Reconfigurable Intelligent Surfaces Based on Single, Group, and Fully Connected Discrete-Value Impedance Networks

Reconfigurable Intelligent Surfaces (RISs) allow to control the propagation environment in wireless networks by properly tuning multiple reflecting elements. Traditionally, RISs have been realized through single connected reconfigurable impedance networks, in which each RIS element is independently controlled by an impedance connected to ground. In a recent work, this architecture has been extended by realizing more efficient RISs with group and fully connected reconfigurable impedance networks. However, impedance networks tunable with arbitrary precision are hard to realize in practice. In this paper, we propose a practical RIS design strategy based on reconfigurable impedance networks with discrete values. Besides, we address the problem of how to group the RIS elements in group connected architectures. We optimize single, group, and fully connected architectures considering finite-resolution elements, and we compare them in terms of received signal power. Through Monte Carlo simulations, supported by theoretical justifications, we show that only a few resolution bits per reconfigurable impedance are sufficient to achieve the performance upper bound. In particular, while four resolution bits are needed to reach the upper bound in single connected architectures, only a single resolution bit is sufficient in fully connected ones, simplifying significantly the future development of these promising RIS architectures.