# Localized and Distributed Beyond Diagonal Reconfigurable Intelligent Surfaces with Lossy Interconnections: Modeling and Optimization

arXiv: https://arxiv.org/abs/2402.05881 · HTML: https://mlsanity.com/paper/2402.05881/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 8 Feb 2024.

## Built on by (most cited first, 1 of 1)

- [Beyond-Diagonal RIS Under Non-Idealities: Learning-Based Architecture Discovery and Optimization](https://mlsanity.com/paper/2510.15701/) (2025)

## Abstract (arXiv)

Reconfigurable intelligent surface (RIS) is a key technology to control the communication environment in future wireless networks. Recently, beyond diagonal RIS (BD-RIS) emerged as a generalization of RIS achieving larger coverage through additional tunable impedance components interconnecting the RIS elements. However, conventional RIS and BD-RIS can effectively serve only users in their proximity, resulting in limited coverage. To overcome this limitation, in this paper, we investigate distributed RIS, whose elements are distributed over a wide region, in opposition to localized RIS commonly considered in the literature. The scaling laws of distributed BD-RIS reveal that it offers significant gains over distributed conventional RIS and localized BD-RIS, enabled by its interconnections allowing signal propagation within the BD-RIS. To assess the practical performance of distributed BD-RIS, we model and optimize BD-RIS with lossy interconnections through transmission line theory. Our model accounts for phase changes and losses over the BD-RIS interconnections arising when the interconnection lengths are not much smaller than the wavelength. Numerical results show that the performance of localized BD-RIS is only slightly impacted by losses, given the short interconnection lengths. Besides, distributed BD-RIS can achieve orders of magnitude of gains over conventional RIS, even in the presence of low losses.