Sensing RISs: Enabling Dimension-Independent CSI Acquisition for Beamforming

Jieao Zhu, Kunzan Liu, Zhongzhichao Wan, Linglong Dai, Tie Jun Cui, H. Vincent Poor

This paper presents a paradigm shift in how we manage Reconfigurable Intelligent Surfaces (RISs) for 6G communications. By moving away from traditional channel estimation and toward a physics-inspired "sensing" approach, the authors demonstrate how to acquire necessary channel information with constant overhead, regardless of how large the RIS becomes.

TL;DR

The authors propose a "Sensing RIS" architecture that uses low-cost power detectors and a simultaneous signaling trick to achieve dimension-independent channel state information (CSI) acquisition. This reduces the pilot overhead from being proportional to the number of RIS elements ($N$) to a constant value (as few as 3 pilot slots), enabling the deployment of extremely large-scale surfaces without the typical training bottleneck.

What Problem Does This Paper Address?

RISs are essentially "smart mirrors" for radio waves, but they have two major flaws: they are passive (they can't "hear" or process signals easily) and they are huge (often containing hundreds or thousands of elements).

To steer a beam toward a user, the system needs to know the Channel State Information (CSI). In traditional systems, the time required to estimate these channels (the "pilot overhead") scales linearly with the number of elements ($N$). As we move toward 6G and "extremely large-scale" RISs, this overhead becomes a dealbreaker—we would spend all our time estimating the channel and have no time left to actually send data.

Methods

The paper introduces three core innovations to break the $O(N)$ scaling law:

  1. Sensing RIS Hardware: Instead of expensive, power-hungry Radio Frequency (RF) chains, the authors propose integrating a simple, low-cost power detector into each RIS element. This allows each element to "sense" the local signal intensity.
  2. Simultaneous Rotational Signaling: Traditionally, the Base Station (BS) and the user take turns sending pilots. Here, they transmit simultaneously with a slight frequency offset. This creates a physical phenomenon called an Interference Random Field (IRF) on the surface of the RIS.
  3. Phase Estimation Algorithms: Since power detectors only measure intensity (amplitude), the phase information must be recovered from the interference pattern. The authors develop three algorithms for this:
  4. DFT Method: A simple approach using Fourier transforms.
  5. Newton-ML: A high-precision Maximum Likelihood estimator that is computationally intensive.
  6. vM-EM (von Mises-Expectation Maximization): An iterative algorithm that provides high precision without the computational burden of complex special functions.

Datasets

The paper relies on numerical simulations rather than a specific external dataset. The authors use standard statistical models for evaluation:

  • Channel Model: i.i.d. Rayleigh fading channels.
  • System Setup: A MISO system with a multi-antenna BS ($4 \times 2$ array), a single-antenna user, and an RIS with $N=200$ elements ($20 \times 10$ array).
  • Parameters: Carrier frequency $f_c = 3.5$ GHz, subcarrier bandwidth of 180 kHz, and standard thermal noise floor models.

Key Insights and Ideas

  • Phase from Intensity: The "aha moment" is realizing that while a power detector loses phase, the interference between two signals converts phase differences into measurable intensity variations. This is analogous to how optical holography works.
  • Dimension Independence: Because every element on the RIS senses the interference field at the same time, the pilot overhead is decoupled from the number of elements. Whether the RIS has 100 or 10,000 elements, you still only need the same small number of pilot slots.
  • Local Processing: Because the sensing is element-wise, the computation can be parallelized. Each element (or small group of elements) can have its own micro-controller to calculate its optimal phase shift locally, making the system highly scalable.

Key Results

  • Overhead Reduction: The proposed IRF method requires only 3 pilot slots per user. For comparison, traditional LMMSE methods required ~200 slots for a 200-element RIS.
  • Asymptotic Optimality: The authors prove that the vM-EM algorithm achieves an error decay rate of $O(\bar{\gamma}^{-1})$, which is the theoretical best possible precision for a phase estimator.
  • Spectral Efficiency: In simulations, the Sensing RIS approach achieved an ergodic spectral efficiency nearly identical to an "Oracle" system (one with perfect, cost-free channel knowledge), significantly outperforming traditional MMSE and LMMSE schemes.

Limitations and Future Work

  • Hardware Complexity: While power detectors are cheaper than RF chains, adding one to every single element still increases the cost and power consumption of the RIS compared to a purely passive version. The authors suggest "sparse sensing" (only putting sensors on some elements) as a future cost-reduction strategy.
  • Multi-User Scaling: The current work focuses primarily on a single-user scenario. While they mention using different "rotational frequencies" to distinguish multiple users, the full implementation and interference management for a high-density user environment remain to be explored.
  • Direct Link Interference: The method assumes the BS-to-RIS channel is relatively stable (quasi-static). If the environment is highly dynamic, the "two-timescale" assumption used to simplify the math might struggle.

Summary updated

Abstract

Reconfigurable intelligent surfaces (RISs) are envisioned as a potentially transformative technology for future wireless communications. However, RISs' inability to process signals and the attendant increased channel dimension have brought new challenges to RIS-assisted systems, including significantly increased pilot overhead required for channel estimation. To address these problems, several prior contributions that enhance the hardware architecture of RISs or develop algorithms to exploit the channels' mathematical properties have been made, where the required pilot overhead is reduced to be proportional to the number of RIS elements. In this paper, we propose a dimension-independent channel state information (CSI) acquisition approach in which the required pilot overhead is independent of the number of RIS elements. Specifically, in contrast to traditional signal transmission methods, where signals from the base station (BS) and the users are transmitted in different time slots, we propose a novel method in which signals are transmitted from the BS and the user simultaneously during CSI acquisition. With this method, an electromagnetic interference random field (IRF) will be induced on the RIS, and we propose the structure of sensing RIS to capture its features. Moreover, we develop three algorithms for parameter estimation in this system, in which one of the proposed vM-EM algorithm is analyzed with the fixed-point perturbation method to obtain an asymptotic achievable bound. In addition, we also derive the Cramér-Rao lower bound (CRLB) and an asymptotic expression for characterizing the best possible performance of the proposed algorithms. Simulation results verify that our proposed signal transmission method and the corresponding algorithms can achieve dimension-independent CSI acquisition for beamforming.
BibTeX
@article{zhu2023sensing,
  author = {Jieao Zhu and Kunzan Liu and Zhongzhichao Wan and Linglong Dai and Tie Jun Cui and H. Vincent Poor},
  title = {{Sensing RISs: Enabling Dimension-Independent CSI Acquisition for Beamforming}},
  journal = {IEEE Trans. Information Theory},
  year = {2023},
  doi = {10.1109/tit.2023.3243836},
}

Loading related content...

Figure