Overview
This study introduces a universal optimization framework for beyond-diagonal reconfigurable intelligent surfaces (BD-RISs), addressing sum-rate maximization and transmit power minimization in multiuser multi-input single-output (MU-MISO) communication systems. The proposed framework is designed to be independent of specific BD-RIS architectures, integrating diverse circuit topologies through the modeling of their scattering matrices via admittance matrices and associated constraints. Optimization is achieved using custom partially proximal alternating direction method of multipliers (pp-ADMM) algorithms.
Research Context
Reconfigurable intelligent surfaces (RISs) are considered a promising technology for future wireless communication systems. Their perceived value stems from their capacity to control the propagation environment in a hardware- and energy-efficient manner. The conventional understanding of RISs has been extended to include beyond-diagonal RISs (BD-RISs). BD-RISs aim to enhance the capabilities of RISs by incorporating tunable interconnections between their individual elements, which is hypothesized to unlock their full potential. While various algorithms have been developed for BD-RIS optimization, these typically target specific architectural designs, and their scattering matrices frequently exhibit specialized structures. A gap in research exists regarding a universal optimization framework capable of accommodating diverse BD-RIS circuit topologies.
Approach
The research bridges the identified gap by developing an architecture-independent framework for BD-RIS optimization. The primary objectives are sum-rate maximization and transmit power minimization within the context of MU-MISO systems. The methodology involves two key steps:
- BD-RIS Architecture Incorporation: The framework integrates different BD-RIS architectures into its models by establishing a connection between the scattering matrix and the admittance matrix. This step also includes the introduction of appropriate constraints that apply to the admittance matrix.
- Optimization Algorithm Implementation: The problems formulated through the architectural incorporation are solved using custom-designed partially proximal alternating direction method of multipliers (pp-ADMM) algorithms. These pp-ADMM algorithms are characterized by their computational efficiency; each subproblem encountered either possesses a closed-form solution or is readily solvable.
Findings
Simulation results obtained from the application of the proposed framework indicate that the approaches achieve a better trade-off between performance and computational efficiency when compared to existing methods for BD-RIS optimization. The custom-designed pp-ADMM algorithms contribute to this efficiency by ensuring that individual subproblems are either solvable in closed-form or are otherwise easily resolvable.