Joint Precoding and Phase-Shift Optimization for Beyond-Diagonal RIS-Aided ISAC System

This paper proposes an alternating optimization framework that jointly designs precoding vectors and beyond-diagonal RIS phase-shift matrices to effectively balance the trade-off between multiuser communication sum rate and sensing beam gain in integrated sensing and communication systems.

The Gist

Here is an explanation of the paper using simple language and everyday analogies.

The Big Picture: The "Smart Mirror" for 6G

Imagine you are trying to talk to a friend in a crowded, noisy room (this is Communication), but you also need to shout out to a specific person across the room to tell them where you are (this is Sensing).

In the future of mobile networks (6G), we want to do both at the same time using the same radio waves. However, buildings and walls often block the signal, making it hard to talk clearly or "see" where things are.

To fix this, scientists use a Reconfigurable Intelligent Surface (RIS). Think of this as a giant, high-tech smart mirror stuck on a wall. Instead of just reflecting light, it reflects radio waves. By changing the angle of the mirror's tiny tiles, it can bend the signal around corners to reach your friend or focus it on a target.

The Problem: The "Old Mirror" vs. The "New Mirror"

The Old Way (Diagonal RIS):
Imagine the smart mirror is made of thousands of tiny, independent tiles. Each tile can only tilt on its own. It's like a crowd of people trying to wave at you, but they can't coordinate with each other. They wave randomly. This works okay, but the signal gets scattered, and you don't get a strong, clear beam.

The New Way (Beyond-Diagonal RIS or BD-RIS):
This paper introduces a "super-mirror." In this new design, the tiles are connected to each other by invisible wires (impedance networks). Now, the tiles can talk to each other. If one tile tilts, it tells its neighbors how to tilt to work together.

• Analogy: Instead of a crowd waving randomly, imagine a synchronized dance team. They move in perfect unison to create a massive, powerful wave that hits exactly where you want it.

The Challenge: The Balancing Act

The researchers faced a tricky problem: How do we make the mirror do two things at once?

1. Talk clearly to 5 different people (Communication).
2. Focus a strong beam on a specific target (Sensing).

Usually, if you focus all your energy on the target, you might not have enough left to talk to the people. If you focus too much on talking, the target might get lost. It's like trying to use a flashlight to read a book and signal a plane at the same time; you have to split the light.

The Solution: The "Magic Recipe"

The authors propose a new mathematical "recipe" to control this super-mirror. Here is how they did it:

1. The "Traffic Cop" (Interference Management):
   When you talk to 5 people at once, their voices often get mixed up (interference). The new method acts like a super-smart traffic cop. It adjusts the mirror so that the signal for Person A goes only to Person A, and Person B's signal goes only to Person B. It untangles the mess, making the conversation clear.

2. The "Spotlight" (Beam Gain Approximation):
   For sensing, they need a tight, powerful spotlight. The method calculates exactly how to tilt the connected tiles to squeeze the signal into a tight beam aimed at the target, making the "sensing" very sharp.

3. The "Dance Partner" (Alternating Optimization):
   Solving the math for the mirror and the transmitter at the same time is incredibly hard (like trying to solve a Rubik's cube while juggling).

   • The Trick: They use a technique called Alternating Optimization.
   • How it works: Imagine two people trying to solve a puzzle together.
     • Step 1: Person A holds the mirror steady, and Person B figures out the best way to send the signal.
     • Step 2: Person B holds the signal steady, and Person A figures out the best way to tilt the mirror.
     • They keep switching back and forth. With every switch, they get closer to the perfect solution. Because they break the big problem into small, manageable steps, they can find the answer very quickly.

The Results: Why It Matters

The researchers ran computer simulations to test their idea.

• The Test: They compared the new "Super Mirror" (BD-RIS) against the "Old Mirror" (Diagonal RIS).
• The Outcome: The Super Mirror won every time.
  • It could talk to users much faster (higher data rates).
  • It could "see" targets much more clearly (better sensing).
  • Most importantly, it offered a flexible trade-off. You could tell the system, "I need 80% focus on talking and 20% on sensing," or vice versa, and it would instantly reconfigure the mirror to do exactly that.

Summary

This paper is about building a super-smart, connected mirror for 6G networks. By letting the mirror's tiny parts work together as a team rather than as individuals, and by using a clever "take-turns" math trick to control them, we can build systems that are much better at both talking (communication) and seeing (sensing) at the same time. This is a huge step forward for technologies like self-driving cars and smart cities, which need to do both things perfectly.

Technical Summary

Here is a detailed technical summary of the paper "Joint Precoding and Phase-Shift Optimization for Beyond-Diagonal RIS-Aided ISAC System".

1. Problem Statement

The paper addresses the performance optimization of Integrated Sensing and Communication (ISAC) systems assisted by Beyond-Diagonal Reconfigurable Intelligent Surfaces (BD-RIS).

