Abstract
This paper investigates the information theoretical limit of a reconfigurable intelligent surface (RIS) aided communication scenario in which the RIS and the transmitter either jointly or independently send information to the receiver. While most previous studies of the RIS focus on its ability to beamform and to boost the received signal-to-noise ratio (SNR), this paper shows that if the information data stream is also available at the RIS and can be modulated through the adjustable phases at the RIS, significant improvement in the degree-of-freedom (DoF) of the overall channel is possible. For an RIS system with $M$ transmit antennas, $K$ receive antennas, and $N$ reflective elements, joint transmission can achieve a DoF of $\min(M+\frac{N}{2}-\frac{1}{2},N,K)$ without a direct path, compared to $\min(M,K)$ for conventional MIMO. The paper establishes a connection between RIS systems and MIMO channels with phase noise, extends results to direct-path cases, and proposes a symbol-level precoding approach for practical modulation.
Introduction
Reconfigurable Intelligent Surfaces (RIS) are typically viewed as "smart mirrors" designed to optimize the wireless environment through passive beamforming. By adjusting the phase shifts of many reflective elements, an RIS can constructively align signals at a receiver, thereby enhancing the Signal-to-Noise Ratio (SNR). However, this conventional use case does not increase the number of independent data streams—the Degree-of-Freedom (DoF)—that the channel can support.
This research explores a paradigm shift: treating the RIS as an active information source. If the RIS has access to the data stream, it can modulate information directly into the environment via its phase shifts. This approach effectively turns the RIS into a set of virtual antennas, potentially offering significant capacity gains.
System Model and Problem Formulation
The study considers a communication system where a transmitter with $M$ antennas sends signals to a receiver with $K$ antennas, aided by an RIS with $N$ elements. The channel is characterized by its multiplicative nature: the received signal is a product of the transmitter's signal and the RIS's phase shifts.
As shown in the system model, the signal travels through a direct path $\mathbf{F}$ and a reflected path involving matrices $\mathbf{G}$ (transmitter to RIS) and $\mathbf{H}$ (RIS to receiver). The RIS configuration is represented by a diagonal matrix $\mathbf{\Theta}$ of phase shifts. The primary challenge in analyzing this setup lies in the non-linear interaction between the transmit signal $\mathbf{X}$ and the RIS phases $\mathbf{\Theta}$.
Information-Theoretic Analysis
To determine the DoF, the authors utilize tools from fractal geometry, specifically information dimension and point-wise dimension under projection. A key insight is the "1/2 DoF penalty" observed in systems without a direct path. Due to global phase ambiguity—the receiver's inability to distinguish between the transmitter's phase and the RIS's phase—0.5 DoF is lost.
Interestingly, each RIS element contributes approximately 0.5 DoF because it only controls phase, not amplitude. Thus, an $N$-element RIS acts like $N/2$ additional virtual transmit antennas. The presence of even a weak direct path can resolve the phase ambiguity, recovering the lost 0.5 DoF.
Key Results: Degree-of-Freedom Regions
The paper establishes explicit DoF formulas for various scenarios:
- Without Direct Path: $DoF = \min(M + \frac{N}{2} - \frac{1}{2}, N, K)$
- With Direct Path (rank $r > 0$): $DoF = \min(M + \frac{N}{2}, N + r, K)$
The DoF regions illustrate the trade-off between the information carried by the transmitter ($DoF_{\mathbf{X}}$) and the RIS ($DoF_{\mathbf{\Theta}}$). In different regimes (a, b, and c), the total system DoF is limited by different physical constraints, such as the number of receive antennas or the total number of reflective elements.
Practical Implementation: Symbol-Level Precoding
To validate these theoretical limits, the authors propose a Symbol-Level Precoding (SLP) strategy. They formulate a non-convex optimization problem to jointly design the transmit signal and RIS phases so the received signal matches a target constellation point. This is solved using an Augmented Lagrangian approach combined with a Riemannian conjugate gradient method to respect the unit-modulus constraint of the RIS elements.
Numerical simulations confirm the theoretical findings, showing a sharp "phase transition" in signal synthesis success as the number of RIS elements reaches the predicted DoF requirements.
Conclusion and Future Directions
This work demonstrates that RIS-based phase modulation can significantly boost the multiplexing capacity of wireless channels. While promising, practical deployment faces challenges including the need for perfect Channel State Information (CSI), hardware limitations like discrete phase levels, and the computational complexity of symbol-level precoding. Future research will focus on low-complexity algorithms and robust designs under imperfect CSI.
Summary updated