🔬 applied physics

Measured LWC-Specific Fog Attenuation and Frequency Scaling in Low-THz Channels

This paper presents controlled fog attenuation measurements at 120, 140, and 160 GHz over a 22 m channel, establishing a specific relationship between liquid water content and attenuation while validating a frequency scaling exponent of 1.347 that aligns with ITU-R P.840 predictions.

Original authors: Jiabiao Zhao, Xiaoxiang Li, Kefeng Huang, Yapeng Ge, Hanchen Liu, Jie Yang, Weidong Hu, Houjun Sun, Jianjun Ma

Published 2026-08-11

📖 4 min read☕ Coffee break read

CC BY 4.0

Original authors: Jiabiao Zhao, Xiaoxiang Li, Kefeng Huang, Yapeng Ge, Hanchen Liu, Jie Yang, Weidong Hu, Houjun Sun, Jianjun Ma

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 the air around us is like a giant, invisible highway for invisible radio waves. For decades, we've been building faster and faster roads for our data, moving from the slow, crowded lanes of old Wi-Fi to the super-highways of 5G. Now, scientists are eyeing the "Terahertz" (THz) band, a super-fast, super-wide stretch of the electromagnetic spectrum that could carry massive amounts of data for everything from instant video downloads to super-sharp radar. Think of it as the ultimate express lane. But there's a catch: just like a real highway can be blocked by a sudden fog bank, these high-speed radio waves can get smothered by tiny water droplets in the air. When fog rolls in, it doesn't just make things blurry for our eyes; it acts like a thick blanket that absorbs and scatters these signals, potentially cutting off our connection.

To design systems that can survive a foggy day, engineers need to know exactly how much "signal loss" a specific amount of water causes. Usually, they guess this based on how "foggy" it looks (visibility), but that's like trying to guess how heavy a cloud is just by looking at its color. A more precise way is to measure the actual weight of the water floating in the air, known as Liquid Water Content (LWC). Until now, we've had very few hard measurements of exactly how much fog weighs down these specific super-fast radio waves, especially in the "low-THz" range (around 120 to 160 GHz) where the first real-world devices are being built. Without this data, engineers are flying blind, unsure if their networks will survive a misty morning.

This paper is like a team of scientists setting up a giant, controlled "fog machine" in a laboratory to finally get those missing numbers. Instead of waiting for nature to provide a foggy day, they built a 22-meter-long indoor tunnel and filled it with a uniform cloud of water droplets. They used a laser to count the size and number of every tiny droplet, allowing them to calculate the exact weight of the water (LWC) in the air. Then, they beamed radio signals at three different super-fast frequencies (120, 140, and 160 GHz) through this fog to see how much signal was lost.

The researchers found that the relationship between the weight of the water and the signal loss is very predictable and linear: more water weight means more signal loss, and they measured exactly how much. They compared their new, hard-won numbers against the standard rules used by engineers worldwide (called ITU-R P.840). They discovered that while their measured numbers were slightly lower than the standard prediction (about 2-3% less), the way the loss grew as the frequency increased matched the standard rules almost perfectly. Specifically, they found that as the frequency goes up, the fog gets worse at a rate described by a power law with an exponent of 1.347, which is nearly identical to the 1.343 predicted by the standard model.

Crucially, the paper argues that this slight difference in the total amount of signal lost isn't because the standard model is wrong about how fog behaves; it's likely because the way they measured the water weight in the lab (using a small laser path) wasn't a perfect match for the entire 22-meter tunnel. However, because the pattern of how the fog affects different frequencies is so consistent, the scientists are confident that the standard model's prediction for how fog scales with frequency is accurate. This gives engineers a reliable "rule of thumb" for designing future THz networks: they can now trust that if they know the water content, they can accurately predict how much signal they'll lose, even as they tune their systems to different super-fast frequencies.

Technical Summary: Measured LWC-Specific Fog Attenuation and Frequency Scaling in Low-THz Channels

Problem Statement
Terahertz (THz) wireless systems, particularly in the low-THz band (130–174.8 GHz), face significant link margin erosion due to atmospheric attenuation from suspended liquid droplets (fog). While the International Telecommunication Union Recommendation (ITU-R) P.840 provides a model for fog attenuation based on Liquid Water Content (LWC), previous experimental validations in the THz regime have largely relied on visibility or optical extinction as proxies. These descriptors do not uniquely determine LWC or the Droplet Size Distribution (DSD), leading to a lack of reported attenuation coefficients explicitly referenced to LWC. Furthermore, existing studies have not systematically separated the absolute attenuation coefficient from its frequency dependence, nor have they provided a validated power-law scaling law for fog attenuation across the low-THz spectrum.

