🔬 optics

Characterising Interleaver Length in a Turbulent Deployed Terrestrial Free-Space Optical Link

This paper analyzes 24 hours of measurement data from a 4.6 km terrestrial free-space optical link to characterize the relationship between interleaver length and atmospheric turbulence, providing indicative distributions for high-capacity transmission up to 350 Gb/s/pol with low outage probabilities.

Original authors: Kadir Gümüş, Vincent van Vliet, Menno van den Hout, Thomas Bradley, Eduward Tangdiongga, Chigo Okonkwo

Published 2026-08-10

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Kadir Gümüş, Vincent van Vliet, Menno van den Hout, Thomas Bradley, Eduward Tangdiongga, Chigo Okonkwo

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

Imagine trying to have a conversation with a friend across a crowded, windy park. You shout your message, but gusts of wind (turbulence) distort your voice, causing you to stutter or lose words entirely. In the world of high-speed internet, scientists are trying to do something similar, but instead of voices, they are beaming invisible laser light through the air to carry massive amounts of data. This is called Free-Space Optical (FSO) communication. It's like a super-fast, wireless fiber-optic cable that doesn't need wires, making it perfect for connecting buildings in a city or setting up emergency networks.

However, just like your voice in the wind, these laser beams get messed up by the atmosphere. The air isn't perfectly still; it swirls with heat and pressure changes, creating "turbulence" that makes the signal flicker and fade. To fix this, engineers use a trick called "interleaving." Think of this like a game of musical chairs for your data. Instead of sending a long sentence word-by-word (which would get ruined if the wind hits at the wrong moment), you shuffle the words around, send them in a scrambled order, and then unscramble them at the other end. This spreads out the damage so that a sudden gust of wind doesn't wipe out a whole sentence, just a few scattered letters that can be easily fixed. But here's the tricky part: how long should you wait to shuffle the words? If you wait too little, the wind still ruins the message; if you wait too long, the conversation gets delayed, and you lose the speed advantage. Figuring out the perfect wait time based on how windy it is outside is the puzzle this paper solves.

The Great Laser Shuffle: How to Keep the Internet Fast When the Air is Bumpy

In a recent study, a team of researchers from Eindhoven University of Technology decided to test this "shuffle" idea in the real world, not just in a computer simulation. They set up a laser link stretching 4.6 kilometers across the city of Eindhoven, connecting their university campus to the High Tech Campus. It was a real-world testbed, running for a full 24 hours to catch the atmosphere at its best and its worst.

The team wanted to find a clear rulebook: If the air is this bumpy (turbulent), how long should we wait to shuffle the data to keep the connection reliable? They measured the "bumpiness" of the air using a special device called a scintillometer, which tracks something called the refractive index structure parameter ( $C_n^2$ ). In plain English, this number tells you how chaotic the air is. They also measured the power of the laser beam hitting the receiver every 10 microseconds, which is incredibly fast—fast enough to catch even the tiniest flicker.

The researchers were looking at a very ambitious goal: sending data at speeds up to 350 Gb/s/pol (that's 350 gigabits per second for a single color of light!). They wanted to see how often the connection would fail (called "outage") and how to keep that failure rate incredibly low, down to 1 in 10,000 times ( $10^{-4}$ ).

What They Found

After crunching 24 hours of data, the team discovered a very strong, predictable relationship between the bumpy air and the shuffle time. It turns out, the windier the air gets, the longer you have to wait to shuffle the data.

The Bumpier the Air, the Longer the Wait: When the turbulence ( $C_n^2$ ) was low (calm air), they didn't need to shuffle the data much at all for speeds up to 325 Gb/s/pol. But as the air got more turbulent, they had to increase the "interleaver length" (the wait time) significantly.
The Trade-off: If they wanted to be super safe and reduce the chance of failure to just 1 in 10,000 ( $10^{-4}$ ), they needed even longer wait times than if they were okay with a 1 in 100 chance of failure ( $10^{-2}$ ).
The Limits of the Shuffle: There was a catch, though. When the air got really turbulent (specifically when $C_n^2$ was greater than $5 \cdot 10^{-15}$ ), simply waiting longer didn't help enough. Even with a long shuffle, they couldn't maintain the high target speeds without the connection failing too often. In these extreme conditions, they found they had to lower their speed expectations to keep the connection stable.

The data showed that for moderate turbulence (between $10^{-15}$ and $5 \cdot 10^{-15}$ ), the required wait time varied wildly. Sometimes a short wait worked, and sometimes a long one was needed, even if the air looked the same. This suggests that just knowing how "bumpy" the air is isn't always enough to pick the perfect shuffle time; other factors might be at play.

Why This Matters

This study is a big step forward because, until now, there wasn't a clear guide for how to set these shuffling timers in real city-based laser links. Most previous studies focused on satellites, where the air is different, or just guessed. This paper provides a practical map for engineers. It shows that by measuring the air's turbulence, we can dynamically adjust the shuffle time to balance speed and reliability.

