⚡ electrical engineering

A Directivity-Dependent Rician K-Factor Model for Indoor Industrial Channels

This paper presents a physics-based, closed-form model linking antenna directivity to the Rician K-factor and delay spread in large indoor industrial environments, validated by 75 GHz ray-tracing simulations, to provide design rules for optimizing wideband mmWave industrial links.

Original authors: Dimitrios C. Tzarouchis

Published 2026-05-06

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Dimitrios C. Tzarouchis

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 Picture: The "Echo Chamber" Problem

Imagine you are in a massive, empty warehouse made of concrete. If you shout, your voice doesn't just travel straight to the listener; it bounces off the walls, the floor, and the ceiling. These bounces (echoes) arrive at the listener's ear at slightly different times.

In the world of wireless internet (specifically the high-speed "E-band" used in factories), this is a problem. When data travels as radio waves, these "echoes" arrive late and muddy the signal, causing errors. The paper calls this "delay spread." The bigger the delay spread, the harder it is to send fast, wideband data.

The Solution: The "Flashlight" vs. The "Lightbulb"

The paper investigates how to fix this using antennas.

The Lightbulb (Omnidirectional Antenna): Imagine a lightbulb that shines light in every direction. In our warehouse, this light hits every wall, creating a chaotic mess of reflections.
The Flashlight (Directive Antenna): Now, imagine a flashlight that focuses its beam into a tight, narrow cone. It shines directly at the listener and ignores most of the walls.

The researchers wanted to create a mathematical "recipe" to predict exactly how much better the "flashlight" works compared to the "lightbulb." They wanted to know: If I make my antenna twice as focused, how much cleaner will the signal become?

The Core Discovery: The "Reverberance Factor"

The authors built a model that links three things:

Antenna Focus (Gain): How tight the beam is.
The "K-Factor": A score that measures how much of the signal is a clear, direct line versus how much is messy, bouncing noise.
The Delay Spread: How "muddy" the signal gets.

They found that as you make the antenna more focused (increase the "gain"), the messy noise drops away. However, it doesn't drop away perfectly. The warehouse isn't a perfect echo chamber; some walls reflect sound better than others.

To account for this, they introduced a "Reverberance Factor" (let's call it the "Room Messiness Score").

If the room were perfectly uniform, the math would be simple.
Because real industrial halls are messy, this factor acts like a "dampener" on the improvement. It tells you that you need more focus than you might expect to get the same result in a messy room.

The "Semicircle" Rule

One of the coolest findings is a geometric rule they discovered. They found that if you plot the "average delay" against the "messiness" (RMS delay spread) for different antenna settings, the points always trace out a perfect semicircle.

Think of it like a slide. No matter how you change the antenna, the relationship between the average delay and the messy delay follows this specific curved path. This allows engineers to predict the future performance of a system just by knowing a few basic numbers about the room.

The "Flashlight" Trade-off

The paper tested this in a giant industrial hall (about the size of a football field) using computer simulations. Here is what they found:

The Good News: Using a highly focused antenna (a "flashlight") drastically reduces the delay spread. It cuts out the late-arriving echoes.
The Catch: The improvement isn't linear. To cut the delay spread in half, you need to increase the antenna's focus significantly more than you might guess.
The "Half-Speed" Rule: They found that the "messiness" (RMS delay spread) shrinks at half the rate of the "average delay."
- Analogy: If you turn up the volume on your flashlight to make the average signal 10 times clearer, the "muddy" part of the signal only gets 3 times clearer (roughly the square root). This means cleaning up the signal is harder than it looks.

Practical Takeaways for Factory Design

The authors created a simple set of rules for engineers designing wireless networks in factories:

You can't do it with just a flashlight: In a huge, concrete factory, even a super-powerful, focused antenna might not be enough to get a "perfectly clean" signal (specifically, getting the delay spread down to 5 nanoseconds).
The "40 dBi" Threshold: To get a decent connection (delay spread under 50 nanoseconds) in this specific type of factory, you need a total antenna power of about 40 dBi.
The "90 dBi" Wall: To get an extremely fast connection (delay spread under 5 nanoseconds), you would need about 90 dBi. The paper suggests this is practically impossible with just antennas; you would need extra electronic help (equalization) to clean up the signal.

Summary

This paper provides a physics-based "map" for engineers. It tells them that while focusing your antenna beam is the best way to stop signal echoes in a factory, there are limits. The map includes a "Room Messiness Score" to help them calculate exactly how powerful their antennas need to be to get a clear connection, saving them from guessing or building systems that are too weak (or unnecessarily expensive).

