Beamforming Under Discrete Phase Shift Constraints

568 concepts455 papers5 proposals

Reconfigurable Intelligent Surfaces (RIS) and Holographic MIMO systems promise to transform wireless environments by dynamically steering electromagnetic waves. In an idealized mathematical model, each unit cell on a surface can shift the phase of an incoming wave by any arbitrary angle $\theta \in [0, 2\pi)$ while maintaining perfect amplitude reflection. This continuous phase control allows for smooth, gradient-based optimization of beamforming vectors to maximize data rates or energy efficiency.

However, as pointed out by Huang et al. and Gong et al., physical hardware cannot deliver infinite phase resolution. Building an RIS with continuous phase control requires complex, expensive, and power-hungry analog components like high-precision varactor diodes. To keep manufacturing costs low and deployments scalable, practical metasurfaces rely on low-resolution digital phase shifters (e.g., 1-bit or 2-bit control), restricting each element to a finite set of discrete phase states.

The Combinatorial Friction of Quantization

Why does this hardware constraint break traditional beamforming algorithms? When phase shifts are continuous, the optimization space is a smooth, continuous manifold. Optimization frameworks can calculate gradients and smoothly navigate toward an optimal configuration.

Introducing discrete phase constraints completely alters the mathematical landscape. As Fischer et al. and Zhu et al. explain, limiting phase shifters to a few bits introduces integer constraints, transforming a continuous optimization problem into a highly non-convex, NP-hard mixed-integer nonlinear program. If an RIS has $N$ elements and each element has $B$ discrete phase states, there are $B^N$ possible surface configurations. For a surface with hundreds of elements, an exhaustive combinatorial search is computationally impossible.

Furthermore, simply rounding a continuous solution to the nearest discrete phase state—a naive quantization approach—often fails. In multi-user scenarios, precise phase alignment is required to null out interference. Coarse rounding destroys this precision, leading to severe signal degradation and multi-user interference, as noted by Yu et al..

Modeling Hardware Impairments and Phase Noise

To design algorithms that work in the real world, researchers must accurately model how discrete quantization and hardware imperfections alter signal propagation.

When a continuous phase target is mapped to a discrete state, the difference manifests as phase quantization error. Additionally, real-world components suffer from manufacturing tolerances and thermal fluctuations, introducing stochastic phase noise. In Spectral and Energy Efficiency of IRS-Assisted MISO Communication with Hardware Impairments, the authors model this phase noise stochastically as a uniformly distributed random variable. They demonstrate a critical physical insight: unlike ideal systems where adding more reflecting elements infinitely boosts capacity, real-world hardware impairments create a hard ceiling on spectral efficiency.

Similarly, Qian et al. analyze the Signal-to-Noise Ratio (SNR) distribution under channel fading and phase noise. They provide a foundational scaling law: as the number of reflecting elements $N$ grows, the collective array gain can counteract channel fading, but the presence of phase noise dictates the ultimate reliability and variance of the practical channel.

How Algorithms Navigate Discrete Spaces

To overcome the combinatorial explosion of discrete phase states, researchers employ two primary algorithmic paradigms: continuous relaxation with smart penalties, and alternating minimization.

Continuous Relaxation and Penalty Methods

Instead of solving the discrete problem directly, a common strategy is to temporarily relax the discrete constraints, allowing the phase shifts to be continuous during the initial optimization steps. However, to prevent the final solution from requiring unachievable continuous values, researchers introduce a mathematical "penalty" to the objective function.

As demonstrated by Chen et al., a penalty-based iterative strategy applies a mathematical force that gradually pushes the continuous variables toward the allowed discrete hardware states as the algorithm converges. This approach allows the optimization to leverage smooth gradient steps early on, while ensuring the final output perfectly matches the physical hardware constraints without requiring an exhaustive combinatorial search. Ma et al. utilize a similar philosophy in their framework, bridging continuous gradient-based updates and discrete hardware constraints to achieve fast convergence in large-scale systems.

Alternating Minimization

Another powerful mechanism is splitting the joint optimization problem into manageable sub-problems. In an RIS-assisted system, the base station's active beamforming and the surface's passive phase shifts are deeply coupled. Yu et al. pioneered the use of Alternating Minimization (AltMin) for hybrid architectures, treating the hardware constraints as a matrix factorization problem. By alternating between optimizing the active digital beamformer (which is continuous) and the passive discrete phase shifters (one element or one subset at a time), the algorithm breaks down the non-convex problem into steps that can be solved efficiently or even in closed form.

The Scaling Law Solution

Why do these hardware-constrained systems ultimately work so well, even with ultra-coarse 1-bit or 2-bit phase shifters? The answer lies in spatial scaling.

While an individual low-resolution element introduces significant phase error, an intelligent surface compensates for quality with quantity. As Sha and Huang et al. explore in the context of continuous holographic apertures and dense metasurfaces, packing a massive number of sub-wavelength elements creates an aggregate beamforming effect. The random quantization errors across hundreds of elements tend to average out in the far field, allowing the surface to synthesize highly directional beams. By explicitly embedding discrete hardware constraints into the optimization loop rather than treating them as an afterthought, modern wireless frameworks successfully unlock the massive energy and spectral efficiency gains of programmable environments using cheap, imperfect hardware.

History & citationVersion 2 · 2026-10-04

Current · Permanent link

  1. Version 2 (latest) · 2026-10-04 · 465 papers
  2. Version 1 · 2026-10-04 · 312 papers
Cite this direction

Latest version (a living page):

@misc{lacuna_discrete_phase_shift_constraints_787ab07d,
  title = {discrete phase shift constraints},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/discrete-phase-shift-constraints/rdr_787ab07ddc324b39804425a008d93634},
  note = {Latest version of a living research direction; cite a version for a fixed text. Accessed 2026-11-22},
}

This exact version (version 2, frozen text):

@misc{lacuna_discrete_phase_shift_constraints_787ab07d_v2,
  title = {discrete phase shift constraints},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/discrete-phase-shift-constraints/rdr_787ab07ddc324b39804425a008d93634/v/2},
  note = {Version 2, frozen 2026-10-04. Accessed 2026-11-22},
}
About

These directions overlap with this one through shared papers.

See something wrong? Report a mistake