⚡ electrical engineering

Cell-free massive MIMO Channels in an Urban Environment -- Measurements and Channel Statistics

This paper presents an extensive measurement campaign for cell-free massive MIMO in an urban environment at 3.5 GHz, utilizing a "virtual AP" technique to provide detailed channel statistics such as pathloss, shadowing, and delay spread.

Original authors: Yuning Zhang, Thomas Choi, Zihang Cheng, Jorge Gomez-Ponce, Issei Kanno, Masaaki Ito, Andreas F. Molisch

Published 2026-02-11

📖 4 min read☕ Coffee break read

CC BY 4.0

Original authors: Yuning Zhang, Thomas Choi, Zihang Cheng, Jorge Gomez-Ponce, Issei Kanno, Masaaki Ito, Andreas F. Molisch

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 you are trying to design a massive, city-wide Wi-Fi network that never drops a signal, no matter which alleyway or plaza you walk into. This is the dream of "Cell-Free Massive MIMO"—a futuristic way of providing internet where instead of one giant cell tower serving a whole neighborhood, hundreds of tiny "mini-antennas" (Access Points) are scattered everywhere, like a swarm of bees working together to keep you connected.

But there is a problem: engineers can't build this "swarm" if they don't know how radio waves actually bounce, dodge, and die in a real city.

This paper is essentially a "High-Tech Weather Report" for radio waves. The researchers went into the streets of Los Angeles to map out exactly how signals behave in an urban jungle.

Here is the breakdown of how they did it and what they found:

1. The "Cherry-Picker" Spy Drone (The Setup)

Usually, when scientists measure signals, they use a stationary tower or a person walking around. That’s too slow and doesn't give enough data.

Instead, the researchers used a "Cherry-Picker" truck (the kind used by electricians to fix streetlights). They mounted a high-tech transmitter on the arm of the truck and drove it through the streets at a height of 13 meters.

The Analogy: Imagine trying to map the shadows in a forest. Instead of standing still, you are driving a giant, moving flashlight through every single street, corner, and alley, while 80 different people (the "users") stand in various spots holding receivers. This allowed them to take over 20,000 different measurements. It’s like taking a high-resolution 3D scan of a moving target.

2. The "Obstacle Course" (What they measured)

Radio waves are like echoes in a canyon. They don't just travel in a straight line; they hit buildings, bounce off statues, and get muffled by trees. The researchers looked at three main things:

Pathloss (The "Fading Voice"): As you move away from a sound, it gets quieter. They measured exactly how much "quieter" the signal gets. They found that if you have a clear line of sight (LOS), the signal is strong. But if a building gets in the way (NLOS), the signal doesn't just get a little quieter—it drops off a cliff, like someone closing a heavy door between you and a speaker.
Shadowing (The "Hide and Seek"): Sometimes, even if you are close, a random object (like a thick tree or a corner) creates a "shadow" where the signal disappears. They measured how unpredictable these "shadows" are.
Delay Spread (The "Echo Chamber"): If you shout in a cathedral, the echo lasts a long time. If you shout in a field, it’s instant. In a city, the signal hits a building, then a statue, then a wall, arriving at your phone at slightly different times. This "smearing" of the signal is called Delay Spread. If the echoes are too messy, your phone can't understand the data.

3. The Big Discoveries

The "Street Canyon" Effect: They noticed that in narrow streets, the radio waves act like water in a pipe—they get "guided" along the street. This is good for keeping the signal alive, but it also means the "echoes" behave very specifically depending on which street you are in.
The "Tree Problem": Even if you can see the antenna, if there are leafy trees in the way, the signal gets "muffled." It’s like trying to listen to music through a thick curtain.
The "Density" Lesson: Because the signal drops so much when a building blocks it, the researchers concluded that to make this "swarm" of antennas work, you can't just have a few big ones. You need a dense swarm of many small ones to ensure that even if one is blocked, another one always has a clear "line of sight" to you.

Why does this matter?

If you are building the 6G networks of the future, you don't want to guess where to put antennas. This paper provides the "GPS Map of Radio Behavior." It tells engineers: "If you put an antenna here, expect this much echo, this much fading, and this much signal loss."

By understanding the "personality" of the city's radio environment, they can design systems that are much more reliable, ensuring your video calls don't freeze just because you turned a corner.