Methodology
The authors conducted controlled fog measurements in an enclosed 13 × 6 × 3 m³ chamber over a 22 m folded path (using a reflector) at three frequencies: 120, 140, and 160 GHz.

Experimental Setup: The system utilized a signal generator, frequency multipliers, and horn-lens assemblies with a power sensor. Fog was generated via an ultrasonic generator, creating droplets in the 1–40 µm diameter range, ensuring the entire population remained within the Rayleigh scattering regime (size parameter $2\pi r/\lambda_0 \ll 1$ ).
Data Acquisition: A laser particle size analyzer measured the DSD along a 2.5 m optical path adjacent to the THz channel. Time-aligned DSD data were used to retrieve the time-dependent LWC ( $M(t)$ ) assuming spherical droplets.
Attenuation Extraction: Total measured attenuation was corrected for incremental water-vapor absorption (calculated via ITU-R P.676-13 based on logged humidity) to isolate fog-induced attenuation ( $A_f$ ).
Regression Analysis: The relationship between $A_f$ and $M(t)$ was analyzed using Ordinary Least Squares (OLS). Crucially, the authors treated the comparison with ITU-R P.840 as an errors-in-variables (EIV) problem. This approach accounts for the uncertainty in the regressor (LWC derived from a local 2.5 m sample representing a 22 m path), separating the bias in the absolute coefficient from the frequency-dependent scaling.

Key Results

Linearity: Attenuation exhibited a strong linear relationship with LWC across all frequencies and runs, with coefficients of determination ( $R^2$ ) ranging from 0.799 to 0.952.
Specific Attenuation Coefficients: The mean measured specific attenuation coefficients ( $K_{meas}$ $K_{m e a s}$ ) were:
- 120 GHz: 5.374 (dB/km)/(g/m³)
- 140 GHz: 6.618 (dB/km)/(g/m³)
- 160 GHz: 7.918 (dB/km)/(g/m³)
  These values represent a 47.3% increase from 120 to 160 GHz.
Comparison with ITU-R P.840: The measured coefficients were consistently lower than the P.840 predictions (by 2.14% to 2.77% on average). The authors attribute this one-signed deficit to regression dilution caused by the spatial mismatch between the local DSD sampling and the full path, rather than a fundamental flaw in the P.840 model. The Gini-Frisch reverse-regression bounds confirmed that the P.840 values lie within the statistical uncertainty of the measurements.
Frequency Scaling: By analyzing the ratio of attenuation coefficients across frequencies, the authors derived a power-law exponent ( $n$ $n$ ) for the frequency dependence: $K(f) \propto f^n$ $K (f) \propto f^{n}$ .
- Measured exponent: $n = 1.347 \pm 0.043$ .
- P.840 implied exponent (at 20°C): $n = 1.343$ .
- The agreement (0.3% difference) is an order of magnitude tighter than the absolute coefficient deviations, validating the frequency scaling law independent of absolute calibration bias.
Temperature Validation: The frequency dependence of the exponent was used to estimate the effective droplet temperature, yielding $20.2 \pm 2.3$ °C, which aligns with the logged chamber temperature of 20°C.

Significance and Contributions
This work provides the first LWC-referenced attenuation data for low-THz fog channels, moving beyond visibility-based estimates. Its primary contributions are:

Reference Data: It supplies specific attenuation coefficients at 120, 140, and 160 GHz for link budgeting, filling a gap where previous studies relied on indirect descriptors.
Methodological Rigor: By framing the model comparison as an errors-in-variables problem, the paper successfully decouples absolute coefficient uncertainty from frequency scaling, offering a robust validation method for atmospheric models.
Compact Scaling Law: The paper establishes a validated power-law scaling ( $K \approx 6.62 \cdot (f/140)^{1.35}$ ) for fog attenuation in the low-THz band. This law is insensitive to common calibration biases and provides a reliable tool for system design.
Validation of Physical Models: The results confirm that the Rayleigh-absorption-dominated models underlying ITU-R P.840 accurately describe the frequency dependence of fog attenuation in this band, while also quantifying the minor systematic biases introduced by experimental spatial sampling.

The authors conclude that these results offer a solid foundation for low-THz fog link budgeting and suggest the potential for using multifrequency links to sense fog microphysics.

More like this