Evaluation of EMF Exposure to Throughput Ratio for Sustainable 5G Networks

This paper proposes a stochastic geometry framework using both Poisson and beta-Ginibre point processes to analyze downlink EMF exposure and introduce a new Radiated Energy per Bit Transmitted metric, demonstrating that beta-GPP offers a more accurate model for real-world 5G multi-connectivity networks and highlighting the importance of energy-aware deployment strategies for sustainability.

The Gist

Imagine you are living in a bustling city, and the air is filled with invisible radio waves carrying your phone calls, videos, and internet data. As we move toward faster 5G networks, the city gets crowded with more and more "transmitters" (base stations) to keep everyone connected. But this raises two big questions for the public:

1. Is it safe? (Are we being exposed to too much electromagnetic radiation?)
2. Is it efficient? (Are we wasting energy just to get a few extra bits of data?)

This paper is like a smart city planner's toolkit designed to answer those questions before we build the next generation of networks. Here is a simple breakdown of what they did and what they found.

1. The Map: Random Chaos vs. Organized Order

To predict how these radio waves behave, the researchers had to create a map of where the base stations are located.

• The Old Way (Poisson Point Process): Imagine throwing darts at a dartboard completely at random. Sometimes they clump together, sometimes they are far apart. This is the "random" model most scientists used before. It's easy to calculate, but it doesn't look like real life. In reality, city planners don't put two cell towers right on top of each other; they space them out to avoid interference.
• The New Way (Beta-Ginibre Point Process): Imagine a group of people at a party who naturally want to keep a little personal space. They aren't perfectly arranged in a grid, but they definitely avoid standing on top of each other. This model captures that "repulsion" or spacing.

The Finding: The researchers tested their models against real data from Orange's network in Paris. They found that the "Party Guest" model (Beta-Ginibre) was much more accurate than the "Random Dart" model. It correctly predicted that because towers are spaced out, the radiation exposure is actually slightly different (and often lower) than the old random models suggested.

2. The Double-Act: The 4G and 5G Dance

We are currently in a transition period. We have old 4G towers and new 5G towers. Often, they work together in a setup called EN-DC (Dual Connectivity). Think of this like a delivery service using both a bicycle (4G) and a motorcycle (5G) to deliver a package at the same time.

• The Benefit: You get the package (data) much faster.
• The Cost: You are now being exposed to radio waves from both vehicles simultaneously.

The paper analyzed this "double-act" scenario. They found that while using both technologies together increases the total amount of radiation hitting a user (because two signals are active), it also drastically increases the speed (throughput).

3. The New Scorecard: "Energy Per Bit"

Traditionally, engineers just looked at "How much power is being radiated?" The authors introduced a new, smarter metric called REBT-DL (Radiated Energy per Bit Transmitted).

The Analogy:
Imagine you are paying for a taxi ride.

• Old Metric: "How much gas did the taxi burn?" (Total Energy).
• New Metric (REBT-DL): "How much gas did it take to get me to my destination?" (Energy per Bit).

If a taxi burns a lot of gas but gets you to the airport in 5 minutes, that's actually very efficient. If it burns a little gas but takes 2 hours, that's inefficient.

The researchers found that 5G is the most efficient taxi. Even though it might use powerful beams (like a laser pointer) that concentrate energy in one spot, it delivers data so fast that the "cost" (energy) per piece of information is the lowest. The "Double-Act" (4G + 5G) is a bit less efficient than pure 5G but still better than just using the old 4G alone.

4. The Beamforming Secret

5G towers use something called Beamforming.

• 4G (Omnidirectional): Like a lightbulb in the middle of a room. It shines light everywhere, even where no one is looking. This wastes energy and creates "background noise" (interference) for everyone.
• 5G (Beamforming): Like a flashlight or a laser pointer. It shines a tight beam only at your phone.

Because 5G focuses the energy, it creates higher peaks of radiation for the person it's talking to (which sounds scary), but it creates less pollution for everyone else in the room. The paper confirms that this "laser" approach is generally safer and more sustainable for the city as a whole, provided the network is designed correctly.

The Bottom Line

This paper tells us that:

1. Real-world planning matters: We can't assume cell towers are randomly scattered; they have a natural spacing that makes networks safer and more efficient than we thought.
2. Speed has a price, but it's worth it: While having 4G and 5G running together increases radiation slightly, the massive boost in speed makes the energy cost per download very low.
3. 5G is the green choice: By using focused beams and high efficiency, 5G networks are actually a more sustainable way to handle our exploding data needs, provided we use the right mathematical tools to plan them.

In short, the authors built a mathematical crystal ball that helps network operators build faster, greener, and safer 5G cities without guessing.

Technical Summary

Here is a detailed technical summary of the paper "Evaluation of EMF Exposure to Throughput Ratio for Sustainable 5G Networks."

1. Problem Statement

The rapid densification of base stations (BSs) required for 5G networks raises two critical, often conflicting, challenges:

• Sustainability vs. Performance: While densification improves throughput, it increases energy consumption and electromagnetic field (EMF) exposure, raising public health concerns and environmental costs.
• Modeling Limitations: Existing stochastic geometry models for network analysis predominantly rely on the Poisson Point Process (PPP). PPP assumes BSs are spatially independent, which fails to capture the "repulsion" (regularity) observed in real-world deployments where operators avoid placing BSs too close to one another.
• Multi-RAT Complexity: As networks evolve toward 5G EN-DC (E-UTRAN New Radio - Dual Connectivity), where 4G and 5G transmit simultaneously, there is a lack of analytical frameworks that jointly assess EMF exposure and throughput in these heterogeneous, multi-connectivity scenarios.
• Metric Gap: Traditional metrics evaluate exposure (power) and efficiency (throughput) separately. There is a need for a unified metric that quantifies the trade-off between radiated energy and data delivery.

