Abstract
This paper introduces a hardware and signal model for dual-sided simultaneously transmitting and reflecting reconfigurable intelligent surfaces (STAR-RISs). Unlike conventional single-sided models, this work considers signals incident on both sides of the surface simultaneously. For elements characterized by scalar surface impedance, the authors prove that transmission and reflection coefficients are identical for both sides. This symmetrical model is then applied to a two-user uplink communication system, comparing non-orthogonal multiple access (NOMA) and orthogonal multiple access (OMA). Analytical derivations of outage probabilities in high-SNR regimes reveal that NOMA outperforms OMA and that the outage probability error floor can be mitigated by tuning the transmission-to-reflection amplitude ratio.
Introduction: Beyond Single-Sided Reflection
Reconfigurable Intelligent Surfaces (RISs) have emerged as a transformative technology for controlling wireless propagation environments. However, traditional RISs are limited to a 180° half-space because they only reflect signals. The advent of STAR-RISs has expanded this to full 360° coverage by allowing signals to both transmit through and reflect off the surface.
While most existing research focuses on single-sided incidence (e.g., downlink from a base station), real-world scenarios like uplink communications involve users on both sides of the surface. This paper addresses the critical gap in understanding how STAR-RIS elements behave when signals strike both sides simultaneously—a condition termed "dual-sided incidence."
Hardware and Signal Modeling
The authors develop a comprehensive model for an electrically thin STAR-RIS element using electromagnetic (EM) principles. By analyzing surface electric ($J_s$) and magnetic ($K_s$) currents, they establish the relationship between incident and radiated fields.
A pivotal finding is the Symmetry Principle: if an element's surface impedance is scalar, its transmission and reflection (T&R) coefficients are reciprocal and identical for both sides ($T_{A \to B} = T_{B \to A}$ and $R_A = R_B$). This symmetry simplifies hardware design, as elements do not require orientation-specific optimization for dual-sided operation.
System Model and Performance Analysis
To evaluate the practical implications of this model, the study investigates a two-user uplink scenario. In this setup, one user is located on the reflection side (Space A) and another on the transmission side (Space B), both communicating with a common base station (BS) through the STAR-RIS.
The authors compare two multiple access schemes:
- NOMA: Users share the same time-frequency resources, and the BS employs Successive Interference Cancellation (SIC).
- OMA: Users are served in orthogonal time or frequency slots.
Due to the complexity of double-Rician cascaded channels, the Central Limit Theorem (CLT) is used to approximate the aggregate channel as a normal distribution. This allows for the derivation of closed-form expressions for outage probabilities in the high-SNR regime.
Key Insights and Results
- NOMA Synergy: Numerical simulations confirm that STAR-RIS-aided NOMA consistently achieves lower outage probabilities than OMA. The STAR-RIS provides an additional degree of freedom—the power ratio between transmitted and reflected signals—which can be tuned to optimize SIC performance.
- The Error Floor Phenomenon: In dual-sided uplink scenarios, the outage probability eventually hits an "error floor" at high SNR. This is not due to thermal noise but stems from the inherent inter-user interference at the surface.
- Tunable Mitigation: The study demonstrates that this error floor can be lowered by adjusting the amplitudes of the transmission and reflection coefficients ($\beta^T$ and $\beta^R$). For instance, reducing the transmission coefficient can improve the performance of the reflection-side user without necessarily degrading the transmission-side user's connectivity.
Conclusion and Future Directions
This work establishes a foundational framework for dual-sided STAR-RIS applications. While the scalar impedance assumption reveals a useful symmetry, future research should explore tensor-based impedances which might introduce non-reciprocal behavior. Additionally, scaling the system to support more users and accounting for hardware impairments like phase-shift errors will be essential for practical deployments.
Summary updated