PaperlayerView exact revision on arXiv

Source-linked AI summary

Reconfigurable-Intelligent-Surface Empowered Wireless Communications: Challenges and Opportunities

Xiaojun Yuan, Ying-Jun Angela Zhang, Yuanming Shi, Wenjing Yan, Hang Liu

arXiv:2001.00364v3cs.ITeess.SP

TL;DR

RIS deployment requires practical solutions to CSI acquisition, passive information transfer, and low-complexity robust design because RISs reconfigure propagation while remaining nearly passive. This article surveys state-of-the-art approaches and open directions, including edge intelligence and physical-layer security, and reports robustness to channel-estimation errors in its discussed designs.

  • Problem

    RIS-aided networks require CSI for transceiver design, but estimating direct, user-RIS, and RIS-BS links through a nearly passive RIS is difficult.

  • Method

    The article reviews three fundamental physical-layer challenges, summarizes state-of-the-art solutions, and identifies open research directions.

  • Results

    The article reports that data-rate loss is negligible at normalized channel-estimation MSE −10 dB, while a message-passing algorithm reaches −15dB normalized MSE within 100 iterations.

  • Takeaways & Limitations

    RIS research should address CSI acquisition, passive information transfer, and robust low-complexity design while exploring edge intelligence and physical-layer security.

Abstract

from arXiv · show

Reconfigurable intelligent surfaces (RISs) are regarded as a promising emerging hardware technology to improve the spectrum and energy efficiency of wireless networks by artificially reconfiguring the propagation environment of electromagnetic waves. Due to the unique advantages in enhancing wireless channel capacity, RISs have recently become a hot research topic. In this article, we focus on three fundamental physical-layer challenges for the incorporation of RISs into wireless networks, namely, channel state information acquisition, passive information transfer, and low-complexity robust system design. We summarize the state-of-the-art solutions and explore potential research directions. Furthermore, we discuss other promising research directions of RISs, including edge intelligence and physical-layer security.

I. INTRODUCTION

RISs aim to make wireless propagation environments controllable, improving efficiency by adjusting electromagnetic-wave propagation. The article reviews their benefits, practical physical-layer challenges, and additional research directions.

  • RISs use adjustable metasurface reflection amplitude and phase shifts to enhance desired signals and mitigate interference.They can be integrated with technologies including ultra-massive MIMO, terahertz communications, AI-empowered wireless networks, and edge intelligence.
  • RISs can combat unfavorable propagation conditions by manipulating the radio propagation environment in future wireless communications.The article connects this capability with millimeter-wave and terahertz systems using increasingly large antenna arrays.
  • RISs offer deployment flexibility, lower energy consumption, full-duplex and full-band transmission, cost effectiveness, and quadratic power scaling.Their nearly passive operation avoids additional energy consumption, while their lack of converters and power amplifiers reduces hardware requirements.
  • The article focuses on CSI acquisition, passive information transfer, and low-complexity robust system design as fundamental physical-layer challenges.It also highlights RIS-aided edge intelligence and physical-layer security as additional research directions.
  • The article organizes its discussion around channel acquisition, passive information transfer, robust and low-complexity design, other research challenges, and conclusions.These topics are covered in Sections II through VI.

II. CHANNEL STATE INFORMATION ACQUISITION IN RIS-AIDED COMMUNICATION SYSTEMS

CSI acquisition is fundamental because RIS-aided transceiver design depends on it, but RIS systems require estimating additional links through a nearly passive device. This creates a difficult cascaded-channel estimation problem.

  • CSI acquisition is fundamental because RIS-aided transceiver designs depend on knowledge of the channel state information.Examples include joint active and passive beamforming and joint transmit-power allocation and beamforming.
  • RIS-aided systems must estimate the direct BS-user link plus the user-RIS and RIS-BS links.The direct channel can be obtained by turning off all RIS elements, whereas the two additional links are harder to estimate.
  • Because RISs have limited RF transmission, reception, and processing capabilities, conventional pilot-assisted estimation cannot rely on the RIS to estimate or transmit pilot signals.The resulting task is cascaded channel estimation.
  • The RIS-assisted massive MIMO setting connects multiple users to a multi-antenna BS through a RIS.The system configuration is illustrated in Fig. 1.

B. State-of-the-Art Solutions

Early CSI-acquisition solutions use active sensors, channel decomposition, or shared-link structure to make cascaded-channel estimation more tractable. These approaches trade hardware and energy costs against estimation efficiency.