2. Methodology

The authors propose a stochastic geometry framework to model and analyze downlink EMF exposure and efficiency in 5G networks.

A. Spatial Modeling

• Point Processes: The paper models BS locations using two distributions:
  1. Poisson Point Process (PPP): The standard tractable model assuming spatial independence.
  2. $\beta$-Ginibre Point Process ($\beta$-GPP): A non-Poisson model that captures spatial repulsion between BSs. The parameter $\beta$ ($0 < \beta \le 1$) controls the degree of repulsion ($\beta=1$is standard GPP,$\beta \to 0$ converges to PPP).
• Network Architecture: The study covers standalone 4G, standalone 5G, and EN-DC configurations. In EN-DC, 5G BSs are modeled as being co-located with 4G sites with a probability $p$, reflecting real-world infrastructure sharing.

B. Propagation and Signal Models

• Path Loss: A bounded path-loss model is used to avoid singularity at the origin, with a guard-zone radius $D$.
• Fading: Small-scale fading is modeled as Rayleigh fading.
• Antenna Patterns:
  • 5G: Uses dynamic beamforming with 120° sectors. The gain is a Bernoulli random variable based on beam alignment probability ($\eta$).
  • 4G: Modeled as omnidirectional transmitters.
• Interference: The network is assumed to be interference-limited.

C. Key Metrics

1. Total EMF Exposure ($E$): The sum of received signal power and interference power from all active links (4G and 5G).
2. Radiated Energy per Bit Transmitted in the Downlink (REBT-DL): A novel user-centric metric defined as:
   $$\text{REBT-DL} = \frac{\text{Total EMF Exposure}}{\text{Total Throughput}}$$
   This metric quantifies the energy cost (in terms of radiated power) required to transmit one bit of data, directly linking environmental impact to network performance.

D. Mathematical Derivation

• The authors derive the Cumulative Distribution Function (CDF) for both EMF exposure and REBT-DL.
• They utilize the Gil–Pelaez inversion theorem to transform the Characteristic Function (CF) of the exposure into the CDF.
• Analytical expressions for the CF are derived for both PPP and $\beta$-GPP under single-tier and EN-DC scenarios.

3. Key Contributions

1. Analytical Framework for EN-DC: The paper provides the first tractable stochastic geometry expressions for EMF exposure and REBT-DL in 5G EN-DC networks, accounting for simultaneous 4G/5G transmissions.
2. Introduction of REBT-DL: It proposes a new sustainability metric that balances radiated energy against throughput, offering a more holistic view of network efficiency than power-only metrics.
3. Validation with Real Data: The theoretical framework is validated against:
   • Monte Carlo simulations.
   • Real-world data: Base station locations from the Orange network in Paris (4G at 2600 MHz, 5G at 3500 MHz).
4. Superiority of $\beta$-GPP: The study demonstrates that $\beta$-GPP provides a significantly better fit to real-world BS distributions than PPP by capturing spatial repulsion.

4. Numerical Results and Findings

The study was conducted using data from Paris, fitting $\beta$ values of 0.75 for 4G and 0.83 for 5G, with a co-location probability ($p$) of 0.7.

• Model Accuracy: The $\beta$-GPP model closely matches experimental data, whereas the PPP model consistently underestimates EMF exposure. This confirms that ignoring spatial repulsion leads to inaccurate exposure predictions.
• Exposure Comparison (4G vs. 5G):
  • Low Thresholds: 4G shows slightly higher exposure for more users due to omnidirectional antennas causing widespread interference.
  • High Thresholds: 5G can exhibit higher peak exposure for users in the main lobe of a beam due to beamforming concentration, despite having lower average interference.
• EN-DC Impact: EN-DC configurations result in the highest aggregate EMF exposure due to simultaneous transmissions from two RATs.
• REBT-DL Performance:
  • Single-tier 5G achieves the best (lowest) REBT-DL, indicating the highest energy efficiency and sustainability.
  • EN-DC has higher exposure than single-tier 5G, but its significantly higher throughput compensates for this, resulting in a better REBT-DL than 4G alone.
  • PPP vs. $\beta$-GPP: The PPP model yields a higher REBT-DL (worse efficiency) because it underestimates throughput (due to lower coverage probability in the model) compared to the more realistic $\beta$-GPP.

5. Significance and Conclusion

• Sustainable Network Design: The paper validates that $\beta$-GPP is essential for accurate network planning. Using PPP can lead to flawed conclusions regarding exposure limits and energy efficiency.
• Transition Strategy: The results suggest that while EN-DC increases total radiated power, it is a sustainable transitional strategy because the throughput gains significantly improve the energy-per-bit ratio compared to legacy 4G.
• Future Directions: The framework serves as a foundation for future work involving advanced beamforming, blockages, uplink analysis, and inter-operator exposure studies.

In summary, this paper provides a rigorous mathematical toolset for evaluating the trade-offs between 5G network densification, user exposure, and energy efficiency, proving that realistic spatial modeling is crucial for designing sustainable future networks.