⚡ electrical engineering

Stacked Intelligent Metasurfaces-Based Electromagnetic Wave Domain Interference-Free Precoding

This paper proposes a hardware-efficient, interference-free multi-stream transmission architecture using stacked intelligent metasurfaces (SIMs) to perform analog wave-domain precoding that compensates for power amplifier nonlinear distortion, while jointly optimizing antenna selection, SIM phase shifts, and power allocation via a recursive oblique manifold algorithm to achieve significant performance gains over traditional methods.

Original authors: Hetong Wang, Yashuai Cao, Tiejun Lv, Jintao Wang, Ni Wei, Jiancheng An, Chau Yuen

Published 2026-01-28

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Hetong Wang, Yashuai Cao, Tiejun Lv, Jintao Wang, Ni Wei, Jiancheng An, Chau Yuen

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

Imagine you are trying to send a group of friends (data streams) to different destinations using a single, powerful flashlight (the transmitter). In a crowded room, the beams of light naturally cross and interfere with each other, making it hard for your friends to see their specific targets clearly.

Traditionally, engineers try to solve this by building a massive, expensive, and power-hungry computer to calculate exactly how to aim the flashlight so the beams don't cross. This is like hiring a team of mathematicians to constantly recalculate the aim every time a friend moves an inch. It's slow, costly, and requires heavy machinery.

This paper proposes a completely different, "magic" way to solve this problem using Stacked Intelligent Metasurfaces (SIMs). Here is the breakdown in simple terms:

1. The Magic Wall (The SIM)

Instead of a heavy computer, the authors use a "smart wall" made of thousands of tiny, adjustable mirrors (meta-atoms) stacked in layers.

The Analogy: Think of this wall not as a computer, but as a living, breathing lens. When the light hits it, the wall doesn't just reflect it; it physically reshapes the light waves as they pass through.
The Innovation: Instead of trying to stop the light beams from crossing (which is hard), this system embraces the crossing. It twists and turns the waves so that when they finally hit your friends, the "messy" crossing actually helps push the signal further away from the danger zone. It turns interference into a helpful push.

2. The "Distortion" Problem (Nonlinear Distortion)

In the real world, the equipment that generates the light (the power amplifier) isn't perfect. It's like a slightly warped lens that bends the light in weird ways depending on how bright the beam is. This is called Nonlinear Distortion (NLD).

The Old Way: Usually, you'd need a separate, complex digital processor to "pre-distort" the signal (mess it up on purpose) so that the warped lens fixes it. This takes a lot of power and hardware.
The Paper's Solution: The authors realized that because the "smart wall" (SIM) is so flexible, it can fix the distortion right there in the light itself. It's as if the wall knows the lens is warped and automatically bends the light in the opposite direction to cancel out the error. No extra computer needed.

3. The "Safety Margin" (Keeping Friends Safe)

To make sure the message gets through, the system tries to maximize the "Safety Margin."

The Analogy: Imagine your friends are standing on a tightrope. The "decision boundary" is the edge of the rope where they might fall off (make a mistake). The "Safety Margin" is how far you can push them toward the center of the rope.
The Goal: The system's job is to use the smart wall to push every single friend as far away from the edge as possible, ensuring they never fall, even if the room is noisy.

4. Choosing the Best Lights (Antenna Selection)

The transmitter has many antennas (light sources), but maybe only a few are needed to serve the friends.

The Strategy: The paper introduces a "greedy" method to pick the best lights. It's like a game of "hot and cold." The system tries adding one light at a time, checking if it pushes the friends further to safety. If it does, it keeps that light. If not, it tries another. This ensures the system uses the fewest, most effective lights possible.

5. The "Smart Brain" (The Algorithm)

Calculating how to twist thousands of tiny mirrors simultaneously is incredibly hard. The authors created a special mathematical tool called the Recursive Oblique Manifold (ROM) algorithm.

The Analogy: Imagine trying to find the highest point on a mountain range that is constantly shifting shape. Instead of climbing every single path, this algorithm is like a super-intelligent hiker who knows exactly which direction to step to climb the fastest, layer by layer, without getting stuck in a valley.

The Results

The paper ran simulations (computer tests) and found:

Better Performance: Their system was significantly better than traditional methods. At a specific signal strength, it performed 20 decibels (dB) better than a system with no special strategies. In simple terms, this is a massive jump in clarity and reliability.
Hardware Savings: It achieves this without needing expensive, power-hungry digital processors to fix signal errors.
Layering Matters: Using more layers of the "smart wall" (more mirrors) made the system much more powerful, allowing it to handle more complex situations.

In Summary:
This paper presents a new way to send wireless signals where the "hardware" (the smart wall) does the heavy lifting of shaping the waves, fixing errors, and using interference to help the signal, rather than fighting it. It's faster, cheaper, and more efficient than the old way of using heavy computers to do the same job.

Technical Summary: Stacked Intelligent Metasurfaces-Based Electromagnetic Wave Domain Interference-Free Precoding

Problem Statement
Future 6G mobile communications require low-cost, low-power, and high-speed transmission with improved spectral and energy efficiency. While Reconfigurable Intelligent Surfaces (RIS) and Stacked Intelligent Metasurfaces (SIMs) offer promising solutions for manipulating electromagnetic (EM) waves, existing precoding architectures face significant limitations. Traditional Interference Exploitation Precoding (IEP), which converts multi-user interference (MUI) into constructive power, typically relies on complex computational hardware and symbol-level optimization, leading to high latency and hardware costs. Furthermore, current SIM-based studies often assume ideal hardware and Gaussian signal inputs, neglecting the nonlinear distortion (NLD) introduced by power amplifiers (PAs) and the constraints of finite alphabet inputs. Additionally, fixed antenna configurations in SIM-assisted systems lack the flexibility to adapt to dynamic user distributions and varying propagation environments.

