⚡ electrical engineering

Practical RIS Gain without the Pain: Randomization and Opportunistic Scheduling in 5G NR

This paper experimentally demonstrates that integrating a reconfigurable intelligent surface (RIS) with a real-time 5G NR system using random phase configurations and the network's inherent proportional fair scheduling can achieve near-optimal throughput and performance comparable to optimized designs without the complexity of real-time channel estimation.

Original authors: L. Yashvanth, Raju Malleboina, Venkatareddy Akumalla, Nekkanti Guna Sai Kiran, Debdeep Sarkar, Chandra R. Murthy

Published 2026-02-16

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: L. Yashvanth, Raju Malleboina, Venkatareddy Akumalla, Nekkanti Guna Sai Kiran, Debdeep Sarkar, Chandra R. Murthy

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

The Big Idea: "The Magic Mirror That Doesn't Need a Brain"

Imagine you are trying to talk to a friend in a large, noisy room full of pillars and walls. Your voice gets blocked, and your friend can't hear you.

The Problem:
Usually, to fix this, you would need a "smart mirror" (called a Reconfigurable Intelligent Surface or RIS) that can bounce your voice perfectly around the obstacles. But here's the catch: To make this mirror work perfectly, you need to know exactly where your friend is standing, measure the air currents, and constantly adjust the mirror's angle in real-time. This takes a lot of computing power, time, and energy. It's like trying to aim a laser pointer at a moving target while blindfolded, but you have to calculate the math for every single movement.

The Paper's Solution:
The researchers at IISc (Indian Institute of Science) asked a crazy question: What if we just let the mirror spin randomly?

Instead of calculating the perfect angle, they built a mirror that randomly changes its angle every few seconds. Surprisingly, they found that when you combine this "dumb," random mirror with the 5G network's built-in traffic cop, it works almost as well as the super-smart, perfectly calculated mirror.

The Analogy: The "Lucky Dice" and the "Smart Waiter"

To understand how this works, let's use a restaurant analogy.

1. The Setup

The Restaurant (The 5G Network): This is the cell tower sending data to your phone.
The Customers (The User Equipment/UEs): You and your friend are sitting at different tables.
The Waiter (The PF Scheduler): The 5G network has a built-in rule called "Proportional Fairness." The waiter's job is to serve food (data) to the customer who is most hungry right now (has the best signal) but also makes sure everyone gets a turn eventually.
The Magic Mirror (The RIS): This is the device that bounces the signal.

2. The Old Way (Optimized RIS)

In the traditional method, the restaurant manager calculates exactly where every customer is sitting. They then manually adjust the Magic Mirror to point a laser beam of food directly at Customer A, then switch it to Customer B, then back to A.

Pros: Perfect delivery.
Cons: The manager is exhausted, the calculation takes too long, and the system is complicated and expensive.

3. The New Way (Randomized RIS)

In this experiment, the researchers told the Magic Mirror: "Just spin around randomly. Point at Table A for 9 seconds, then Table B for 9 seconds, then maybe Table C, then back to A. Don't think about it."

At first, this sounds terrible. Sometimes the mirror points at the wrong table. But here is the magic trick:

The Waiter is Smart: The 5G Waiter (the scheduler) is always watching.
The Opportunity: When the mirror randomly points at You, your signal becomes super strong. The Waiter sees this and says, "Oh! Customer A has a great connection right now! Let's send them a huge plate of food immediately!"
The Result: Even though the mirror is "dumb" and pointing randomly, the Waiter is "smart" enough to only serve food when the mirror happens to be pointing at the right person.

What Did They Actually Do?

The researchers built a real-life 5G test lab in India.

They built a custom "Smart Mirror": It's a flat panel with 1,024 tiny switches (like pixels) that can flip between two states (ON/OFF) to change how radio waves bounce.
They set up a 5G network: Using real 5G software (OpenAirInterface) and real phones (User Equipment).
They tested two scenarios:
- Scenario A: No mirror at all (just a blocked signal).
- Scenario B: A mirror that changes its angle randomly every 9 seconds.

The Results: "Good Enough" is Great

They measured four things:

Signal Strength (RSRP): The mirror made the signal 7 times stronger (about 7-8 dB better).
Speed (Throughput): The data speed went up by about 20-25%.
Errors (BLER): The number of failed messages stayed low because the system adjusted the speed automatically.
The "Genie" Comparison: They compared their random mirror against a "Genie" (a theoretical perfect mirror that knows exactly where to point).

The Shocking Finding:
The "dumb" random mirror, when paired with the smart 5G Waiter, achieved 90-95% of the performance of the perfect "Genie" mirror.

Why Does This Matter?

Think of it like this:

The Old Way: You hire a team of 100 engineers to constantly calculate the perfect path for your Wi-Fi. It's expensive and slow.
The New Way: You just let the Wi-Fi bounce around randomly, but you have a very smart router that says, "I'll only download the big file when the signal is good."

The Takeaway:
You don't need a super-complex, expensive brain for the mirror. You just need a mirror that changes randomly and a 5G network that is smart enough to grab the opportunity when it arises. This makes 5G (and future 6G) much cheaper, simpler, and easier to build in the real world.

Summary in One Sentence

By letting a "dumb" mirror flip randomly and relying on the 5G network's natural ability to pick the best moment to send data, we can get almost all the speed benefits of high-tech smart mirrors without the headache of complex calculations.

