STAR Beyond Diagonal RISs with Amplification: Modeling and Optimization

This paper proposes a physically consistent signal model and an alternating optimization framework for downlink sum-rate maximization in STAR beyond-diagonal RISs with per-element amplification, achieving substantial performance gains over conventional passive systems while ensuring hardware feasibility and passivity.

The Gist

Imagine you are trying to have a conversation with a friend in a crowded, noisy room, but there is a giant, solid wall between you. In the world of wireless signals (like your phone connecting to a cell tower), this wall blocks the signal, making the connection weak or non-existent.

To fix this, engineers invented RIS (Reconfigurable Intelligent Surfaces). Think of an RIS as a "smart mirror" made of thousands of tiny tiles. Instead of just reflecting light like a normal mirror, these tiles can be programmed to bend, focus, and steer radio waves around the wall to reach your friend.

However, traditional smart mirrors have two big problems:

1. They are passive: They can only bounce the signal they receive. If the signal is weak when it hits the mirror, it comes out weak. They can't make it louder.
2. They are "diagonal": Each tile works alone. Tile #1 only talks to Tile #1. They don't coordinate with their neighbors, which limits how well they can focus the beam.

The New Idea: The "Super-Active" Smart Mirror

This paper introduces a revolutionary upgrade called the STAR BD-RIS with Amplification. Let's break down what that means using a simple analogy.

1. STAR: The "Two-Way" Mirror

Traditional mirrors only reflect light back. This new mirror is Simultaneous Transmitting and Reflecting (STAR).

• The Analogy: Imagine a glass wall that is smart enough to send half the light through to the other side of the room while reflecting the other half back.
• Why it matters: It can serve people on both sides of the wall at the same time, creating a full 360-degree coverage zone.

2. BD-RIS: The "Team Player" Tiles

"Beyond Diagonal" (BD) means the tiles stop working in isolation.

• The Analogy: In the old system, every tile was a solo musician playing their own note. In this new system, the tiles are a full orchestra. Tile #1 talks to Tile #2, #3, and #4, coordinating perfectly to create a powerful, focused beam of sound (or signal).
• Why it matters: This coordination allows for much sharper, more precise signal focusing, cutting through interference like a laser.

3. Amplification: The "Microphone" Upgrade

This is the biggest breakthrough. The new mirror isn't just a mirror; it has tiny, low-power amplifiers built into every single tile.

• The Analogy: Imagine a relay race. In the old passive system, Runner A passes a whisper to Runner B, who passes a whisper to Runner C. By the end, the message is faint. In this new Active system, every runner has a microphone and a speaker. Runner A whispers, Runner B amplifies it to a normal voice, and passes it on. Runner C amplifies it again. The message arrives loud and clear.
• Why it matters: It fixes the "double fading" problem. In wireless, signals get weak twice (going to the mirror, then going to the user). This new mirror boosts the signal in the middle, ensuring it arrives strong even over long distances.

The Challenge: The "Power Budget"

You might ask, "If they amplify the signal, won't they use too much electricity or break the laws of physics?"

The authors solved this by creating a very strict set of rules (a mathematical model) for how the mirror operates:

• The "No Free Lunch" Rule: The mirror can't create energy out of thin air. It has a total power limit (like a battery).
• The "Per-Tile" Limit: Each tiny tile has its own small limit so it doesn't overheat.
• The "Splitting" Logic: The mirror has to decide, for every tile, how much energy to send through the wall and how much to bounce back. It's like a water faucet that splits a stream into two pipes; the paper figures out the perfect angle to open the faucet so both pipes get enough water without bursting the pipes.

How They Solved It: The "Tuning" Algorithm

Designing a mirror with thousands of moving parts, amplifiers, and splitting rules is incredibly hard. It's like trying to tune a piano with 10,000 keys all at once.

The authors developed a smart computer algorithm (called WMMSE Alternating Optimization) that acts like a master tuner:

1. It adjusts the amplifiers (turning the volume up or down).
2. It adjusts the splitting (deciding how much goes left vs. right).
3. It adjusts the coordination between the tiles (the orchestra tuning their instruments).
4. It repeats this process, getting slightly better every time, until it finds the perfect setting to maximize the conversation speed (sum-rate).