  • Initial CSI-acquisition approaches are grouped into three categories.The supplied passages describe active-channel-sensor, channel-decomposition, and structure-based directions, with the latter developed in the following section.
  • 1) Active-channel-sensor based CSI acquisition:: Active-channel-sensor CSI acquisition inserts sensors with baseband processing into the passive-element array.Sensors alternate between channel-sensing mode and reflection mode.
  • 1) Active-channel-sensor based CSI acquisition:: Active sensors increase RIS hardware cost and energy consumption through additional baseband processing units and sensing operation.These costs can burden a device intended to remain nearly passive.
  • 2) Channel-decomposition based CSI acquisition:: Channel decomposition estimates simpler subchannels by activating one RIS element at a time and combining the resulting estimates.The method decomposes the cascaded channel into rank-1 components associated with individual RIS elements.
  • 2) Channel-decomposition based CSI acquisition:: A three-phase decomposition method activates users sequentially and exploits their shared RIS-BS link to reduce required pilot length.The supplied passage gives the expression as K + N + max, with the remainder truncated.

3) Structure-learning based CSI acquisition:

Structure-learning methods exploit sparsity, low-rankness, and slowly varying RIS-BS channels to reduce CSI-acquisition overhead. The section also frames training length as a function of RIS size and estimation requirements.

  • 3) Structure-learning based CSI acquisition:: RIS cascaded channels exhibit sparsity and low-rankness that can be exploited to reduce CSI-acquisition overhead.Large arrays provide high angular resolution for sparse representations, while limited scattering paths produce low-rank cascaded channels.
  • 3) Structure-learning based CSI acquisition:: Controlling RIS-element on/off states can artificially introduce signal sparsity to assist cascaded-channel estimation.This provides an additional structural property for learning-based acquisition.
  • 3) Structure-learning based CSI acquisition:: Table I reports minimum training lengths of channel-estimation algorithms versus the number of RIS elements.The table’s stated comparison concerns how training requirements vary with RIS size.
  • 3) Structure-learning based CSI acquisition:: The evaluation considers a RIS-aided uplink MIMO system with 200 BS antennas and 100 single-antenna users.Minimum training lengths are determined when channel MSEs fall below −60 dB, and one method uses an active RIS-element rate of 0.15.
  • 3) Structure-learning based CSI acquisition:: The RIS-BS channel varies more slowly than the user-BS and user-RIS channels because the RIS is usually deployed at a fixed location.This quasistatic property is used with channel sparsity in a matrix-calibration-based approach.

C. Research Challenges

RIS channel modelling and acquisition remain open because conventional far-field assumptions may not fit large surfaces, while practical CSI acquisition faces uncertainty and nonlinear joint estimation demands.

  • 1) Channel modelling and channel acquisition:: RIS channel models remain insufficiently understood because surfaces can be comparable in size to their distance from propagation sources.This challenges conventional far-field assumptions used in MIMO modelling.
  • 1) Channel modelling and channel acquisition:: More precise RIS propagation models are needed to characterize RIS-aided MIMO channels accurately.
  • 1) Channel modelling and channel acquisition:: Cascaded-channel estimation has used matrix factorization and completion, but the associated channel model is described as primitive.More realistic models may motivate tensor factorization and structured signal reconstruction.
  • 2) System design under CSI uncertainty:: Most RIS system designs assume perfect CSI, although practical cascaded-channel acquisition is difficult and motivates designs robust to outdated CSI.Acquisition delay is especially relevant because user-RIS and RIS-BS coefficients add substantial information-transfer overhead.
  • 2) System design under CSI uncertainty:: Jointly estimating user-RIS and RIS-BS channels while detecting user data is desirable but forms a highly nonlinear signal estimation problem.

3) Theoretical limits:

RIS research still lacks fundamental performance limits and must address information transfer requirements spanning control, maintenance, CSI assistance, and green IoT. Passive transfer is preferred over dedicated transmitters because it avoids extra cost and power consumption.

  • 3) Theoretical limits:: The required pilot overhead and achievable pilot reduction from channel sparsity or low-rankness remain unclear in general RIS-aided massive MIMO systems.
  • A. Why Passive Information Transfer:: RISs must transfer control, maintenance, and CSI-assistance information in addition to enhancing primary end-to-end communications.
  • A. Why Passive Information Transfer:: Passive information transfer is proposed for green IoT because dedicated RIS transmitters would be costly and consume extra power.

B. State-of-the-Art Solutions

State-of-the-art approaches use an RF signal generator or joint passive beamforming and information transfer, including spatial modulation across RIS elements.

  • B. State-of-the-Art Solutions:: One approach treats the RIS as an access point, using a nearby RF generator whose carrier the RIS modulates while optimizing received SNR.
  • B. State-of-the-Art Solutions:: The PBIT technique jointly transmits RIS information and enhances primary communication quality through passive beamforming.
  • B. State-of-the-Art Solutions:: Spatial modulation can encode RIS information through the on/off states or indices of RIS elements in a completely passive manner.
  • C. Research Challenges:: Passive information transfer, particularly its joint design with passive beamforming, remains an emerging direction with open challenges.

