Electromagnetically Consistent Optimization Algorithms for the Global Design of RIS

TL;DR

This paper develops electromagnetically consistent optimization algorithms for designing reconfigurable intelligent surfaces, modeling them as impedance boundaries, and reformulating complex non-convex problems into polynomially solvable approximations with guaranteed convergence.

Contribution

It introduces a novel modeling approach for RIS as impedance boundaries and proposes polynomial-time algorithms with proven convergence for their optimal design.

Findings

Algorithms converge monotonically in the objective value.

Reformulated problems are solvable as linear quadratically constrained or semidefinite programs.

The approach balances performance and implementation complexity.

Abstract

The reconfigurable intelligent surface is an emerging technology for wireless communications. We model it as an inhomogeneous boundary of surface impedance, and consider various optimization problems that offer different tradeoffs in terms of performance and implementation complexity. The considered non-convex optimization problems are reformulated as a sequence of approximating linear quadratically constrained or semidefinite programs, which are proved to have a polynomial complexity and to converge monotonically in the objective value.