Technical Summary: Cell-free Massive MIMO Channels in an Urban Environment

1. Problem Statement

As cellular networks transition from 4G to 5G and 6G, Cell-free Massive MIMO (CF-mMIMO) has emerged as a critical architecture. Unlike conventional cellular systems where users connect to a single concentrated base station, CF-mMIMO distributes numerous access points (APs) across a geographical area, connecting them to a central processing unit. This architecture aims to provide higher data rates, improved reliability, and uniform coverage by eliminating inter-cell interference.

However, existing channel models are largely insufficient for CF-mMIMO. Most current models focus on the link between a user equipment (UE) and a single base station, failing to capture the complex interrelations between multiple UEs and a vast, distributed array of APs. To optimize the design and deployment of CF-mMIMO, accurate, measurement-based channel models that account for urban propagation characteristics are required.

2. Methodology

The authors conducted an extensive measurement campaign in an urban environment (USC University Park Campus, Los Angeles) using a unique "virtual AP" technique.

Measurement Setup:
- Transmitter (AP side): A "cherry-picker" vehicle was used to raise a single antenna to a typical AP height of 13 meters. By moving the vehicle along a precise trajectory, the researchers created a "virtual array" of more than 20,000 possible AP locations.
- Receiver (UE side): An 8-element distributed switched array remained static at ground level (1 m height), covering a 60 m area.
- Signal Characteristics: Measurements were performed at a 3.5 GHz carrier frequency with a 350 MHz bandwidth, providing high delay resolution (2.857 ns).
Data Processing:
- The team used a multi-tone waveform and applied Fast Fourier Transforms (FFT) to extract the Channel Impulse Response (CIR) and Power Delay Profile (PDP).
- Advanced post-processing included noise thresholding, delay gating, and the elimination of "precursors" (direct signal leakage within the sounder hardware).
- Clock Drift Correction: Because GPS synchronization can drift during vehicle movement, a two-stage correction method was implemented using real-time video and geometric distance calculations to maintain phase accuracy.

3. Key Contributions

Unprecedented Scale: The study provides one of the most extensive wideband datasets for CF-mMIMO, involving 80 UE locations and over 20,000 AP locations.
New Virtual Array Method: A modified wideband virtual array technique that allows for high-precision tracking of AP locations in an outdoor urban setting.
Comprehensive Statistical Characterization: The paper provides detailed statistics for pathloss, shadowing, delay spread (DS), and window parameters ( $Q$ -window and $Q$ -tap), which are essential for system-level simulations.
Geometric Interpretation: The study links observed multipath components (MPCs) to specific urban geometries (e.g., building reflections, rooftop diffraction, and foliage obstruction).

4. Key Results

Pathloss and Shadowing:
- LOS (Line-of-Sight): Pathloss coefficient $\alpha \approx 2.95$ ; Shadowing standard deviation $\sigma_S \approx 5.1$ dB.
- NLOS (Non-Line-of-Sight): Pathloss coefficient $\alpha \approx 10.4$ ; Shadowing standard deviation $\sigma_S \approx 16.6$ dB.
- Note: The high LOS coefficient is attributed to antenna gain variations caused by grazing angles of incidence at larger distances.
Delay Dispersion:
- The Root Mean Squared (RMS) Delay Spread was found to be approximately $-80.6$ dB for LOS and $-72.6$ dB for NLOS.
- Contrary to some traditional models, the DS in this urban microcell setup was found to be almost independent of distance in LOS scenarios and showed a negative slope in NLOS scenarios, likely due to the "waveguiding" effect of street canyons.
Sparsity and Windowing: The difference between $Q$ -window (contiguous taps) and $Q$ -tap (arbitrary taps) parameters indicates that the PDPs exhibit significant sparsity.

5. Significance

This research provides the empirical foundation necessary for the practical deployment of CF-mMIMO. The findings suggest that:

Deployment Density: To ensure high performance, networks must be dense enough to guarantee that UEs maintain LOS to at least one AP, given the massive (up to 50 dB) pathloss difference between LOS and NLOS.
Interference Management: The significant attenuation in NLOS links suggests that interference from non-line-of-sight nodes may be naturally limited, potentially allowing for smaller, more efficient AP clusters.
Modeling Evolution: The study highlights that traditional "one-size-fits-all" models are insufficient; future models (like the proposed CUNEC model) must account for street-specific variations and the unique geometry of urban microcells.