• Context: Traditional RISs use a diagonal phase-shift matrix where reflecting elements are independent. This limits beamforming gain in massive MIMO systems. BD-RIS overcomes this by interconnecting elements via impedance networks, allowing for full-rank or block-diagonal reflection matrices (specifically Fully-Connected BD-RIS and Group-Connected BD-RIS).
• Challenge: The authors aim to optimize a multiuser ISAC system where a Base Station (BS) serves $K$ users and senses a target simultaneously. The core difficulty lies in the non-convex nature of the joint optimization problem involving:
  • Communication: Maximizing sum rate by managing multiuser interference.
  • Sensing: Maximizing beam gain toward a specific target direction.
  • Constraints: The BD-RIS phase-shift matrix $\Theta$ must satisfy unitary and symmetric constraints ($\Theta^H\Theta = I, \Theta^T = \Theta$), and the objective function involves high-order (quartic) terms due to the coupling of the precoding vector $W$ and the RIS matrix $\Theta$.

2. Methodology

The authors propose a novel optimization framework that transforms the highly non-convex problem into a solvable form using Alternating Optimization (AO) and specific approximation techniques.

A. System Modeling

• Architecture: An $M$-antenna BS communicates with $K$ single-antenna users and senses a point-like target via an $N$-element BD-RIS. Direct links are assumed blocked.
• Signal Model: The received signal includes the desired signal, multiuser interference, and noise. The sensing performance is characterized by the beam gain in the target direction.
• Objective Function: A weighted sum minimization problem is formulated to balance communication and sensing:
  $$\min_{W, \Theta} \eta \|F - P\|_F^2 + (1-\eta)|F_s - P_s|^2$$
  Where:
  • $F$ is the multiuser beam gain matrix (aiming for diagonalization to suppress interference).
  • $F_s$ is the sensing beam gain (aiming to match a desired gain $P_s$).
  • $\eta \in [0,1]$ is the trade-off weight.

B. Proposed Algorithm

To solve the non-convex problem, the authors employ an Alternating Optimization (AO) algorithm that decouples the variables into three sub-problems, each admitting closed-form solutions:

1. Beamforming Optimization ($W$):

   • With $\Theta$ fixed, the problem is reformulated into a quadratic form.
   • Solved using the Lagrangian method, yielding a semi-closed form solution involving matrix inversion and a Lagrange multiplier found via bisection search.

2. Phase-Shift Matrix Optimization ($\Theta$):

   • Due to the complex constraints of BD-RIS, an Auxiliary Variable ($\psi = \text{vec}(\Theta)$) is introduced.
   • The problem is solved using the Alternating Direction Method of Multipliers (ADMM).
   • The update step for $\Theta$ is formulated as a projection problem onto the set of symmetric unitary matrices.
   • Key Innovation: A Symmetric Unitary Projection is derived using Singular Value Decomposition (SVD) to obtain the closed-form solution for $\Theta$.

3. Desired Beam Phase Optimization ($\theta, \phi$):

   • Auxiliary phase variables are optimized to approximate the desired beam patterns.
   • Solved in parallel using a geometric proposition that aligns the phase of the complex term with the desired phase, yielding a simple modulo operation solution.

3. Key Contributions

• Novel Framework: Proposes a joint optimization framework combining multiuser interference management (via beam gain matrix diagonalization) and sensing beam gain approximation.
• Complexity Reduction: Successfully transforms the original non-convex quartic optimization problem into a series of convex sub-problems with closed-form solutions, significantly reducing computational complexity compared to iterative numerical methods.
• BD-RIS Advantage: Demonstrates that the interconnection of RIS elements (BD-RIS) provides superior flexibility and performance over traditional Diagonal RIS (D-RIS).
• Scalability: The proposed projection method is shown to be extensible to Group-Connected BD-RIS (GBD-RIS) architectures by applying the projection to diagonal blocks.

4. Simulation Results

The authors validate the framework through simulations with $K=5$ users, $M=8$ antennas, and an $8 \times 4$ BD-RIS.

• Trade-off Flexibility: By varying the weight $\eta$, the system can smoothly transition between communication-centric ($\eta=1$, highly diagonal channel gain matrix) and sensing-centric ($\eta=0$, concentrated beam pattern) modes.
• Performance Comparison:
  • BD-RIS vs. D-RIS: BD-RIS-aided systems significantly outperform traditional Diagonal RIS in the trade-off curve between achievable sum rate and sensing beam gain.
  • FBD-RIS vs. GBD-RIS: Fully-Connected BD-RIS (FBD-RIS) achieves the best performance due to higher spatial degrees of freedom (DoF), followed by Group-Connected BD-RIS (GBD-RIS), with traditional D-RIS performing the worst.
• Interference Suppression: The results confirm that BD-RIS effectively suppresses multiuser interference by optimizing the phase shifts to diagonalize the effective channel matrix.

5. Significance

This paper is significant for the development of 6G networks for several reasons:

1. Hardware Efficiency: It leverages the emerging BD-RIS architecture to overcome the limitations of traditional RIS, proving that inter-element connectivity is crucial for high-performance ISAC.
2. Algorithmic Efficiency: The derivation of closed-form solutions for a complex non-convex problem makes real-time implementation feasible, addressing a major bottleneck in current RIS research.
3. System Design: It provides a practical methodology for dynamically balancing the conflicting requirements of sensing accuracy and communication throughput, a critical requirement for future autonomous driving and smart city applications.

In conclusion, the work establishes that BD-RIS is a superior enabler for ISAC systems and provides a mathematically rigorous, computationally efficient algorithm to realize its potential.