Generalized Group Selection Strategies for Self-sustainable RIS-aided Communication

This paper proposes and analyzes various group selection strategies for self-sustainable RIS-aided D2D communication under spatially correlated channels, deriving outage probability expressions and asymptotic bounds for both power splitting and time switching configurations using linear and non-linear energy harvesting models.

The Gist

The Big Picture: The "Smart Mirror" That Runs on Light

Imagine you are trying to talk to a friend across a crowded room, but there is a giant wall blocking your view. You can't shout over it, and you can't walk around it.

In the world of wireless communication (like your phone connecting to the internet), this "wall" is a physical obstacle or a lack of signal. To fix this, scientists invented something called a Reconfigurable Intelligent Surface (RIS). Think of an RIS as a giant, smart mirror made of thousands of tiny tiles.

Normally, a mirror just reflects light. But this smart mirror can be programmed to catch a signal (like a Wi-Fi wave), bounce it off the wall, and aim it perfectly at your friend's phone, creating a new path for the conversation.

The Problem: The Mirror Needs Power

Here's the catch: To change the direction of the signal, the mirror needs electricity. If you have to plug this giant mirror into a wall socket, it's expensive, hard to install, and not very "green."

The authors of this paper are asking: "What if the mirror could power itself?"

They propose a Self-Sustainable RIS. This mirror has a special ability: it can "eat" a tiny bit of the radio signal hitting it to generate the electricity it needs to work. It's like a solar panel that runs on the very light it is reflecting.

The Challenge: Too Many Tiles, Not Enough Time

This smart mirror isn't just one big sheet; it's made of hundreds of small groups of tiles (let's call them Teams).

• The Issue: If you try to control every single tile individually, it takes too much time and energy to figure out how to set them up. It's like trying to conduct an orchestra of 1,000 musicians by giving each one a separate instruction sheet.
• The Solution: The paper suggests grouping these tiles into smaller Teams. Each Team acts as a single unit. This makes the system faster and cheaper.

But now, a new problem arises: Which Team should we use?
Imagine you have 20 Teams of tiles. You can't use all of them at once for one conversation. You have to pick the best one.

• Random Choice: Picking a team by closing your eyes and pointing. (This often leads to a bad connection).
• The Paper's Idea: A smart strategy to pick the k-th best team. This means: "If the absolute best team is busy, let's pick the second best. If that's busy, let's pick the third." It's like a backup plan that ensures you always get a good connection, even if the top choice is unavailable.

The Two Ways to "Eat" the Signal (Powering the Mirror)

Since the mirror powers itself from the signal, it has to decide how to split its attention between Listening/Reflecting (sending data) and Charging (harvesting energy). The paper looks at two ways to do this:

1. The "Splitter" Strategy (Power Splitting):
   Imagine a river flowing into a dam. You split the water: 30% goes to the turbine to make electricity, and 70% flows over the dam to power the town.

   • In this mode, the mirror does both at the same time. It takes a slice of the signal to charge its battery and uses the rest to reflect the message.

2. The "Switcher" Strategy (Time Switching):
   Imagine a light switch. For the first half of the second, the mirror is in "Charging Mode" (absorbing all the signal to get power). For the second half, it switches to "Reflecting Mode" (sending the message).

   • It takes turns doing one thing, then the other.

The "Correlation" Factor: The Crowd Effect

The paper also talks about Spatial Correlation.
Imagine the tiles on the mirror are standing very close together, shoulder-to-shoulder. If one tile gets hit by a signal, its neighbor gets hit almost the same way because they are so close. They aren't independent; they are a "team" in the literal sense.

The authors realized that if you ignore this closeness, your math is wrong. You have to account for the fact that the tiles are "influencing" each other. They found that when tiles are close together, you actually get a better signal boost, but you have to calculate it differently.

The Results: Why This Matters

The authors did a lot of complex math (using tools like "Order Statistics" and "Extreme Value Theory" – which are just fancy ways of saying "predicting the best and worst outcomes in a large group") to prove their ideas work.

