⚡ electrical engineering

Measurement-Based Massive MIMO Channel Characterization and Performance Evaluation at FR3 (8 and 15 GHz) Under Equal Physical Aperture

This study characterizes FR3 (8 and 15 GHz) Massive MIMO channels under equal physical aperture constraints, revealing that while the 15 GHz band suffers a 3.0 dB coverage deficit due to increased sparsity, it significantly outperforms the 8 GHz band in spectral efficiency by leveraging higher element counts to overcome channel sparsity, with system performance proving largely insensitive to specific array topologies.

Original authors: Enrui Liu, Pan Tang, Haiyang Miao, Qi Zhen, Jianhua Zhang

Published 2026-04-20

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Enrui Liu, Pan Tang, Haiyang Miao, Qi Zhen, Jianhua Zhang

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 "Goldilocks" Frequency

Imagine the world of wireless internet (like 5G and the upcoming 6G) as a highway system.

Low frequencies (Sub-6 GHz) are like old, wide country roads. They are slow to carry a lot of data, but they travel very far and can easily go around hills and through trees.
High frequencies (Millimeter-wave) are like super-fast, narrow race tracks. They can carry massive amounts of data, but they are fragile; a single tree or a brick wall can block the signal completely.

The "FR3" bands (8 GHz and 15 GHz) discussed in this paper are the "Goldilocks" zone. They are faster than the country roads but not as fragile as the race tracks. They offer a sweet spot for covering cities with high speed.

However, there is a catch: Physics is strict. As you go faster (higher frequency), the signal gets weaker and loses its ability to bounce around obstacles.

The Experiment: The "Same Size Window" Rule

The researchers wanted to know: If we build a base station antenna, can we just switch from the 8 GHz band to the 15 GHz band and get better performance?

To make a fair test, they applied a strict rule: The "Window" must stay the same size.
Imagine the antenna array is a window on a building.

At 8 GHz, the "bricks" (antenna elements) are big. You can only fit 32 bricks in that window.
At 15 GHz, the "bricks" are tiny (because the waves are shorter). You can fit 128 tiny bricks in that exact same size window.

The question was: Does having 4x more tiny bricks at 15 GHz make up for the fact that the signal is weaker and harder to push through walls?

The Findings: The Two-Part Story

1. The Coverage Problem (The "Cell Edge" Struggle)

The Theory: If you have 4x more tiny bricks, you can focus the signal like a laser beam. Theoretically, this should perfectly cancel out the extra weakness of the 15 GHz signal.

The Reality: It helps, but it's not magic.

The Analogy: Imagine trying to shout to a friend across a noisy street.
- At 8 GHz, you have a megaphone with 32 holes. The sound bounces off buildings and reaches your friend even if they are behind a corner.
- At 15 GHz, you have a megaphone with 128 tiny holes. You can focus the sound into a tight beam. If your friend is standing in the open, they hear you clearly (maybe even better than before).
- But, if your friend is hiding behind a thick wall (the "cell edge"), the tight beam can't bend around the corner. The 128 holes can't fix the fact that the wall blocks the sound.

The Result: Even with the super-focused beam, the 15 GHz signal was still about 3 dB weaker at the very edge of the coverage area compared to 8 GHz. The "tiny bricks" couldn't fully overcome the physics of the wall.

2. The Capacity Win (The "Data Highway" Win)

The Theory: Even if the signal is weaker at the edge, does the 15 GHz system carry more data when it works?

The Reality: Yes, absolutely.

The Analogy: Think of the antenna elements as lanes on a highway.
- The 8 GHz system has 32 lanes.
- The 15 GHz system has 128 lanes in the same amount of space.
- Even though the 15 GHz highway is a bit bumpier (sparser channels), having 4x the lanes means you can send 4x more cars (data) at the same time.

The Result: The 15 GHz system crushed the 8 GHz system in terms of speed and capacity. The sheer number of tiny antennas allowed it to pack way more data into the same space, overcoming the "bumpiness" of the channel.

