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