Here is what they found:

• Smart Selection Wins: Choosing the "k-th best" group is much better than picking randomly. It's like having a backup generator; if the main one fails, you don't lose power.
• More Groups = Better Performance: If you have a huge mirror with many groups to choose from, your chances of finding a perfect group go up. It's like having a larger pool of candidates to hire from; you're more likely to find the perfect employee.
• The "Square" Shape is Best: When arranging the tiles on the mirror, a square shape works better than a long, thin strip. It's like how a square room feels more balanced and efficient than a long hallway.

The Bottom Line

This paper is about building a self-powered, smart mirror for the future internet (6G and beyond).

• It solves the problem of how to power the mirror without wires.
• It solves the problem of how to pick the best part of the mirror to use at any given time.
• It proves that by being smart about how we group and select these tiles, we can make wireless communication faster, more reliable, and energy-efficient.

In short: They figured out how to make a self-charging, intelligent mirror that knows exactly which part of itself to use to keep your video call from dropping, even when the signal is weak.

Technical Summary

1. Problem Statement

The paper addresses the challenges of deploying Reconfigurable Intelligent Surfaces (RIS) in Device-to-Device (D2D) communication networks, specifically focusing on two critical issues:

• Energy Sustainability: Conventional RISs require external power for phase-shifting operations. The paper investigates self-sustainable RISs that harvest energy from incoming signals to power their operations.
• Channel Estimation Overhead & Spatial Correlation: As RIS element counts grow, channel estimation becomes prohibitively expensive. To mitigate this, RIS elements are grouped into smaller sub-surfaces. However, existing literature often ignores spatial correlation (the statistical dependence between adjacent elements), which becomes significant when element spacing is less than half a wavelength.
• Selection Strategy: In a multi-group RIS, the "best" group might be occupied. The paper seeks to optimize system performance by selecting the $k$-th best available group based on specific metrics, rather than just the absolute best or a random group.

The goal is to derive analytical frameworks for group selection strategies that balance Information Transfer (IT) and Energy Harvesting (EH) under spatially correlated Rician fading channels.

2. Methodology

System Model

• Topology: A single-antenna source ($S$) communicates with a single-antenna destination ($D$) via an RIS ($R$) consisting of $N$ elements partitioned into $B$ non-overlapping groups ($R_i$), each containing $M$ elements ($N = B \times M$).
• Channel Model: The links ($S \to R_i$ and $R_i \to D$) experience Rician fading with spatial correlation. The correlation matrix is modeled using a sinc function based on inter-element spacing.
• Energy Harvesting (EH): The RIS operates in a self-sustainable mode using two configurations:
  1. Power Splitting (PS): The received signal is split into an EH stream (fraction $\rho$) and an IT stream (fraction $1-\rho$).
  2. Time Switching (TS): The RIS alternates between EH mode (fraction $\zeta$ of time) and IT mode.
• EH Models: The analysis considers both a simplified linear EH model and a practical non-linear EH model (based on circuit-specific parameters $a, b, c$).

Analytical Framework

• Order Statistics: The authors utilize higher-order statistics to model the selection of the $k$-th best group. They derive the Probability Density Function (PDF) and Cumulative Distribution Function (CDF) for the $k$-th best Signal-to-Noise Ratio (SNR) and harvested energy.
• Composite Channel Approximation: Due to the complexity of the product of two non-central $\chi^2$ distributions (representing the composite channel), the authors approximate the resulting distribution using a Gamma distribution via moment matching.
• Extreme Value Theory (EVT): To analyze scenarios with a massive number of groups (Large Intelligent Surfaces), the authors apply EVT to derive the asymptotic behavior of the $k$-th best group as the number of groups approaches infinity.

