Novel 1-bit Hybrid Reconfigurable Intelligent Surface

Sajedeh Keshmiri, Suren Jayasuriya, Mohammadreza F. ImaniarXiv 2025 · arXiv

TL;DR

Researchers have developed a low-cost, 1-bit Hybrid Reconfigurable Intelligent Surface (HRIS) that can simultaneously steer radio signals and "sense" their direction of arrival. By using a simple parallel plate waveguide and clever phase randomization, the design overcomes the common 1-bit problem of unwanted interference patterns (quantization lobes) while maintaining high sensing accuracy.

What Problem Does This Paper Address?

Reconfigurable Intelligent Surfaces (RIS) are a cornerstone of future 6G networks, acting as "smart mirrors" that redirect signals to eliminate dead zones. However, most RIS are "blind"—they require external feedback from a base station to know where a user is located before they can steer a beam toward them.

Existing "hybrid" designs that add sensing capabilities to the RIS are often prohibitively complex, requiring expensive RF chains for every element or using structures (like substrate-integrated waveguides) that result in weak signals and difficult manufacturing. Furthermore, simple "1-bit" RIS designs (which only have two states: 0 or 180 degrees) usually suffer from "quantization lobes"—unwanted secondary beams that waste energy and cause interference. This paper aims to create a sensing-capable RIS that is simple, low-cost, and efficient.

Methods

The authors propose a hardware-software co-design:

  • Hardware Design: The surface consists of 19 square patch elements loaded with PIN diodes (providing 1-bit control). To enable sensing, they etched small rectangular slots into the ground plane. These slots couple a tiny portion of the incident signal into a Parallel Plate Waveguide (PPWG) located beneath the patches. Only two coaxial connectors at the ends of the PPWG are needed to collect the multiplexed signal from the entire array.
  • Phase Randomization: To kill the unwanted quantization lobes, the researchers varied the lengths of the coupling slots (2mm vs. 6mm) across the array. This introduces "pre-coded" random phase delays that break up the periodic patterns responsible for interference.
  • Signal Processing: The team used two distinct approaches to determine the Angle of Arrival (AoA):
  • Compressive Sensing: Utilizing a Conjugate Gradient Squared (CGS) solver to reconstruct the signal direction from the multiplexed data.
  • Machine Learning: A 3-layer Multi-Layer Perceptron (MLP) was trained to map the voltages at the coaxial ports directly to the incident angle.

Datasets

The study relies on high-fidelity synthetic data:

  • Full-wave Simulations: Ansys HFSS was used to model the electromagnetic behavior of the unit cells and the 19-element array at 5.6 GHz.
  • MLP Training Data: The authors generated a dataset of 25 reference angles (from -60° to 60°). This was expanded 300-fold by adding Gaussian noise to simulate various Signal-to-Noise Ratios (SNR), creating a robust training set for the neural network.

Key Insights and Ideas

  • PPWG over SIW: Moving from Substrate-Integrated Waveguides (SIWs) to a Parallel Plate Waveguide (PPWG) is a major simplification. It provides a stronger sensing signal and makes the device much easier to manufacture on standard PCB substrates.
  • Randomization as a Feature: Usually, 1-bit quantization is a limitation. Here, the authors turn it into a strength by using randomized slot sizes to implement "phase randomization." This suppresses grating lobes without needing expensive multi-bit phase shifters or varactor diodes.
  • Sensing with Minimal Hardware: The design proves that you don't need an RF chain for every antenna element. By multiplexing signals into a single waveguide and using "spatial masks" (random on/off patterns of the diodes), they can extract high-resolution directional data from just two measurement points.

Key Results

  • Sensing Accuracy: The system achieved an Angle of Arrival (AoA) accuracy of ±3°.
  • Robustness: The MLP classifier proved highly resilient to noise, maintaining accurate classification even at low SNR levels where traditional mathematical solvers (CGS) began to fail.
  • Efficiency: Adding the sensing slots had a negligible impact on the primary job of the RIS; the reflection coefficient was reduced by less than 0.1 dB, meaning almost all the energy is still available for beamsteering.
  • Beamsteering: Simulations in HFSS confirmed that the randomized slot distribution successfully suppressed quantization lobes, producing clean, directed beams at various target angles (e.g., -10°, -30°, -55°).

Limitations and Future Work

  • 1D vs. 2D: The current study focuses on a 1D array (steering in the yz-plane). Real-world applications will require expanding this to a 2D surface for full hemispherical coverage.
  • Optimization: While the authors showed that random slot distribution works, they noted that finding the optimum distribution to further minimize side lobes is a task for future research.
  • Prototyping: The results are currently simulation-based. While the physics are well-grounded in previous experimental work by the same group, a physical prototype of this specific 1-bit PPWG design is the logical next step.

Summary updated

Abstract

This paper proposes a novel 1-bit hybrid reconfigurable intelligent surface (HRIS) designed to sense the incident signal's angle of arrival (AoA) and redirect it toward desired directions. This device consists of independently tunable resonant patch elements loaded with PIN diodes. To introduce sensing capabilities, a portion of the signal incident on each element is coupled to a parallel plate waveguide (PPWG) through small rectangular slots. Two coaxial connectors are then used to collect the signal coupled to the PPWG. We use compressive sensing techniques and a multi-layer perceptron to analyze this signal and detect AoA. Further, pre-coded phase randomization is implemented by varying slot sizes to suppress undesired quantization lobes. The proposed HRIS is simple, low-cost, and can pave the way for intelligent wireless communication, power transfer, and sensing without needing feedback loops.
BibTeX
@misc{keshmiri2025novel,
  author = {Sajedeh Keshmiri and Suren Jayasuriya and Mohammadreza F. Imani},
  title = {{Novel 1-bit Hybrid Reconfigurable Intelligent Surface}},
  year = {2025},
  eprint = {2507.04675},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2507.04675},
  howpublished = {arXiv},
}
Figure