SCROOGE: A Physics-Aware Framework for Efficient Orchestration of RIS-Assisted Networks

1. Problem Statement

Reconfigurable Intelligent Surfaces (RISs) are a cornerstone technology for 6G Programmable Wireless Environments (PWEs), capable of manipulating electromagnetic (EM) wave propagation. However, integrating RISs into operational networks faces significant challenges:

• Orchestration Complexity: Existing approaches rely on online optimization to jointly determine RIS configurations, beamforming, and power allocation. In multi-user scenarios with large RIS arrays, this introduces high latency and requires continuous control signaling, which is unsuitable for stringent 6G requirements.
• Lack of Physics-Awareness: Many current methods use idealized statistical fading models (e.g., Rayleigh/Rician) that fail to capture critical physical phenomena such as spatially varying illumination, near-field effects, mutual coupling between RIS elements, and structured multipath from walls.
• Inefficient Resource Management: There is a lack of mechanisms that jointly handle resource allocation, energy efficiency (deactivating unused elements), and admission control based on the actual physical contribution of RIS elements to user performance.

2. Methodology: The SCROOGE Framework

The authors propose SCROOGE, a physics-aware orchestration framework that shifts the computational burden from the real-time operating phase to an offline synthesis phase.

A. Offline Physics-Based Codebook Compilation

Before network operation, a high-fidelity physics-based EM simulation is performed to generate a codebook database. For every potential user location, the system computes:

1. Optimal Codebook Entries (CE): The ideal RIS phase configuration to maximize Signal-to-Noise Ratio (SNR).
2. Per-Element Influence Scores ($v_{k,n}$): A metric quantifying how much each specific RIS element contributes to the macroscopic response for a specific user.
3. System Model: The simulation uses a 3D indoor environment modeling:
   • Direct illumination from the transmitter.
   • Mutual coupling between adjacent RIS unit cells.
   • Specular reflections from uncovered walls (image-source model).
   • Coherent superposition of reradiated fields to calculate exact SNR maps.

B. Online Orchestration Mechanisms

During network operation, SCROOGE uses the precomputed codebook and influence scores to make low-latency decisions without online optimization. It addresses three key functions:

1. Tier-Aware Resource Allocation:

   • Goal: Construct a single "Common Configuration" (CC) to serve multiple users simultaneously while respecting user priority tiers (e.g., pricing classes).
   • Mechanism: Instead of treating all RIS elements equally, SCROOGE uses a weighted voting system. The weight for a user $k$ at element $n$ is a blend of:
     • Tier Priority: Represented by a payment factor ($PF_k$).
     • Physical Influence: The influence score ($v_{k,n}$) of the element.
   • Logic: If an element has low influence, the decision relies purely on tier priority. If an element has high influence, the decision incorporates both tier priority and the element's physical importance, preventing "tier inversions" where low-priority users suffer disproportionately due to critical element misalignment.

2. Energy-Efficient Deactivation:

   • Goal: Reduce hardware power consumption by turning off RIS elements that contribute negligibly to the active users.
   • Mechanism: For each element, the system calculates the maximum influence across all active users. If this value falls below a threshold ($\tau_{off}$), the element is switched to a zero-state (OFF).
   • Benefit: This removes non-contributing degrees of freedom, potentially reducing scattering interference and saving energy with negligible SNR loss.

3. Physics-Aware Admission Control:

   • Goal: Decide whether to admit a new user request based on the current RIS state and the user's tier.
   • Mechanism: The system compares the candidate user's optimal configuration with the currently deployed Common Configuration (CC).
   • Logic: It checks the phase difference on the top $y\%$ of the candidate's most influential elements. Admission is granted only if the phase mismatch is within a tier-dependent tolerance ($x$). Higher-tier users face stricter tolerance requirements to guarantee QoS, while lower-tier users are admitted more leniently.

3. Key Contributions

• Framework Design: Introduction of SCROOGE, the first framework to exploit offline physics-derived descriptors (influence scores) for real-time network orchestration (allocation, energy, admission) without online optimization loops.
• Novel Mechanisms:
  • An influence-aware aggregation rule that balances economic tier priorities with physical element importance.
  • A physics-based energy saving mechanism that selectively deactivates RIS elements based on global influence.
  • A tier-dependent admission control that uses physical compatibility metrics to enforce QoS.
• High-Fidelity Evaluation: The framework is evaluated using a rigorous physics-based EM simulation (not statistical fading) that includes mutual coupling, near-field effects, and wall reflections, providing a realistic assessment of RIS orchestration.

4. Performance Results

The framework was evaluated via 2,000 Monte Carlo simulations in a 3D indoor environment with up to 18 simultaneous users and RIS arrays of 57,600 elements.

• Resource Allocation (Fairness & SNR):

  • SCROOGE significantly outperforms a non-physics-aware baseline in tier-consistency (correlation between user tier and SNR loss).
  • It reduces SNR loss across all tiers. For example, in 1-bit quantization, SCROOGE reduced the SNR loss for Tier 1 users from 18.56 dB (baseline) to 18.38 dB, and for Tier 5 from 12.31 dB to 10.26 dB.
  • The "influence-aware" voting prevents the systematic disadvantage of lower-tier users often seen in purely price-driven baselines.

• Energy Efficiency:

  • SCROOGE successfully deactivates a substantial fraction of RIS elements: ~58% at low load (4 users) and ~24% even at high load (18 users).
  • This deactivation results in negligible SNR degradation (difference of ~0.1–0.2 dB). In some cases (K=6, 8), deactivation slightly improved SNR by suppressing conflicting field components (EM regularization effect).

• Admission Control:

  • The mechanism sharply separates accepted and rejected users.
  • Acceptance Rates: Tier 1 (43.3%) to Tier 5 (96.0%), reflecting stricter QoS gates for premium users.
  • QoS Enforcement: For all tiers, the mean SNR loss of accepted users remained below their specific baseline QoS target, while rejected users consistently exceeded it.

5. Significance

• Scalability for 6G: By moving complex synthesis offline, SCROOGE enables low-latency, low-signaling overhead orchestration suitable for massive RIS deployments in 6G.
• Practicality: The framework moves beyond idealized channel models, addressing real-world constraints like mutual coupling and spatial non-uniformity, making it a viable blueprint for actual deployment.
• Holistic Management: It demonstrates that "physics-aware" data can effectively bridge the gap between physical layer capabilities and network-level management goals (fairness, energy, security), offering a new paradigm for Programmable Wireless Environments.