Proposed Strategies

The paper proposes three selection mechanisms:

1. Random Group Selection (RGS): A group is chosen randomly from the set of groups that satisfy minimum energy and data rate requirements.
2. SNR-Based Group Selection (SBGS): Selects the group with the $k$-th highest end-to-end SNR.
3. Energy-Based Group Selection (EBGS): Selects the group with the $k$-th highest harvested energy (provided it also meets the data rate threshold).

3. Key Contributions

1. Generalized $k$-th Best Selection Framework: The paper provides a complete analytical characterization of selecting the $k$-th best group based on both SNR and harvested energy, covering both PS and TS configurations.
2. Parameter Bounds: The authors derive rigorous analytical bounds for the system parameters:
   • Power Splitting Factor ($\rho$): Bounds ensuring sufficient energy is harvested while maintaining a minimum data rate.
   • Time Switching Factor ($\zeta$): Similar bounds for the TS configuration.
   • These bounds are derived for both linear and non-linear EH models, considering worst-case and best-case channel conditions.
3. Spatial Correlation Analysis: Unlike previous works, this paper explicitly incorporates spatial correlation into the group selection analysis, demonstrating how inter-patch spacing affects outage performance.
4. Asymptotic Analysis: Using Extreme Value Theory, the paper analyzes the system performance when the number of RIS groups is very large, providing insights for Large Intelligent Surface (LIS) applications.
5. Comprehensive Outage Analysis: The paper derives closed-form expressions for both Data Outage Probability (failure to meet data rate) and Energy Outage Probability (failure to harvest sufficient energy).

4. Results and Validation

The analytical results were validated against extensive Monte Carlo simulations ($10^6$ iterations). Key findings include:

• Performance of Selection Strategies: The SNR-based (SBGS) and Energy-based (EBGS) strategies significantly outperform Random Group Selection (RGS). Furthermore, selecting the 1st best group yields the best performance, but the $k$-th best strategy offers a robust trade-off when the optimal group is unavailable.
• Impact of Spatial Correlation:
  • Counter-intuitively, the paper shows that while spatial correlation increases outage probability compared to an uncorrelated scenario for a fixed number of elements, it allows for denser packing of elements (smaller inter-patch distance).
  • Denser packing (enabled by correlation modeling) increases the total number of available patches, which ultimately improves the data rate and system gain.
• Grouping vs. No Grouping: Grouping-based strategies (GBS) improve energy efficiency ($E_{eff}$) compared to using the entire RIS for a single link, allowing the RIS to serve multiple users simultaneously. However, increasing the number of groups reduces the elements per group, which can lower the data rate. An optimal group size exists.
• EH Model Impact: The non-linear EH model provides more realistic bounds for $\rho$ and $\zeta$. As received power increases, the required splitting factor ($\rho$) or time switching factor ($\zeta$) decreases to meet energy needs, leaving more resources for data transmission.
• Asymptotic Behavior: As the number of groups ($B$) increases, the outage probability decreases monotonically. The EVT-based analysis confirms that having a large pool of groups significantly enhances the likelihood of finding a high-quality group.

5. Significance

This paper is significant for the design of 6G and beyond wireless networks for several reasons:

• Sustainability: It provides a theoretical foundation for self-sustainable RIS, a crucial requirement for deploying massive RISs in remote or hard-to-wire locations.
• Practicality: By addressing spatial correlation and non-linear energy harvesting, the proposed models are more aligned with real-world hardware constraints than idealized linear/uncorrelated models.
• Scalability: The application of Extreme Value Theory offers a roadmap for designing systems with massive RISs (LIS), where selecting the absolute best group is computationally expensive, but selecting a near-optimal ($k$-th best) group is efficient and highly effective.
• Resource Optimization: The derived bounds for $\rho$ and $\zeta$ offer practical guidelines for network operators to configure RIS parameters dynamically based on channel conditions and energy requirements.

In conclusion, the paper establishes a robust analytical framework for optimizing self-sustainable, grouping-based RIS systems, proving that intelligent group selection strategies can significantly enhance reliability and energy efficiency in correlated fading environments.