⚡ electrical engineering

Multi-User Symbol Detection with XL Reception: Dynamic Metasurface Antennas with Low Resolution ADCs

This paper proposes an efficient joint design for hybrid analog-digital combining in uplink multi-user XL DMA systems with low-resolution ADCs by formulating and solving a non-convex MSE minimization problem via Bussgang decomposition, demonstrating that such architectures achieve accurate symbol detection with favorable hardware complexity trade-offs.

Original authors: Rahul K. Pal, Soumya P. Dash, Barathram Ramkumar, George C. Alexandropoulos

Published 2026-04-09

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Rahul K. Pal, Soumya P. Dash, Barathram Ramkumar, George C. Alexandropoulos

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

The Big Picture: The "Super-Listening" Wall

Imagine you are at a massive, noisy concert. You want to hear the specific lyrics of four different singers (the users) who are all singing at once. In a traditional setup, you would need a separate, expensive, high-tech microphone for every single singer, plus a giant computer to process all those sounds. This is like current "Massive MIMO" technology: it works great, but it's heavy, expensive, and eats up a lot of electricity.

This paper proposes a new way to listen using Dynamic Metasurface Antennas (DMAs). Think of a DMA not as a bunch of individual microphones, but as a giant, smart, living wall made of thousands of tiny, tunable tiles.

The Problem: Too Much Noise, Too Little Power

The researchers are trying to solve a specific problem: How do we listen to many people at once using this "smart wall" without needing a supercomputer and perfect, expensive equipment?

The Hardware Constraint: The "smart wall" is divided into strips (like rows of tiles). Each strip connects to only one processing chain (one "ear"). This saves money and power.
The "Low-Resolution" Problem: To save even more power, the researchers want to use cheap, low-quality digital converters (ADCs) to turn the sound into data. Imagine trying to describe a complex painting using only "Black," "White," and "Gray" (1-bit) instead of millions of colors. This creates "quantization noise"—a fuzzy distortion that makes it hard to hear the singers clearly.

The Solution: The "Smart Filter" and the "Mathematical Magic Trick"

The paper's goal is to design a system that can still hear the singers clearly, even with these cheap, fuzzy converters. They do this by optimizing two things simultaneously:

The Analog Filter (The Wall's Shape): Before the sound hits the computer, the "smart wall" tiles physically twist and turn the sound waves. The researchers figure out exactly how to bend these waves so that the singers' voices line up perfectly and the background noise gets canceled out.
The Digital Filter (The Computer's Brain): Once the sound is digitized (even with low quality), a computer algorithm cleans up the rest of the mess.

The Challenge: Designing the wall and the computer brain at the same time is incredibly hard. It's like trying to tune a guitar while simultaneously writing the sheet music for a symphony. The math is "non-convex," which is a fancy way of saying the solution landscape is full of hills and valleys, making it easy to get stuck in a "good enough" answer rather than the "best" one.

The Magic Trick (Bussgang Decomposition):
To solve this, the authors use a mathematical tool called Bussgang decomposition.

The Analogy: Imagine you are trying to guess the shape of a hidden object by looking at its shadow through a foggy window. The shadow is distorted (quantized). The Bussgang trick is like a special pair of glasses that mathematically "de-fogs" the shadow, allowing you to treat the distorted signal as if it were a clean, linear signal with a little bit of extra static noise added to it. This turns a nightmare of complex math into a manageable puzzle.

The Results: What Did They Find?

The researchers ran simulations to see how well this system works. Here are their main discoveries, translated:

Cheap vs. Expensive: Even with very low-resolution "ears" (1-bit or 2-bit converters), the system works surprisingly well. However, as you upgrade to 3-bit converters, the performance jumps to near-perfect levels. It's like upgrading from a grainy black-and-white TV to a high-definition one; the jump from 2-bit to 3-bit makes a huge difference.
More "Ears" is Better: The number of strips (microstrips) on the wall matters a lot. If you have more strips than there are singers, the system can separate the voices perfectly. It's like having enough distinct microphones to isolate each singer.
Bigger Tiles Don't Help (Once You're Big Enough): Increasing the number of tiny tiles on each strip didn't actually improve the sound quality once the system was already large. It's like adding more pixels to a photo that is already blurry because of the lens; the lens (the number of strips) is the bottleneck, not the pixels (the tiles).

The Takeaway

This paper shows that we can build massive, high-capacity communication systems (like 6G networks) that are cheap and energy-efficient. By using a "smart wall" architecture and clever math to handle low-quality digital converters, we can listen to many users at once without needing a supercomputer in every cell tower.

In short: They figured out how to build a super-listening wall that uses cheap parts and still hears everything clearly, proving that you don't need expensive hardware to get great performance if you have the right math.

1. Problem Statement

The paper addresses the challenge of uplink multi-user symbol detection in eXtremely Large (XL) MIMO systems utilizing Dynamic Metasurface Antennas (DMAs). While DMAs offer a cost- and energy-efficient alternative to conventional massive MIMO by reducing the number of Radio-Frequency (RF) chains, practical deployment faces two critical constraints:

Hardware Structural Constraints: DMAs consist of multiple microstrips (waveguides) where elements are coupled within a strip but independent across strips. This imposes non-convex structural constraints on the analog combining weights.
Quantization Effects: To further reduce power and cost, the RF chains are equipped with low-resolution Analog-to-Digital Converters (ADCs) (e.g., 1-bit to 3-bit). The non-linear distortion introduced by coarse quantization complicates signal detection.

