Optimizing Rate-CRB Performance for Beyond Diagonal Reconfigurable Intelligent Surface Enabled ISAC

Xiaoqi Zhang, Liang Liu, Shuowen Zhang, Weifeng Zhu, Haijun Zhang

TL;DR

This paper tackles the challenge of simultaneously optimizing communication rate and sensing accuracy in Integrated Sensing and Communication (ISAC) systems that utilize Beyond Diagonal Reconfigurable Intelligent Surfaces (BD-RIS). They propose a novel adaptive Riemannian steepest ascent algorithm, enhanced with a dynamic log-barrier method, to efficiently solve this complex optimization problem.

What Problem Does This Paper Address?

Integrated Sensing and Communication (ISAC) systems promise to revolutionize wireless networks by combining communication and sensing capabilities. Reconfigurable Intelligent Surfaces (RIS) are a key enabler, capable of steering signals to improve coverage and sensing in challenging environments. However, traditional RIS have limitations. This paper focuses on a more advanced "Beyond Diagonal RIS" (BD-RIS), which offers greater flexibility in its scattering matrix design.

The core challenge addressed is how to jointly optimize the base station's (BS) beamforming and the BD-RIS's scattering matrix to maximize the sum rate for users while simultaneously ensuring a certain level of target localization accuracy, quantified by the Cramér-Rao bound (CRB). This is a complex optimization problem due to the non-convexity of the objective function, the unitary constraint on the BD-RIS scattering matrix (which lies on a manifold), and the coupling between the BS and BD-RIS designs. Existing methods struggle with these constraints, particularly the unitary matrix requirement.

Methods

The authors propose an alternating optimization (AO) approach to decompose the problem into two sub-problems: optimizing the BS beamforming and optimizing the BD-RIS scattering matrix.

  1. BS Beamforming Optimization: For a fixed BD-RIS configuration, the BS beamforming matrix is optimized using a weighted minimum mean square error (WMMSE) framework. This involves iteratively updating user coefficients and the BS beamforming vectors using Lagrangian dual methods.
  2. BD-RIS Scattering Matrix Optimization: This is the more novel part. To handle the unitary constraint on the BD-RIS scattering matrix (which resides on a Stiefel manifold), they employ an adaptive Riemannian steepest ascent method.
  3. Log-Barrier Method: To incorporate the CRB constraint into the objective function, they use a dynamic log-barrier method. This transforms the constrained problem into an unconstrained one by adding a logarithmic penalty term that grows as the CRB approaches the threshold. The barrier parameter is dynamically updated to balance accuracy and stability.
  4. Riemannian Steepest Ascent: The problem is then mapped onto the Stiefel manifold (for unitary matrices). The algorithm computes the Riemannian gradient and uses a geodesic-based search to update the BD-RIS scattering matrix, ensuring it remains unitary throughout the optimization. This method is designed to be more efficient than traditional gradient-based methods on manifolds.

Datasets

Not discussed in the paper. The evaluation is performed using simulated system parameters.

Key Insights and Ideas

  • BD-RIS Flexibility is Key: The "beyond diagonal" nature of the BD-RIS, allowing for more complex scattering matrices than simple phase shifts, is crucial for simultaneously improving communication and sensing performance.
  • Manifold Optimization is Necessary: The unitary constraint on the BD-RIS scattering matrix means the problem lives on a geometric manifold. Standard optimization techniques are insufficient; Riemannian geometry is required.
  • Dynamic Log-Barrier for Constraints: Using a dynamic log-barrier method is an effective way to handle the CRB constraint, allowing the optimizer to approach the feasible region smoothly.
  • Adaptive Riemannian Steepest Ascent: This algorithm offers a robust way to navigate the Stiefel manifold and find optimal solutions for the BD-RIS scattering matrix, outperforming traditional methods.
  • Trade-off Management: The proposed method effectively balances the communication rate and sensing accuracy, demonstrating that joint optimization can achieve better performance than fixing one aspect and optimizing the other.

Key Results

  • The proposed BD-RIS enabled ISAC system significantly outperforms conventional RIS-aided ISAC systems in terms of UE sum rate.
  • The adaptive Riemannian steepest ascent method with dynamic log-barrier shows faster convergence and better performance compared to fixed-factor schemes and conjugate gradient methods.
  • The fully connected BD-RIS configuration offers the best performance but comes with higher complexity. Group-connected and single-connected configurations provide a trade-off between performance and complexity.
  • Increasing the number of BD-RIS elements generally improves UE rate, while increasing the number of groups can slow down convergence.
  • The system demonstrates a good trade-off between sensing accuracy (CRB threshold) and communication rate, with the proposed method maintaining a stable UE rate across different CRB requirements.

Limitations and Future Work

  • Complexity: While the proposed method is efficient, the fully connected BD-RIS configuration still presents high computational complexity.
  • Channel State Information: The paper assumes perfect channel state information (CSI), which might not be realistic in practical scenarios.
  • Future Work: The authors suggest applying their framework to other BD-RIS aided systems with different types of constraints on the scattering matrix. Investigating practical CSI acquisition techniques and exploring more resource-efficient BD-RIS configurations would be valuable extensions.

Summary updated

Abstract

This letter considers a beyond diagonal reconfigurable intelligent surface (BD-RIS) aided integrated sensing and communication (ISAC) system, where the BD-RIS can help a multi-antenna base station (BS) serve multiple user equipments (UEs) and localize a target simultaneously. We formulate an optimization problem that designs the BS beamforming matrix and the BD-RIS scattering matrix to maximize UEs' sum rate subject to a localization Cramer-Rao bound (CRB) constraint and an additional unitary matrix constraint for the scattering matrix. Because unitary matrices form a manifold, our problem belongs to constrained manifold optimization. This letter proposes a log-barrier based Riemannian steepest ascent method to solve this problem effectively. Numerical results verify the effectiveness of our algorithm and the performance gain of the BD-RIS aided ISAC systems over the conventional RIS aided ISAC systems.
BibTeX
@article{zhang2025optimizing,
  author = {Xiaoqi Zhang and Liang Liu and Shuowen Zhang and Weifeng Zhu and Haijun Zhang},
  title = {{Optimizing Rate-CRB Performance for Beyond Diagonal Reconfigurable Intelligent Surface Enabled ISAC}},
  journal = {IEEE Communications Letters},
  year = {2025},
  doi = {10.1109/lcomm.2025.3599597},
}
Figure