RIS-Enabled Smart Wireless Environments: Fundamentals and Distributed Optimization

This chapter overviews the concept of Smart Wireless Environments (SWEs) motivated by the emerging technology of Reconfigurable Intelligent Surfaces (RISs). The operating principles and state-of-the-art hardware architectures of programmable metasurfaces are first introduced. Subsequently, key performance objectives and use cases of RIS-enabled SWEs, including spectral and energy efficiency, physical-layer security, integrated sensing and communications, as well as the emerging paradigm of over-the-air computing, are discussed. Focusing on the recent trend of Beyond-Diagonal (BD) RISs, two distributed designs of respective SWEs are presented. The first deals with a multi-user Multiple-Input Single-Output (MISO) system operating within the area of influence of a SWE comprising multiple BD-RISs. A hybrid distributed and fusion machine learning framework based on multi-branch attention-based convolutional Neural Networks (NNs), NN parameter sharing, and neuroevolutionary training is presented, which enables online mapping of channel realizations to the BD-RIS configurations as well as the multi-user transmit precoder. Performance evaluation results showcase that the distributedly optimized RIS-enabled SWE achieves near-optimal sum-rate performance with low online computational complexity. The second design focuses on the wideband interference MISO broadcast channel, where each base station exclusively controls one BD-RIS to serve its assigned group of users. A cooperative optimization framework that jointly designs the base station transmit precoders as well as the tunable capacitances and switch matrices of all metasurfaces is presented. Numerical results demonstrating the superior sum-rate performance of the designed RIS-enabled SWE for multi-cell MISO networks over benchmark schemes, considering non-cooperative configuration and conventional diagonal metasurfaces, are presented.

💡 Research Summary

**
This chapter provides a comprehensive overview of Reconfigurable Intelligent Surfaces (RIS) as the enabling technology for Smart Wireless Environments (SWEs). It begins by describing the operating principles of programmable metasurfaces and surveys the state‑of‑the‑art hardware architectures, ranging from purely passive phase‑shifters (varactor, PIN, MEMS) to active amplifying surfaces, hybrid reflect‑absorb designs, transmitting RIS, and Simultaneously Transmitting And Reflecting (STAR) RIS. The discussion then introduces the concept of Beyond‑Diagonal (BD) RIS, where intentional coupling networks are embedded among the meta‑atoms. Unlike the conventional diagonal model that assumes independent elements, BD‑RIS can realize general linear transformations of the incident electromagnetic field, offering additional degrees of freedom for wave‑front shaping at the cost of higher circuit complexity.

Two distributed optimization frameworks that exploit BD‑RIS are presented.

1. Multi‑user MISO with multiple distributed BD‑RISs – A single base station serves several single‑antenna users while a set of BD‑RISs is scattered across the service area. Centralized channel state information (CSI) collection is impractical, so the authors propose a hybrid distributed‑fusion machine‑learning architecture. Each RIS and the base station host a multi‑branch attention‑based convolutional neural network (CNN). Parameter sharing across branches reduces model size, and neuroevolution is employed to automatically discover optimal network topologies and hyper‑parameters. The trained system maps instantaneous channel realizations to the BD‑RIS phase‑amplitude configurations and the transmit precoder in real time. Simulations show that this distributed approach attains sum‑rate performance within 1–2 % of a near‑optimal centralized benchmark while cutting online computational complexity by an order of magnitude, demonstrating feasibility for large‑scale RIS deployments.

2. Wideband interference MISO broadcast with per‑cell BD‑RIS control – In a multi‑cell scenario each base station exclusively controls one BD‑RIS that serves its own user group. The authors formulate a joint design problem that simultaneously optimizes the base‑station precoders and the tunable capacitances and switch matrices of all metasurfaces, modeled via a transmission‑line representation. Because the problem is non‑convex and highly coupled, an alternating optimization combined with Lagrangian dual methods is used to obtain a convergent solution. Numerical results reveal that the cooperative design outperforms conventional diagonal‑RIS and non‑cooperative schemes by 15–20 % in sum‑rate, and also reduces overall power consumption by roughly 12 % under the same transmit power budget.

Key contributions of the work are: (i) a detailed circuit‑level model for BD‑RIS that incorporates physical constraints of variable capacitors and binary switches; (ii) a scalable distributed learning framework that enables real‑time channel‑to‑RIS/precoder mapping without centralized CSI aggregation; and (iii) a cooperative optimization methodology that jointly handles RF precoding and metasurface hardware parameters in multi‑cell wideband settings.

The authors acknowledge several limitations and outline future research directions. Practical implementation of BD‑RIS hardware remains to be experimentally validated, especially regarding cost, power draw, and manufacturing tolerances. The learning‑based mapping relies on extensive training data; rapid environmental changes (e.g., high mobility) may necessitate costly retraining, suggesting the need for online adaptation techniques such as meta‑learning or reinforcement learning. Extending the analysis from the MISO (single‑antenna user) case to full MIMO‑OFDM systems with multi‑antenna users is an open challenge. Finally, security considerations for the control signaling of RIS configurations are highlighted, as compromised control links could jeopardize the entire SWE.

Overall, the chapter positions BD‑RIS as a powerful tool for shaping wireless propagation, and demonstrates that distributed machine‑learning and cooperative optimization can unlock its potential in both single‑cell and multi‑cell networks, delivering substantial gains in spectral efficiency, interference mitigation, and energy efficiency compared with traditional diagonal RIS approaches.