Reconfigurable Intelligent Surfaces: Architectures, Models, and Applications

I. INTRODUCTION AND SCOPE OF THE REVIEW

• Communication-theoretic RIS studies often rely on idealized models, while electromagnetic studies are detached from system-level performance, limiting hardware–algorithm co-design –[13].The review identifies this disconnect as increasingly critical as RIS moves toward practical deployment scenarios.Source: §I · p. 1
• The review bridges electromagnetic design principles with wireless communication performance, emphasizing architectures, design and analysis paradigms, and deployment-relevant properties [5],.Those properties include reconfigurability, scalability, efficiency, and controllability.Source: §I · p. 1
• Its scope covers representative communication use cases, practical RIS design and analysis, and challenges and future directions including hardware limits, modeling accuracy, intelligent control, and standardization [3],,, [10]–[13].The article aims to provide researchers and practitioners with a roadmap to RIS technology and its role in future networks.Source: §I · p. 1

II. IMPROVEMENTS IN WIRELESS COMMUNICATIONS WITH PROGRAMMABLE METASURFACES

• In severe blockage or NLoS conditions, RIS can redirect incident signals to shadowed users and create virtual line-of-sight paths; studies report SNR gains of several decibels, including with low-resolution phase control [3],,.These gains are especially valuable in urban and indoor scenarios where relays or added access points may be impractical or costly.Source: §II · p. 2
• By directing reflections toward intended receivers, RIS can reduce multi-user interference and improve spatial selectivity, increasing achievable data rates [7],.With base-station beamforming, RIS can complement massive MIMO and extend its effective aperture without additional active RF chains [9].Source: §II · p. 2
• Mostly passive or semi-passive RIS elements consume negligible power compared with active relays, while 1-bit or 2-bit phase-quantized designs can capture most theoretical gains with minor degradation versus continuous phase,, [7].These findings motivate simple, scalable designs with low hardware complexity.Source: §II · p. 2
• Adaptive wavefront control can reduce multipath fading and Doppler effects and improve reliability, particularly in vehicular, UAV, millimeter-wave, and terahertz scenarios with rapid, severe channel variation [7],, [10].RIS provides adaptability alongside traditional physical- and network-layer solutions.Source: §II · p. 2
• RIS makes propagation part of system design, supporting optimization of secrecy capacity, localization accuracy, and latency; it can also hinder unintended reception and support positioning and sensing [1],, –[9].These capabilities connect smart radio environments with communication, security, and sensing functions.Source: §II · p. 2
• RIS can improve received signal power, spectral and energy efficiency, reliability, and security, with many gains achievable using simple, scalable architectures and limited phase resolution, [7].The review frames these improvements as spanning multiple wireless performance metrics rather than a single use case.Source: §II · pp. 2–3

• RIS design must satisfy hardware-level electromagnetic constraints and system-level wireless performance objectives, motivating hybrid methodologies that combine physical modeling with communication-centric abstractions.Source: §III · p. 3
• Physics-based design engineers unit cells for transformations such as anomalous reflection, refraction, focusing, or polarization control, but optimization becomes computationally prohibitive as surfaces grow.It commonly evaluates unit-cell geometry, tunability, bandwidth, and efficiency through full-wave simulations or circuit-based models.Source: §III · p. 3
• Communication-based design uses phase shifts, reflection coefficients, or scattering responses for tractable system optimization, but idealized models can omit hardware effects, losses, and mutual coupling.These omissions become especially relevant in wideband operation, oblique incidence, and near-field interactions.Source: §III · p. 3
• Ray-optics models efficiently represent redirection on large surfaces, while wave-optics methods capture diffraction, coupling, and resonance at higher computational cost; hybrid strategies balance these trade-offs.Source: §III · p. 3
• Learning-based methods use neural networks to model physical-design-to-electromagnetic-response mappings, reducing repeated full-wave simulations while retaining reasonable accuracy.At system level, learning-assisted schemes address configuration, channel estimation, and joint transceiver–RIS optimization, with reduced overhead supporting adaptation in dynamic environments.Source: §III · p. 3
• Co-design frameworks unify electromagnetic realizations and wireless-system objectives through multilevel modeling, hybrid simulation, and learning-assisted optimization.The section presents these approaches as a way to bridge theoretical performance gains and deployable RIS systems.Source: §III · p. 4

• RIS scalability is constrained by configuration, control-signaling, and optimization complexity as surfaces grow, with CSI and configuration overhead potentially prohibitive in dense networks with dynamic channels.The challenge is exacerbated at millimeter-wave and terahertz frequencies, where coherence times are short and hardware imperfections more pronounced.Source: §IV · p. 4
• Practical RIS elements have finite phase resolution, insertion losses, mutual coupling, and frequency-dependent responses that depart from idealized models.Source: §IV · p. 4
• RIS control-circuit power, especially in active or semi-active designs, raises energy-efficiency and thermal-management concerns in large-scale deployments.Source: §IV · p. 4
• Element grouping, hierarchical control, and approximation-based optimization reduce configuration burden, trading marginal performance loss for substantial overhead reductions and enabling operation with partial or imperfect CSI.Source: §IV · p. 4
• Machine-learning and reinforcement-learning approaches target configuration, channel estimation, and joint transceiver–RIS optimization, reducing training overhead and supporting near-real-time adaptation.Hybrid physics-based and data-driven strategies are described as balancing interpretability, generalization, and robustness.Source: §IV · p. 4
• Low-loss tunable elements, efficient biasing, and compact control electronics may improve RIS efficiency and reliability, while active and hybrid architectures trade increased power and noise for amplification or signal regeneration.Future directions include integration with sensing, cell-free massive MIMO, and network orchestration; realistic field trials and common models, metrics, and control interfaces are identified as important for deployment and interoperability.Source: §IV · p. 4

• The review bridges electromagnetic design principles and system-level wireless communication perspectives through its survey of RIS architectures, modeling frameworks, and design methodologies.Source: §V · p. 5
• It discusses scalability, hardware non-idealities, channel acquisition, and control overhead alongside algorithmic, hardware, and learning-assisted approaches to mitigate them.Source: §V · p. 5
• The review identifies continued co-design across electromagnetic, circuit, and system layers as necessary for transitioning RIS concepts into practical, efficient, and scalable infrastructure.It also points to 6G integration with intelligent network control, integrated sensing and communication, and smart radio environments.Source: §V · p. 5