The Results: A Giant Leap Forward

The paper ran simulations to see how this new "Super-Active" mirror compares to the old "Passive" one.

• Low Power Scenarios: When the signal is weak (like in a basement or far away), the new mirror is a game-changer. It performed 1,100% better than the old mirror in some cases. It's the difference between a whisper and a shout.
• High Power Scenarios: Even when the signal is strong, the new mirror still wins, offering about 75% better performance.
• Scalability: The bigger the mirror (more tiles), the bigger the advantage.

In a Nutshell

This paper proposes a new type of "smart wall" for 6G networks. It's not just a mirror; it's a smart, amplifying, two-way, team-coordinating surface. By combining these features with a smart algorithm to manage power, it can drastically improve internet speeds and coverage, especially in places where signals are currently weak or blocked. It bridges the gap between passive mirrors (cheap but weak) and active relays (strong but expensive), offering the best of both worlds.

Technical Summary

Here is a detailed technical summary of the paper "STAR Beyond Diagonal RISs with Amplification: Modeling and Optimization."

1. Problem Statement

The paper addresses the limitations of current Reconfigurable Intelligent Surface (RIS) technologies in 6G and beyond networks. Specifically, it targets the constraints of passive RIS architectures:

• Passive Limitations: Conventional RISs (both Diagonal and Beyond-Diagonal) are passive, meaning they cannot amplify signals. They merely redistribute incident energy, leading to the "double-fading" effect (signal attenuation over the BS-RIS link and the RIS-User link), which severely limits coverage and sum-rate, especially in large-scale or high-path-loss scenarios.
• Active Limitations: While Active RISs (with amplification) exist, they are typically modeled as Diagonal RISs (D-RIS) or lack a rigorous physical model that accounts for hardware constraints (like per-element power caps and noise) when combined with Beyond-Diagonal (BD) coupling.
• The Gap: There is a lack of a physically consistent signal model and optimization framework for Simultaneous Transmitting and Reflecting (STAR) BD-RISs that incorporate per-element amplification. Existing models for passive STAR BD-RISs cannot be directly applied to active systems because active components violate unconditional power conservation laws.

Objective: To develop a physically consistent model for an Active STAR BD-RIS (capable of simultaneous reflection, transmission, and amplification) and solve the sum-rate maximization problem under practical hardware constraints (per-element emission caps, total power budgets, and lossless splitting).

2. Methodology

A. System Modeling

The authors propose a novel signal model for an Active STAR BD-RIS with $N$ cells, where each cell contains one transmitting and one reflecting element. The signal flow is decoupled into three distinct modules:

1. Per-Element Amplification: A diagonal gain matrix $\mathbf{A} = \text{diag}(\beta_1, \dots, \beta_N)$ amplifies the incident signal. This stage also introduces internal noise.
2. Lossless Power Splitting: Diagonal matrices $\mathbf{E}_R$ and $\mathbf{E}_T$ split the amplified energy between reflection and transmission branches, enforcing element-wise energy conservation ($\varsigma_i^2 + (1-\varsigma_i^2) = 1$).
3. Beyond-Diagonal Coupling: Passive coupling matrices $\mathbf{\Phi}_R$ and $\mathbf{\Phi}_T$ (acting on the reflective and transmissive branches, respectively) model the inter-element electromagnetic coupling. These matrices are constrained to be right-unitary ($\mathbf{\Phi}^H\mathbf{\Phi} = \mathbf{I}$) to ensure the coupling network remains passive.

Key Constraints:
Unlike passive RISs where power conservation is unconditional, the active model enforces feasibility based on the operating covariance of the surface:

• Per-Element Emission Cap: The total power emitted by each element (reflection + transmission) must not exceed a hardware limit $P_{\max,i}$.
• Aggregate Power Budget: The total radiated power across the RIS must satisfy a regulatory/thermal limit.
• Passivity of Coupling: The coupling matrices must remain unitary to ensure no additional power is generated by the coupling network itself.