While the results are based on one specific day and one specific 4.6 km link, the authors believe these insights apply to other similar systems. They aren't claiming to have solved every problem in the world of laser internet, but they have provided a solid foundation for building systems that can adapt to the weather, keeping our future wireless connections fast and steady even when the air is playing tricks on them.

Technical Summary: Characterising Interleaver Length in a Turbulent Deployed Terrestrial Free-Space Optical Link

Problem Statement
Terrestrial free-space optical (FSO) communication offers high data rates using coherent optical technologies but suffers from reliability issues due to atmospheric turbulence. Unlike satellite links, terrestrial beams traverse the dense lower atmosphere, causing significant fluctuations in received optical power and signal-to-noise ratio (SNR). While data interleaving is a known strategy to mitigate burst errors by spreading them over time—thereby averaging capacity and improving stability at the cost of latency—existing research has largely focused on satellite systems. There is a lack of quantitative frameworks linking atmospheric turbulence conditions to the specific interleaver lengths required to meet reliability targets in practical, deployed terrestrial FSO systems. Consequently, adaptive configuration of interleavers in real-world deployments remains undefined.

Methodology
The authors experimentally investigated the correlation between interleaver length ( $t_{int}$ ) and the refractive index structure parameter ( $C_n^2$ ), a metric for turbulence strength. The study utilized 24 hours of measurement data from the Reid Photonloop testbed, a 4.6 km terrestrial FSO link spanning the city of Eindhoven, Netherlands.

Experimental Setup: The transmitter, located at the Eindhoven University of Technology campus, used a 1542 nm external cavity laser amplified to 29 dBm. The receiver, at the High Tech Campus, employed a high-speed power meter to measure received power ( $P_{rec}$ ) every 10 $\mu$ s.
Turbulence Measurement: A Scintec BLS900 NEO scintillometer measured $C_n^2$ with a 10-minute averaging time, with its transmitter and receiver swapped relative to the FSO link endpoints.
Data Processing: Power data was segmented into one-minute intervals to calculate capacity ( $C$ ) for a single polarization, assuming a 50 GHz bandwidth and a receiver noise figure of 4.4 dB. The system capacity was modeled as $C = W \log_2(1 + \text{SNR})$ , where the total SNR accounts for both back-to-back performance and EDFA-amplified shot-noise limits.
Analysis: The authors applied sliding-window averaging to simulate ideal interleaving with lengths ( $t_{int}$ ) up to 100 ms. They calculated the outage probability ( $P_{out}$ ), defined as the probability that system capacity falls below a target value, for various target capacities (300, 325, and 350 Gb/s/pol) and target outage probabilities ( $10^{-2}$ , $10^{-3}$ , and $10^{-4}$ ).

Key Results
The analysis of 1,328 valid minutes of data (yielding 141 data points after filtering for scintillation measurement gaps) revealed the following:

Correlation between Turbulence and Interleaver Length: A clear positive correlation exists between $C_n^2$ and the required $t_{int}$ . As turbulence intensity increases, longer interleavers are necessary to maintain target capacities and outage probabilities.
Impact of Outage Targets: Stricter outage probability requirements (e.g., reducing $P_{out}$ from $10^{-2}$ to $10^{-4}$ ) necessitate longer interleaver lengths.
Capacity vs. Interleaver Requirements:
- For low turbulence ( $C_n^2 < 10^{-15}$ ), no interleaving is required to achieve capacities up to 325 Gb/s/pol.
- At moderate turbulence ( $10^{-15} \le C_n^2 \le 5 \cdot 10^{-15}$ ), the distribution of required $t_{int}$ becomes very wide for capacities $\ge 325$ Gb/s/pol. In this range, $C_n^2$ data alone is insufficient to reliably determine $t_{int}$ to meet outage targets.
- At high turbulence ( $C_n^2 > 5 \cdot 10^{-15}$ ), the interleaver cannot handle all fading events; significantly reducing the target capacity is required to achieve low outage probabilities.
Distribution Characteristics: For a target capacity of 350 Gb/s/pol, the distribution of required interleaver lengths is generally small at low turbulence, with the exception of specific outliers caused by unusually long fades.

Significance and Claims
The paper claims to address the gap in establishing a quantitative relationship between atmospheric turbulence and interleaver requirements for terrestrial FSO. By providing an indicative distribution of interleaver lengths for capacities up to 350 Gb/s/pol and outage probabilities down to $10^{-4}$ , the authors offer practical design guidelines for adaptive interleaving. The results suggest that while a strong correlation exists between $C_n^2$ and $t_{int}$ , the variability of required interleaver lengths increases with turbulence, highlighting the need for dynamic balancing of reliability and latency in future terrestrial FSO systems. The authors note that while the data is from a single day on a specific link, the analysis provides a basis for dimensioning interleavers in similar terrestrial deployments.

Technical Summary: Characterising Interleaver Length in a Turbulent Deployed Terrestrial Free-Space Optical Link

More like this