Source-linked AI summary
Channel Estimation for RIS Assisted Wireless Communications: Part I -- Fundamentals, Solutions, and Future Opportunities
Xiuhong Wei, Decai Shen, Linglong Dai
TL;DR
RIS-assisted communication can improve coverage and capacity, but accurate CSI is difficult to obtain because many RIS elements are passive and cannot process pilot signals. This paper’s first part surveys channel-estimation fundamentals, three overhead-reduction approaches, and future opportunities, while identifying high pilot overhead as an unresolved challenge addressed in the second part.
Problem
Accurate CSI is needed for reliable RIS beamforming, but passive RIS elements and large channel dimensions make channel estimation challenging and increase pilot overhead.
Method
The paper reviews channel-estimation fundamentals and compares solutions exploiting two-timescale channels, multi-user correlation, and channel sparsity.
Results
The first part provides an overview of channel-estimation fundamentals, overhead-reduced solutions, challenges, and future research opportunities.
Takeaways & Limitations
The survey identifies high pilot overhead as a key challenge and points to further channel-characteristic exploitation as a route for reducing it.
Abstract
from arXiv · showhide
The reconfigurable intelligent surface (RIS) with low hardware cost and energy consumption has been recognized as a potential technique for future 6G communications to enhance coverage and capacity. To achieve this goal, accurate channel state information (CSI) in RIS assisted wireless communication system is essential for the joint beamforming at the base station (BS) and the RIS. However, channel estimation is challenging, since a large number of passive RIS elements cannot transmit, receive, or process signals. In the first part of this invited paper, we provide an overview of the fundamentals, solutions, and future opportunities of channel estimation in the RIS assisted wireless communication system. It is noted that a new channel estimation scheme with low pilot overhead will be provided in the second part of this paper.
I. INTRODUCTION
RIS can improve wireless coverage and capacity with low hardware cost and energy consumption, but accurate CSI is essential for reliable beamforming. Channel estimation is difficult because RIS elements are passive and numerous, creating high pilot overhead; this first part surveys fundamentals, reduced-overhead solutions, and future opportunities.
- RIS can enhance wireless coverage and capacity with low hardware cost and energy consumption.
- Accurate CSI is essential for reliable beamforming in RIS-assisted wireless communication systems.
- Passive RIS elements cannot transmit, receive, or process pilot signals, preventing conventional channel-estimation schemes from being directly applied.
- The large number of RIS elements increases channel dimension and sharply raises pilot overhead, making channel estimation a key challenge.
- This first part surveys channel-estimation fundamentals, overhead-reduced solutions, and future opportunities.
II. FUNDAMENTALS OF CHANNEL ESTIMATION IN THE RIS ASSISTED SYSTEM
The system model considers uplink communication from multiple single-antenna users to an M-antenna BS assisted by an N-element RIS. It defines the channels, RIS reflection vector, received signal, noise, and TDD reciprocity used for channel estimation.
- A. System Model: The uplink system contains one M-antenna BS, one N-element RIS, and K single-antenna users.
- A. System Model: The direct user-to-BS channel is h_d,k, the RIS-to-BS channel is G, and the user-to-RIS channel is h_r,k.
- A. System Model: The received BS signal combines each user’s direct and RIS-reflected links, transmitted symbol, and additive noise.
- A. System Model: The RIS reflection vector θ contains one coefficient per element, with each coefficient controlling amplitude and phase.
- A. System Model: Under TDD reciprocity, the downlink channel can be obtained from the estimated uplink channel.
B. Channel Estimation Problem
The direct channel can use conventional estimation, whereas passive RIS hardware makes the RIS-related channels difficult to estimate. Because beamforming optimizes the effective reflected link, schemes commonly estimate the cascaded channel instead of its individual components.
- Conventional schemes can estimate the direct channel, but passive RIS elements make G and h_r,k difficult to estimate.
- The cascaded channel H_k = Gdiag(h_r,k) represents the user-to-BS channel through the RIS.
- Beamforming algorithms optimize the effective reflecting link, whose power equals ∥H_kθ∥^2.
- Consequently, most existing estimation schemes estimate H_k directly rather than separately estimating G and h_r,k.
- Orthogonal pilot transmission makes uplink channel estimation for different users independent.
C. Basic Channel Estimation Schemes
Basic schemes first estimate the direct channel conventionally, then estimate the RIS-related cascaded channel by sequential RIS activation or DFT-designed reflections. These approaches can recover the cascaded channel, but their pilot overhead is large for high-dimensional RIS systems.
- The direct channel can be estimated with classical methods such as least squares when RIS elements are perfectly turned off.The paper notes that “turn off” is an approximation unless a special RIS setting absorbs the incident wave.
- The ON/OFF protocol estimates the cascaded channel one RIS-element column at a time across N stages.Each stage activates one element, removes the previously estimated direct-channel contribution, and applies least squares to estimate the corresponding column.
- The ON/OFF protocol may reduce estimation accuracy because only one RIS element reflects the pilot signal in each stage.
- The DFT protocol keeps all RIS elements on and uses one DFT-matrix reflecting vector per stage before estimating the cascaded channel with least squares.
- 64 × 256 unknown channel coefficients versus 64 × 1 in a conventional system illustrates why the required pilot overhead is huge and can reduce effective capacity gains.
III. OVERHEAD-REDUCED CHANNEL ESTIMATION SOLUTIONS
This section presents three pilot-overhead reduction strategies for RIS channel estimation, exploiting two-timescale channel behavior, multiuser correlation, and channel sparsity. It details two-timescale estimation, including dual-link estimation of the BS–RIS channel and conventional estimation of user channels, while noting remaining challenges.
- Three channel-estimation solution types reduce pilot overhead by exploiting two-timescale channel behavior, multiuser correlation, and channel sparsity.