B. Optimization Framework

The problem of maximizing the downlink sum-rate is highly non-convex due to the coupling of beamformers, amplification gains, splitting ratios, and coupling matrices. The authors propose an Alternating Optimization (AO) framework based on the Weighted Minimum Mean Square Error (WMMSE) transformation.

The algorithm iteratively updates the following variables:

1. MMSE Combiners & Weights: Updated in closed-form for each user.
2. Digital Beamformers (BS): Optimized via a waterfilling-like update governed by a single dual variable (Lagrange multiplier) to satisfy the BS power constraint.
3. Amplification Matrix ($\mathbf{A}$): Optimized as a convex quadratic program subject to per-element and total power constraints.
4. Power Splitting Ratios ($\mathbf{E}_R, \mathbf{E}_T$): Optimized using a Cyclic Coordinate Descent method with a global acceptance test. This ensures monotonic improvement by testing a set of candidate values (including stationary points, balance points, and local perturbations) and accepting only those that strictly reduce the global objective.
5. BD-RIS Coupling Matrices ($\mathbf{\Phi}_R, \mathbf{\Phi}_T$): Optimized on the Complex Stiefel Manifold. The authors derive Riemannian gradient steps and use QR/Polar retraction to project updates back onto the manifold, preserving the unitary (passive) constraint at every iteration.

Key Feature: The approach decouples the optimization of reflective and transmissive responses, enabling efficient distributed implementation.

3. Key Contributions

1. Novel Physical Model: The first physically consistent signal model for Active STAR BD-RIS that explicitly separates amplification, lossless splitting, and passive coupling while enforcing hardware-aware power constraints.
2. WMMSE-Based AO Algorithm: A provably convergent algorithm that transforms the non-convex sum-rate problem into a tractable WMMSE problem. It provides closed-form or low-complexity updates for all system variables.
3. Manifold Optimization: The derivation of Riemannian gradient descent with QR/polar retraction for optimizing the beyond-diagonal coupling matrices, ensuring they remain on the complex Stiefel manifold (preserving passivity).
4. Global Acceptance Strategy: A specialized coordinate descent algorithm for power splitting that avoids local minima by evaluating a comprehensive set of candidate values and enforcing global objective improvement.
5. Distributed Implementation: The framework naturally decouples the reflective and transmissive optimization, facilitating distributed hardware implementation.

4. Simulation Results

The authors conducted extensive simulations comparing the proposed Active STAR BD-RIS against a conventional Passive STAR BD-RIS.

• Sum-Rate Gains: The active architecture significantly outperforms the passive one.
  • At low transmit power (10 dBm) with $N=64$ elements, the active system achieved ~1100% higher sum-rate compared to the passive system.
  • At high transmit power (35 dBm), the gain moderated but remained substantial at ~75.9%.
  • Gains were observed across various RIS sizes ($N=16$ to $N=81$) and user configurations (1 or 2 users per zone).
• Scalability: The performance gap widens as the RIS size increases, confirming the scalability of the active architecture.
• Low-Power Regime: The active design is particularly effective in low-power regimes where passive RISs suffer from severe double-fading attenuation. The amplification compensates for this loss, yielding improvements of over an order of magnitude in some cases.
• BS Antenna Count: Increasing the number of BS antennas improves both systems, but the relative advantage of the active RIS remains stable (~93-116%), suggesting that expanding the RIS with active elements is more impactful than simply adding more BS antennas.

5. Significance

This paper establishes a new class of Multi-Functional RIS architectures. By integrating low-power amplification into the Beyond-Diagonal STAR framework, the proposed solution:

• Overcomes Double-Fading: Effectively mitigates the signal attenuation inherent in passive RIS deployments.
• Enables Full-Space Coverage: Simultaneously serves users on both sides of the RIS with controllable power gain.
• Bridges the Gap: Provides a scalable solution that sits between passive RISs and fully active relay systems, offering high spectral efficiency with manageable power consumption.
• Theoretical Rigor: Offers a mathematically sound optimization framework with provable convergence, addressing the complex interplay between amplification, coupling, and hardware constraints.

In conclusion, the proposed Active STAR BD-RIS represents a significant advancement for 6G networks, promising substantial coverage extension and spectral efficiency gains, particularly in challenging propagation environments.