinging signals in desired directions. However, they introduce an overhead in the communication process since the optimal configuration of a RIS depends on the channels to and from the RIS, which must be estimated. In this paper, we propose a novel fast iterative configuration (FIC) protocol to determine the optimal RIS configuration that exploits the small number of paths of millimetre-wave (mmWave) channels and an adaptive choice of the explored RIS configurations. In particular, we split the elements of the RIS into a number of subsets equal to the number of channel taps. For each subset, then an iterative procedure finds at each iteration the optimal RIS configuration in a codebook exploring a two-dimensional grid of possible angles of arrival and departure of the path at the RIS. Over the iterations, the grid is made finer around the point identified in previous iterations. Numerical results obtained using an urban channel model confirm that the proposed solution is fast and provides a configuration close to the optimal in a shorter time than other existing approaches.
Paper Structure (8 sections, 20 equations, 7 figures)

This paper contains 8 sections, 20 equations, 7 figures.

Table of Contents

  1. Introduction
  2. System model
  3. Channel Estimation Protocol
  4. ris Configuration With A Single-Path Channel
  5. ris Configuration With Multi-Path Channel
  6. Estimation Time
  7. Numerical results
  8. Conclusions

Figures (7)

  • Figure 1: System model.
  • Figure 2: Example of multi-path scenario with $M=3$.
  • Figure 3: Average normalized achievable rate loss varying $L_i$ with $L_G = L_H = 2$ paths and element/antenna spacing $d=\frac{\lambda}{2}$.
  • Figure 4: Average normalized achievable rate loss varying $L_i$ with $L_G = L_H = 3$ paths and element/antenna spacing $d=\frac{\lambda}{2}$.
  • Figure 5: Average normalized achievable rate loss varying the number of channel estimates per iteration $K = 1$, 2, 3, 4, and 5 with $L_i=9$$L_G = L_H = 3$ paths and element/antenna spacing $d=\frac{\lambda}{2}$
  • ...and 2 more figures