Methodology
The paper proposes a hardware-efficient, interference-free transmitter architecture leveraging SIMs to perform precoding directly within the analog EM wave domain. The core methodology involves:

System Architecture: A Multi-User Multiple Input Single Output (MU-MISO) downlink system is modeled where a Base Station (BS) with $M$ antennas is covered by an $L$ -layer transmissive SIM. The system integrates modulation and precoding in the wave domain, utilizing the SIM's high degrees of freedom (DoF) to manipulate signal phases and amplitudes.
NLD-Aware Modeling: The study incorporates a memoryless nonlinear Saleh model to characterize the amplitude and phase distortions caused by PAs. Unlike traditional approaches that use digital pre-distortion, this architecture compensates for NLD directly in the wave domain by optimizing the SIM phase shifts.
Frame-Level IEP Strategy: To avoid the prohibitive complexity of symbol-by-symbol optimization, the authors formulate a frame-level IEP problem. The objective is to maximize the minimum "safety margin" (the distance between the received signal and the decision boundary) for all users across a channel coherence block, rather than suppressing interference.
Optimization Algorithms:
- Recursive Oblique Manifold (ROM) Algorithm: To solve the non-convex, non-differentiable problem of optimizing SIM phase shifts across multiple coupled layers, the authors decompose the problem layer-by-layer. They transform the complex-valued objective into a real-valued form and employ Riemannian manifold optimization on an oblique manifold to find optimal phase shifts.
- ROM-Based Alternating Optimization (ROM-AO): For the joint optimization of Transmit Antenna Selection (AS), SIM phase shifts, and Power Allocation (PA), a ROM-AO framework is developed. This includes a greedy safety margin-based AS algorithm to select the optimal subset of antennas and a manifold-based approach for PA.

Key Contributions

SIM-Enabled IEP Architecture: The paper introduces a novel architecture where signal processing and pre-distortion are executed in the EM wave domain using multi-layer transmissive metasurfaces, enabling high-speed operation without dedicated digital hardware for precoding.
NLD Compensation: Unlike prior SIM studies, this work explicitly models and compensates for power amplifier nonlinearities within the wave domain, eliminating the need for complex digital pre-distortion modules at the BS.
Frame-Level Optimization: A frame-level IEP scheme is formulated to reduce control overhead and computational complexity compared to symbol-level precoding, utilizing a max-min safety margin objective.
Advanced Optimization Frameworks: The development of the ROM algorithm to handle complex phase shift coupling across SIM layers and the ROM-AO framework for jointly optimizing antenna selection, phase shifts, and power allocation.
Antenna Selection (AS): A flexible DoF-driven AS scheme is explored, demonstrating that selecting optimal transmit antennas significantly enhances system performance in dynamic environments.

Simulation Results
Numerical simulations validate the proposed schemes under various configurations (e.g., varying numbers of SIM layers, meta-atoms, and SNR levels):

NLD Compensation: The NLD-aware scheme significantly outperforms NLD-unaware counterparts. At an SER of $10^{-3}$ , the NLD-aware SIM design provides an SNR advantage of nearly 4 dB over the NLD-unaware version. Constellation diagrams show that the NLD-aware scheme effectively concentrates points around ideal locations, whereas NLD-unaware schemes exhibit angular shifts.
Layer and Element Scaling: Increasing the number of SIM layers ( $L$ ) and meta-atoms ( $N$ ) improves performance. A 6-layer SIM achieves an 8.46 dB SER gain over a single-layer RIS at 6 dB SNR. However, performance saturates beyond certain thresholds (e.g., $L=7$ or specific $N$ values), indicating optimal resource allocation points.
Algorithm Performance: The proposed ROM algorithm demonstrates robust convergence. Compared to benchmarks like Adaptive Particle Swarm Optimization (APSO) and random phase shifts, the ROM-based approach achieves significantly lower SER.
Joint Optimization Gains: The joint strategy of Antenna Selection and Power Allocation yields substantial improvements. Specifically, at a 12 dB SNR, the strategy with both AS and PA achieves a 20 dB performance gain in terms of the safety margin compared to a system without these strategies. The AS strategy is shown to have a more significant impact on performance than PA alone.
Comparison with BLP: The proposed IEP scheme outperforms traditional Block-Level Precoding (BLP) techniques like Zero-Forcing (ZF), offering a 12.60 dB SER gain at 10 dB SNR by constructively utilizing interference rather than nulling it.

Significance
The paper claims that the proposed SIM-enabled IEP architecture represents a significant advancement in wireless transmission by shifting precoding operations from the digital domain to the analog wave domain. This shift not only reduces hardware complexity and power consumption but also enables the direct compensation of hardware impairments (NLD) that typically degrade performance. By maximizing the safety margin through wave-domain manipulation, the system achieves superior reliability and spectral efficiency. The work demonstrates that leveraging the high DoF of stacked metasurfaces, combined with intelligent antenna selection and power allocation, can effectively overcome the limitations of traditional interference suppression methods and hardware non-idealities, offering a viable path toward efficient 6G communication systems.