The core problem is to design a hybrid Analog/Digital (A/D) combiner that minimizes the Mean Squared Error (MSE) between the transmitted symbols and the detected signals, despite the non-convex hardware constraints and quantization noise.

2. Methodology

The authors propose a framework based on Alternate Optimization (AO) to jointly design the analog DMA weights and the digital linear filter. The methodology involves the following key steps:

System Modeling:
- The DMA receiver is modeled with $N_v$ microstrips, each containing $N_e$ tunable metamaterial elements ( $N = N_v \times N_e$ ).
- The analog combining matrix $Q$ is constrained to a specific set of complex weights representing the tunable metamaterial response.
- The quantization process at the $b$ -bit ADCs is modeled using the Bussgang decomposition. This linearizes the quantization effect, expressing the quantized output $z$ as a linear function of the input $y$ plus uncorrelated quantization noise ( $z = F_b y + g$ ), where $F_b$ is the Bussgang gain matrix.
Problem Formulation:
- The objective is formulated as an MSE minimization problem over the digital combiner $W$ and analog combiner $Q$ .
- The problem is inherently non-convex due to the structural constraints on $Q$ and the non-linear relationship between $Q$ and the quantization noise covariance.
Optimization Strategy:
1. Digital Combiner ( $W$ ): For a fixed analog combiner $Q$ , the optimal digital combiner is derived in closed form as the Linear Minimum MSE (LMMSE) estimator: $W_{opt} = C_z^{-1} C_{zs}$ .
2. Analog Combiner ( $Q$ ):
  - The authors utilize matrix quadratic transformation to reformulate the non-convex MSE objective.
  - To handle the complexity of the Bussgang gain in the optimization, they employ a large-user approximation (assuming $K \to \infty$ ). This simplifies the covariance matrices and allows the Bussgang gain $\rho_b$ to be treated as a scalar.
  - The problem is further transformed using vectorization and the Kronecker product to express the objective as a quadratic function of the weight vector.
  - Finally, the non-convex constraint on the weights is relaxed using Semidefinite Relaxation (SDR). The problem is converted into a convex optimization problem over a positive semidefinite matrix $U$ , which is then solved to recover the optimal analog weights.

3. Key Contributions

Joint Hybrid Design Framework: The paper presents the first comprehensive framework for joint analog and digital combiner design specifically for DMA-based XL MIMO systems with low-resolution ADCs.
Handling Quantization and Hardware Constraints: By integrating the Bussgang decomposition with the specific structural constraints of DMAs (microstrip independence), the authors derive a tractable solution for a problem that is otherwise intractable due to non-convexity.
Large-User Approximation: The introduction of the large- $K$ approximation simplifies the Bussgang gain calculation, making the optimization of the analog weights computationally feasible without significant loss of accuracy in high-user regimes.
SDR-Based Solution: The conversion of the non-convex analog weight optimization into a convex Semidefinite Relaxation problem provides a robust method to find near-optimal solutions.

4. Numerical Results

The authors evaluated the proposed scheme using numerical simulations with varying parameters:

ADC Resolution ( $b$ ):
- At low SNR, all ADC resolutions perform similarly as noise dominates.
- At high SNR, higher resolution ADCs significantly reduce MSE.
- 1-bit ADCs suffer from high non-linear distortion, resulting in the highest MSE.
- 3-bit ADCs achieve performance nearly identical to infinite-resolution (ideal) ADCs, demonstrating that very coarse quantization is sufficient for accurate detection in this architecture.
Number of Microstrips ( $N_v$ ):
- Increasing $N_v$ (the number of RF chains) significantly improves performance by providing more spatial degrees of freedom to separate users and suppress interference.
- Performance saturates when $N_v$ exceeds the number of users $K$ .
Elements per Microstrip ( $N_e$ ):
- Increasing $N_e$ (scaling the physical aperture of the DMA) does not change the MSE.
- This confirms that while a larger aperture captures more signal energy, the limiting factor is the number of RF chains ( $N_v$ ) and the digital processing capability, not the number of analog elements per waveguide.

5. Significance

This work is significant for the future deployment of 6G and beyond networks for several reasons:

Scalability: It validates that DMAs are a viable, scalable solution for XL-MIMO, capable of supporting massive antenna arrays without the prohibitive power and cost of full digital RF chains.
Cost-Efficiency: The results prove that low-resolution ADCs (even 3-bit) can be used in DMA receivers without sacrificing detection accuracy, drastically reducing hardware complexity and power consumption.
Design Guidelines: The finding that $N_e$ (elements per strip) does not impact MSE while $N_v$ (number of strips) does, provides crucial design guidance. System architects can focus on increasing the number of RF chains rather than the density of elements within waveguides to improve performance.
Robustness: The proposed framework demonstrates that reliable multi-user detection is achievable even under severe hardware constraints (quantization and structural coupling), bridging the gap between theoretical DMA potential and practical implementation.