Technical Summary: A Directivity-Dependent Rician K-Factor Model for Indoor Industrial Channels

Problem Statement
The deployment of E-band (71–86 GHz) links in large, partially reverberant industrial environments (e.g., warehouses, assembly halls) faces significant challenges due to high delay spreads. These environments act as reverberant cavities, producing multipath components that severely degrade wideband performance. While empirical studies at various frequencies (28 GHz to 304 GHz) have confirmed that directive antennas suppress late-arriving multipath and reduce the Root-Mean-Square (RMS) delay spread ( $\sigma_\tau$ ), existing literature lacks a physics-based, closed-form model. Current approaches rely on empirical fits that cannot predict $\sigma_\tau$ or mean excess delay ( $\bar{\tau}$ ) from measurable room parameters or antenna gain. There is a specific need for a unified model that maps total system gain ( $G$ ) to delay statistics using a single, physically interpretable scalar to quantify the departure from an ideal isotropic cavity.

Methodology
The authors derive a physics-based model linking antenna directivity to the Rician K-factor and delay statistics through the following steps:

K-Factor Derivation: The Rician K-factor (ratio of Line-of-Sight to scattered power) is modeled as a function of total system gain ( $G = G_{Tx} + G_{Rx}$ ). The authors introduce a reverberance factor ( $a_K \in [0, 1]$ ) to account for the sub-linear growth of scattered power with gain caused by specular reflections in anisotropic fields. The relationship is defined as $K(G) = K_0 \cdot 10^{a_K G/10}$ , where $K_0$ is the baseline K-factor for an omnidirectional link.
General Identity: A general mathematical identity is derived connecting $\sigma_\tau$ , $\bar{\tau}$ , and $K$ for any arbitrary scatter Power Delay Profile (PDP). This identity reveals that the pair $(\bar{\tau}, \sigma_\tau)$ traces a semicircle parameterized by $K$ , with the radius determined solely by the scatter profile's coefficient of variation ( $CV_s$ ).
Exponential Scatter Model: The authors apply this framework to the physically motivated special case of an exponentially decaying PDP (Sabine–Rice model), where $CV_s = 1$ . This yields explicit closed-form expressions for $\bar{\tau}(G)$ and $\sigma_\tau(G)$ .
Validation via Simulation: The model is validated using Shooting-and-Bouncing-Ray (SBR) simulations in MATLAB. The scenario is the FILL industrial hall in Austria (approx. 57,300 m³, concrete enclosure) at 75 GHz. Simulations were run for 100 random Tx/Rx link placements using four antenna configurations: omnidirectional (0 dBi) and three horn-pair configurations (20, 30, and 40 dBi total gain).
Extraction Methods: Two independent methods were used to extract K-factors from the simulation data:
- Method A: Moment-based extraction using the derived general identity.
- Method B: Direct power integration of the first-arriving ray versus the sum of all subsequent rays.

Key Results

Delay Suppression: Increasing antenna gain from 0 dBi to 40 dBi suppressed multipath power by over 40 dB within 100 ns. The mean excess delay ( $\bar{\tau}$ ) decreased from 104.1 ns to 9.2 ns, and the RMS delay spread ( $\sigma_\tau$ ) decreased from 125.2 ns to 38.2 ns.
Model Validation: The derived three-parameter model $\{\tau_s, K_0, a_K\}$ successfully fit the simulation data. For the FILL hall, the extracted parameters were $\tau_s \approx 127$ ns, $K_0 = 0.32$ , and $a_K = 0.40$ (with $R^2 = 0.91$ for the K-factor fit).
Asymptotic Behavior: The results confirm the theoretical prediction that for high K-factors, $\sigma_\tau$ decays at half the rate of $\bar{\tau}$ with respect to antenna gain (in dB). Specifically, $\bar{\tau} \propto 10^{-a_K G/10}$ while $\sigma_\tau \propto 10^{-a_K G/20}$ .
Discrepancy Analysis: The moment-based K-extraction (Method A) yielded higher K-values than the power-ratio method (Method B) at directive gains. The paper attributes this to the scatter PDP deviating from the exponential model ( $CV_s > 1$ ) under directive conditions, where diffuse scatter is suppressed more than strong specular reflections.

Significance and Design Implications
The paper claims to provide the first compact, closed-form model linking antenna gain to delay spread in reverberant industrial channels. Its primary significance lies in enabling design rules for antenna-gain specification:

The model allows engineers to calculate the minimum total system gain required to meet specific delay-spread targets.
For the simulated FILL hall, achieving $\sigma_\tau \leq 50$ ns requires approximately 40 dBi total gain, while achieving $\sigma_\tau \leq 5$ ns would require approximately 90 dBi.
The authors note that the 5 ns target is effectively unreachable with passive antennas alone, highlighting the necessity for complementary equalization or extreme beamforming.
The derived expressions facilitate the joint optimization of antenna gain and equalizer overhead, as the reduction in $\sigma_\tau$ directly translates to fewer equalizer taps and reduced computational cost in single-carrier modems.

The paper concludes that while the model is validated for a specific empty concrete hall, the framework is general. Future work suggested by the authors includes experimental validation in furnished halls, characterizing the gain-dependence of the scatter coefficient of variation ( $CV_s$ ), and integrating the model with standardized factory channel models (e.g., 3GPP).