1. Problem Statement

Reconfigurable Intelligent Surfaces (RIS) are a promising technology for enhancing next-generation wireless networks by controlling radio propagation environments. However, traditional RIS deployment faces significant practical hurdles:

High Overhead: Achieving optimal performance requires complex Channel State Information (CSI) estimation, phase optimization, and extensive control signaling.
Implementation Complexity: Integrating RIS hardware with real-time 5G New Radio (NR) systems is difficult, and few experimental studies exist that validate gains in a fully integrated, real-time testbed.
The Trade-off: While theoretical research suggests optimized RIS designs yield high gains, the overhead required to calculate and implement these optimizations often negates the practical benefits in dynamic 5G environments.

The authors address the question: Can an RIS provide near-optimal performance gains without the complexity of real-time phase optimization?

2. Methodology

The authors propose and experimentally validate a low-complexity approach that replaces explicit RIS optimization with randomization coupled with the inherent opportunistic scheduling of 5G NR.

A. Experimental Testbed

Hardware: An in-house built 1-bit coded RIS (32x32 elements, 1024 unit cells) operating at 5 GHz with 300 MHz bandwidth. It uses PIN diodes for switching between two phase states (ON/OFF).
Software: A real-time 5G NR testbed built using the OpenAirInterface (OAI) framework.
- Base Station (gNB): Intel i7 computer running OAI 5G RAN stack.
- User Equipment (UE): Two UEs connected via directional horn antennas.
- Channel: The direct BS-UE link is physically blocked to isolate the RIS effect.
Scheduling: The system utilizes the standard 5G Medium Access Control (MAC) layer Proportional Fair (PF) scheduler. This scheduler allocates resources to the UE with the best instantaneous channel conditions relative to their average throughput.

B. The Proposed Approach

Instead of calculating the optimal phase shifts for a specific UE, the RIS phase configuration is randomized:

Random Switching: The RIS alternates between two distinct beamforming states (aligned with UE-1 and UE-2) with equal probability ( $p=0.5$ ) at fixed time intervals ( $T_s = 9$ seconds).
Opportunistic Exploitation: The 5G PF scheduler naturally detects which UE currently has the strongest channel (due to the RIS beam alignment) and prioritizes data transmission to that UE.
Parameter Tuning: The performance depends on the PF scheduler's averaging window parameter ( $\alpha$ ). A small $\alpha$ prioritizes instantaneous channel quality (opportunism), while a large $\alpha$ prioritizes fairness.

3. Key Contributions

Real-Time 5G NR Integration: The paper presents the first experimental demonstration of an in-house RIS integrated with a real-time 3GPP-compliant 5G NR testbed using OAI.
Randomization Strategy: It proposes a novel method where random RIS configurations, when combined with native 5G scheduling, achieve performance comparable to optimized designs, eliminating the need for CSI estimation and phase optimization.
Comprehensive Metric Analysis: The study provides empirical data on Reference Signal Received Power (RSRP), Block Error Rate (BLER), Modulation and Coding Scheme (MCS) index, and throughput under random RIS configurations.
Validation of Multi-User Diversity: The work demonstrates that the system effectively exploits multi-user diversity, capturing RIS beamforming gains without explicit per-element control.

4. Key Results

The experiments were conducted with two UEs (UE-1 at 30°, UE-2 at 45°) under a single-UE and a multi-UE scheduling scenario.

RIS Gain vs. No RIS:
- RSRP: The presence of the RIS increased RSRP by approximately 7–8 dB (a ~7x increase in signal power).
- Throughput: Data rates increased by 20–25% for both UEs compared to the no-RIS scenario.
Performance of Randomized RIS with PF Scheduling:
- Dynamic Adaptation: As the RIS randomly switched beams, the PF scheduler automatically selected the UE with the high RSRP.
- BLER & MCS: The system maintained a target BLER (~0.1). When the RIS aligned with a UE, the channel quality improved, causing the scheduler to increase the MCS index (higher order modulation), thereby boosting throughput. When the beam misaligned, the MCS dropped to maintain reliability.
- Throughput vs. $\alpha$ :
  - With a large $\alpha$ (fairness-focused), throughput was lower as the scheduler ignored instantaneous channel gains.
  - With a small $\alpha$ (opportunistic), the scheduler prioritized the UE with the aligned RIS beam.
  - Crucial Finding: A randomized RIS with a properly tuned PF scheduler achieved near-optimal throughput, comparable to a "genie-aided" scenario where the RIS is perfectly optimized for the scheduled UE in every slot.
Scheduling Efficiency: The scheduler allocated approximately 40% of the time slots to each UE when the RIS was aligned with them, demonstrating that the system maintains fairness while maximizing aggregate throughput.

5. Significance and Conclusion

This paper provides a practical proof of concept that integrating RIS into 5G networks does not necessarily require complex, high-overhead optimization algorithms.

Low Complexity: By leveraging the existing opportunistic scheduling mechanisms of 5G NR, the system captures the majority of RIS benefits (beamforming gains) simply by randomizing the RIS states.
Scalability: This approach avoids the computational burden of real-time CSI estimation and phase calculation, making RIS deployment more feasible for large-scale, real-world 5G networks.
Future Work: The authors note that while the current study uses a 2-UE indoor setup, the methodology extends to multi-antenna and large-scale user deployments. Future work will focus on the analytical relationship between the RIS switching interval ( $T_s$ ) and the scheduler parameter ( $\alpha$ ).

In summary, the paper demonstrates that "Randomization + Opportunistic Scheduling" is a sufficient and highly effective strategy for RIS-aided 5G, delivering high performance without the "pain" of complex optimization.