The promise of Reconfigurable Intelligent Surfaces (RIS) and Large Intelligent Surfaces (LIS) lies in their ability to turn the wireless environment from a passive, unpredictable medium into an active, software-defined participant. By adjusting the phase shifts of hundreds or thousands of tiny reflecting elements, an RIS can constructively combine signals at a intended receiver or destructively cancel them at an eavesdropper.
Mathematically, optimization models often assume that these phase shifts can be adjusted continuously across a $[0, 2\pi)$ range. However, translating infinite precision into physical hardware requires complex, power-hungry, and expensive radio frequency (RF) components. In practice, hardware constraints force engineers to use discrete, low-resolution phase shifters (e.g., 1-bit, 2-bit, or 3-bit control).
A major research direction has emerged to answer a critical question: Why does an intelligent surface constrained by coarse, discrete phase shifts perform almost identically to an ideal, continuous-phase system? The answer lies in spatial averaging, structural scaling laws, and a distinct phase quantization saturation threshold.
Why Coarse Phase Shifts Work: Spatial Averaging and Scaling Laws
To understand why low-resolution quantization works, it helps to look at how an intelligent surface forms a beam. A single reflecting element with a large phase error will redirect a signal in the wrong direction. However, an RIS achieves its massive beamforming gains not from any single element, but through the coherent combining of reflections across a massive aperture.
When an RIS contains a high density of elements, the phase quantization errors across the surface behave like independent, zero-mean noise spatial variables. When the reflections converge at the receiver, this noise averages out.
Ideal Phase Vector: [ 45°, 90°, 135°, 180° ]
2-Bit Quantized Vector: [ 90°, 90°, 180°, 180° ]
└───┬───┘ └───┬───┘
Phase errors average out over a massive array
The structural resilience of this mechanism is governed by specific scaling laws:
- Asymptotic SNR Scaling: In an analysis of reconfigurable surfaces, Minchae Jung et al. demonstrated that the fundamental quadratic scaling of the signal-to-noise ratio (SNR)—where performance scales with the square of the number of elements ($N^2$)—is preserved even under extreme 1-bit phase control. This implies that hardware simplification does not sacrifice the core asymptotic gains of the surface.
- The 1.6x Aperture Rule: If binary (1-bit) quantization does cause a performance drop, it can be perfectly compensated for by scaling the physical size of the surface. Wei Shi et al. established a precise "1.6x rule" for secure communications, proving that the secrecy outage degradation from 1-bit hardware can be fully mitigated by increasing the total number of reflecting elements by approximately 60%.
- Beam Shape Preservation: Modeling the physical electromagnetics of large surfaces, Marzieh Najafi et al. showed that when element spacing is sufficiently small, low-bit quantization profiles effectively mimic continuous beamforming characteristics. Even under 1-bit quantization, the fundamental shape of the main reflection lobe remains intact.
The Phase Quantization Saturation Threshold
Because spatial averaging mitigates phase errors, increasing the bit-depth of phase shifters yields sharply diminishing returns. The literature identifies a clear hardware saturation threshold, typically occurring between 2 and 4 bits.
System Performance
^
│ ┌─── 3-Bit / 4-Bit Saturation (Near-Optimal)
│ ┌─────┘
│ ┌─────┘
│ ┌┘
│ │
└──┴─────────────────────────────────────────> Phase Resolution (Bits)
1-Bit 2-Bit 3-Bit 4-Bit Continuous
In early foundational work on multi-user MISO communication, Chongwen Huang et al. demonstrated that a 2-bit resolution is sufficient to achieve energy efficiency levels comparable to ideal, continuous-phase systems. Expanding on this for hybrid beamforming architectures, Boya Di et al. showed that moving to 2 or 3 bits captures nearly the entire achievable sum-rate limit, with further resolution increases offering negligible utility.
This saturation effect holds true across different network topologies. For instance, in decentralized cell-free massive MIMO networks, Shaocheng Huang et al. confirmed that 3-bit resolution matches continuous phase control performance. Similarly, Yali Chen et al. mapped the performance of device-to-device (D2D) networks under RIS assistance and found that system throughput completely saturates at 4 bits.
Robustness Across Diverse Communication Paradigms
The validity of low-resolution phase optimization is not limited to standard line-of-sight beamforming; it extends across highly sensitive, non-linear, and high-frequency wireless applications.
Non-Orthogonal Multiple Access (NOMA)
NOMA networks rely on successive interference cancellation (SIC), a process highly sensitive to phase misalignments. Despite this, Xidong Mu et al. and Jiakuo Zuo et al. demonstrated that 3-bit phase shifters provide enough precision to align user channels and manage multi-user interference effectively, matching the performance of infinite-precision hardware.
Terahertz and Near-Field Beamfocusing
As networks transition to millimeter-wave (mmWave) and Terahertz (THz) frequencies, communications shift into the near-field region, requiring precise beamfocusing rather than just directional beam steering.
- Yuanwei Liu et al. showed that 4-bit discrete phase shifters achieve near-field beamfocusing accuracy nearly identical to continuous hardware.
- For THz wavefront control, Hamidreza Taghvaee demonstrated that discretizing the phase space into just four distinct states (2-bit resolution) with a 90-degree separation is sufficient for effective signal redirection.
Integrated Sensing, Imaging, and Security
The robustness of low-resolution quantization also extends to multi-functional wireless systems:
- Sensing: In Integrated Sensing and Communication (ISAC) systems, where the surface must simultaneously track targets and communicate data, Yongqing Xu et al. showed that a 2-bit quantization constraint introduces almost no performance degradation.
- Imaging: Ying He demonstrated that high-resolution WiFi imaging can be successfully achieved using simple 1-bit phase control, provided the element density is sufficiently high.
- Security: For physical-layer cryptographic key generation, Han Han et al. proved that 3-bit discrete surfaces induce artificial randomness with the same statistical efficacy as infinite-resolution alternatives.
Algorithmic Frameworks for Discrete Optimization
While the physical system tolerates low-resolution phase shifts, optimizing these discrete states presents a challenging combinatorial problem. Because discrete phase sets are non-convex, researchers have developed specialized algorithmic strategies to bypass brute-force searches.
One prominent approach is successive refinement, as explored by Jiakuo Zuo et al., which iteratively updates individual reflecting elements one-by-one while keeping others fixed, rapidly converging to near-optimal discrete configurations.
Alternatively, researchers use structural trade-offs to simplify control. Chau Yuen proposed Stacked Intelligent Metasurfaces (SIM), which shift the optimization burden from high-resolution phase control to structural depth. By stacking multiple layers of simple, binary-state (1-bit) elements, a 3D wave-domain processor is formed that achieves advanced radiation pattern synthesis without high-precision components.
Ultimately, this body of research demonstrates that the spatial degrees of freedom provided by large surfaces inherently compensate for coarse hardware controls, making low-resolution phase quantization a viable foundation for next-generation wireless networks.