- Two-timescale channel estimation: Two-timescale estimation separately targets the slowly varying BS–RIS channel G and the faster-varying user–RIS and direct user–BS channels.The BS and RIS are typically fixed, whereas user mobility makes the user-related channels vary on a smaller timescale.
- Two-timescale channel estimation: A dual-link pilot strategy estimates the high-dimensional BS–RIS channel G over a large timescale by transmitting pilots from the BS to the RIS and reflecting them back.The strategy uses N + 1 sub-frames, addressing the challenge that RIS elements are passive and lack signal-processing capability.
- Alternative schemes and challenges: Alternative BS–RIS channel methods use two users near the RIS, while the two-timescale approach still faces full-duplex requirements and user-scheduling complexity.The two-user method estimates cascaded channels before calculating entries of G; the cited passage truncates the full list of its limitations.
- Two-timescale channel estimation: After G is acquired, conventional uplink pilots with known RIS configuration Φ enable BS-side estimation of the direct channel hd and user–RIS channel hr.The received pilots traverse both hd and the effective reflecting channel GΦhr, and LS is cited as an example estimator.
B. Multi-User Correlation Based Channel Estimation
Multi-user correlation reduces RIS cascaded-channel estimation overhead by reusing the user-independent RIS-to-BS channel, but the approach assumes noiseless BS reception and degrades at low SNR.
- Overhead reduction: The method can significantly decrease pilot overhead because far fewer channel coefficients must be estimated.Its stated assumption is that the same RIS-to-BS channel is available across users.
- Multi-user correlation: Cascaded channels for different users are correlated because all users share the same RIS-to-BS channel G.For each RIS element, the cascaded channel can be expressed using a user-dependent scalar and a shared channel vector.
- Estimation procedure: The scheme first estimates one typical user’s cascaded channel, then estimates only one unknown scalar per RIS element for each additional user.This replaces estimation of an M-dimensional column vector with estimation of a single coefficient for each RIS element.
- Overhead reduction: Only N scalars are required to obtain an additional user’s M × N cascaded channel.The scalar reduction follows from the shared RIS-to-BS channel structure.
- Limitation: The proposed scheme assumes no receiving noise at the BS, so estimation accuracy degrades in typical low-SNR channel-estimation conditions.The limitation arises because the noiseless-reception assumption does not match low-SNR operation.
C. Sparsity Based Channel Estimation
Angular-domain sparsity turns RIS cascaded-channel estimation into a sparse signal-recovery problem, enabling compressive-sensing methods and further overhead reductions through sub-surfaces and wideband extensions.
- Angular-domain sparsity: RIS cascaded channels are typically sparse in the angular domain because limited scattering produces few propagation paths.The angular cascaded channel’s nonzero count depends on the product of path counts on the RIS–BS and user–RIS links.
- Sparse recovery: Angular cascaded-channel estimation can be formulated as sparse signal recovery and addressed with compressive-sensing algorithms such as OMP.These methods reduce pilot overhead, although conventional CS methods have unsatisfactory accuracy at low SNR.
- Sub-surface design: Dividing the RIS into sub-surfaces with shared channel coefficients significantly decreases the number of coefficients to estimate.Combining sub-surface division with existing overhead-reduced schemes can reduce pilot overhead further.
- Wideband extension: The narrowband estimation ideas can extend to wideband OFDM by estimating each sub-carrier separately while using the same RIS reflecting vector.The passage also points to common angular-domain sparsity across sub-carriers as a basis for further processing.
IV. CHALLENGES AND FUTURE OPPORTUNITIES FOR CHANNEL ESTIMATION
The paper identifies key challenges in RIS-assisted channel estimation and uses them to motivate corresponding future research opportunities.
- Challenges and future opportunities: Future research opportunities are discussed by first identifying key channel-estimation challenges in RIS-assisted wireless communication systems.The section is explicitly organized around challenges and their corresponding opportunities.
A. Ultra-Wideband Channel Estimation
Ultra-wideband RIS-assisted communication introduces beam squint, making channel estimation more difficult because one physical angle maps to multiple spatial angles.
- Ultra-wideband challenge: Beam squint in ultra-wideband RIS-assisted communication transforms a single physical angle into multiple spatial angles, creating a serious channel-estimation challenge.The paper identifies beam-squint pattern matching as a related wideband estimation idea from conventional wireless systems.
B. Spatial Non-Stationarity
As RIS arrays grow much larger, RIS-related channels exhibit spatial non-stationarity, in which incident-wave direction and power vary across the array.
- RIS arrays may be hundreds of times larger than those in most prior scenarios, creating large RIS arrays.
- Larger arrays introduce spatial non-stationarity in RIS-related channels.The passage characterizes this as a new channel property associated with significant array-size increases.
- Spatial non-stationarity means the incident electromagnetic wave's direction and power vary across the RIS array.
C. RIS Assisted Cell-Free Network
RIS-assisted cell-free networks increase the number of channels requiring estimation as the number of RISs grows. Existing overhead-reduced methods still require substantial pilot overhead, motivating further exploitation of RIS channel characteristics and a subsequent paper part addressing this challenge.
- Adding more RISs to a cell-free network increases the number of channels that must be estimated.This setting is considered alongside the goal of improving network capacity with low power consumption.
- Multi-user correlation is identified as one possible way to address the growing channel-estimation burden in RIS-assisted cell-free networks.
- Existing overhead-reduced channel-estimation methods still require high pilot overhead because RIS elements are passive.
- The paper introduces three typical types of overhead-reduced channel-estimation solutions and identifies high pilot overhead as a key challenge.
- A feasible solution for reducing high pilot overhead is reserved for the second part of the invited paper.