Low Resolution Phase Quantization for Intelligent Surfaces

113 concepts95 papers4 proposals

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.

History & citationVersion 2 · 2026-10-04

Current · Permanent link

  1. Version 2 (latest) · 2026-10-04 · 95 papers
  2. Version 1 · 2026-10-04 · 163 papers
Cite this direction

Latest version (a living page):

@misc{lacuna_discrete_phase_shift_optimization_572a4d67,
  title = {discrete phase shift optimization},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/discrete-phase-shift-optimization/rdr_572a4d675dfe4158827dd53b1e4c4628},
  note = {Latest version of a living research direction; cite a version for a fixed text. Accessed 2026-11-20},
}

This exact version (version 2, frozen text):

@misc{lacuna_discrete_phase_shift_optimization_572a4d67_v2,
  title = {discrete phase shift optimization},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/discrete-phase-shift-optimization/rdr_572a4d675dfe4158827dd53b1e4c4628/v/2},
  note = {Version 2, frozen 2026-10-04. Accessed 2026-11-20},
}
About

These directions overlap with this one through shared papers.

See something wrong? Report a mistake