1) RIS design:

RIS design must balance primary communication enhancement against passive information delivery while handling stochastic, bilinear, and capacity-related challenges in PBIT systems.

  • 1) RIS design:: Dividing RIS elements between beamforming and information transfer creates a direct functionality-allocation tradeoff.
  • 1) RIS design:: As the RIS-element activation probability rises from 0.5 to 1, primary rate increases from about 13 bits to 14 bits while per-element information decreases from 1 bit to 0.
  • 1) RIS design:: PBIT passive-beamforming design generally requires stochastic optimization because RIS information randomizes the reflecting coefficients.
  • 1) RIS design:: Joint active and passive beamforming is particularly challenging because RIS information introduces stochastic optimization requirements.
  • 1) RIS design:: PBIT receivers face bilinear signal detection because transmitter and RIS signals are multiplied together.
  • 1) RIS design:: The multiplicative multiple access channel underlying PBIT has poorly understood capacity, complicating joint coding, beamforming, detection, and decoding.

IV. LOW-COMPLEXITY ROBUST SYSTEM DESIGN

RIS systems can achieve strong performance with imperfect CSI, but Bayesian channel estimation and joint RIS-transceiver optimization impose substantial computational costs. Robustness to estimation error supports early stopping before full algorithmic convergence.

  • RIS optimization combines high computational cost from Bayesian cascaded-channel factorization with large-scale non-convex joint phase-shift and transceiver design.These costs motivate low-complexity robust system-design methods.
  • Robustness Against Channel Estimation Errors: A normalized MSE of −10 dB causes negligible achievable-data-rate loss because many RIS elements compensate for inaccurate phase calibration.The resulting error tolerance permits stopping channel estimation once an acceptable MSE is reached.
  • Robustness Against Channel Estimation Errors: −15dB normalized MSE is reached within the first 100 message-passing iterations, whereas −20dB requires more than 400 iterations.The comparison illustrates the computational cost of pursuing additional estimation accuracy.
  • Robustness Against Channel Estimation Errors: The algorithm can safely stop at the 100-th iteration without full convergence because achievable data rate is robust to the remaining estimation error.This stopping rule follows from combining the convergence behavior with the rate robustness observation.
  • Future analysis should characterize performance metrics versus CSI accuracy and analytically track MSE evolution in iterative Bayesian inference.These understandings are identified as necessary for deciding how much computational cost to spend on cascaded channel estimation.

B. Robustness Against Low-Resolution Phase Shifts

Finite RIS phase-shift resolution creates a mixed-integer non-convex optimization problem, while low-resolution quantization can reduce hardware and signaling complexity with limited capacity loss.

  • Finite phase-shift quantization makes RIS optimization an intractable mixed-integer non-convex problem.The hardware has only discrete phase-shift values rather than continuously adjustable shifts.
  • Capacity degradation remains below 1bit/s/Hz when quantization resolution decreases from infinity to 2 bits.This result motivates studying robust system performance under low-resolution hardware.
  • Low-resolution quantization enables low-complexity optimization by restricting the number of candidate solutions.The reduced action space also supports reinforcement-learning approaches such as actor-critic mixed-integer programming solvers.

V. OTHER CHALLENGES OF RIS-AIDED WIRELESS COMMUNICATIONS

RISs offer additional research opportunities in edge intelligence and physical-layer security by controlling propagation conditions, while introducing new optimization and systems challenges. The article surveys these directions alongside its three core physical-layer issues.

  • RIS-Aided Edge Intelligence: RISs can improve edge-caching and edge-computing feasibility by coping with rank-deficient channels and mitigating co-channel interference.They can also boost received signal power for over-the-air computation in edge-learning user scheduling.
  • RIS-Aided Edge Intelligence: Edge intelligence uses edge storage, computation offloading, and radio-access-network learning to reduce traffic, computation latency, and privacy risks.Its deployment remains constrained by network topology and edge-device energy budgets.
  • RIS-Aided Physical-Layer Security: RISs can manipulate propagation to enhance legitimate-user signals and cancel signals reaching eavesdroppers.This creates new physical-layer-security communication models and generally non-convex optimization problems.
  • The article examines CSI acquisition, passive information transfer, and low-complexity robust design, while briefly introducing edge intelligence and physical-layer security.For the three fundamental issues, it summarizes challenges, state-of-the-art solutions, and open research directions.
Loading 2001.00364v3…