The Surprise: Shape Doesn't Matter Much

The researchers also asked: Does the shape of the antenna window matter? Should it be a long skinny strip (1x32) or a square block (4x8)?

The Result: It barely matters!

The Analogy: Imagine you have 32 people holding flashlights. Does it matter if they stand in a single line or a square?
The Finding: As long as you have the same number of people (antennas), the total light (signal) they can catch is almost the same, regardless of how they are arranged.
Why this is great news: Engineers don't need to build weird, long, skinny antennas that are hard to mount on buildings. They can build compact, square panels that are easier to install, cheaper to ship, and look better, without losing performance.

The Bottom Line

This paper tells us that moving to the new 15 GHz bands for 6G is a smart move, but it changes the rules of the game:

Don't expect perfect coverage everywhere: The 15 GHz signal is still a bit weaker at the very edge of the city compared to older bands. You might need more base stations to cover the same area.
Do expect massive speed: If you are within range, the 15 GHz system is a beast. The ability to pack more antennas into the same space gives us huge speed boosts.
Design freedom: We can build these antennas in compact, square shapes. We don't need to sacrifice engineering practicality for performance.

In short: The 15 GHz band trades a little bit of "reach" for a massive amount of "speed," and thanks to smart engineering, we can build the antennas to fit anywhere.

1. Problem Statement

With the commercialization of 6G, Frequency Range 3 (FR3) (specifically bands around 7–8 GHz and 14–15 GHz) has emerged as a critical spectrum candidate, offering a balance between the wide bandwidth of sub-6 GHz and the high capacity of mmWave. However, a fundamental design challenge exists:

Propagation vs. Hardware Trade-off: As frequency increases, path loss and penetration loss increase, and channels become sparser (fewer multipath components). Conversely, shorter wavelengths allow for more antenna elements to be packed into the same physical space.
The Gap: Existing studies often compare frequencies using inconsistent array sizes or idealized channel models. There is a lack of empirical data evaluating system performance under the realistic constraint of equal physical aperture (fixed base station panel size). It remains unclear whether the increased array gain from packing more elements at higher frequencies can fully offset the inherent propagation degradation in real-world urban environments.

2. Methodology

The authors conducted a comprehensive measurement campaign to address these gaps using a unified approach:

Measurement Platform: A unified dual-band channel sounding platform (Time-Division Multiplexing) was used, capable of operating at both 8 GHz and 15 GHz.
- Transmitter: 128-element Uniform Planar Array (UPA).
- Receiver: 40-element Omni-directional Antenna (ODA) array.
- Synchronization: GPS-disciplined Rubidium clock for nanosecond-level precision.
Scenario: Urban Macro (UMa) environment at the Beijing University of Posts and Telecommunications (BUPT) campus, covering both Line-of-Sight (LoS) and Non-Line-of-Sight (NLoS) conditions.
Experimental Design (Equal Aperture Constraint):
- To ensure a fair comparison, the physical size of the transmit array was kept constant.
- 8 GHz: Used a 32-element array.
- 15 GHz: Used a 128-element array.
- Rationale: Since wavelength $\lambda$ at 15 GHz is roughly half that of 8 GHz, a 4x increase in element count ( $N \propto 1/\lambda^2$ ) maintains the same physical aperture area ( $A_{array}$ ).
Data Processing:
- System calibration to remove RF chain and antenna effects.
- SAGE Algorithm: Used for high-resolution parameter estimation (delay, angle, power) of multipath components.
- Performance Metrics: Analyzed Power Delay Profiles (PDP), Delay Spread, Angular Spread, Path Loss, Coverage (CDF of received power), and Spectral Efficiency (SE).

3. Key Contributions

Unified Cross-Frequency Framework: Established the first measurement-based comparison of 8 GHz and 15 GHz Massive MIMO under a strict equal physical aperture constraint, isolating the effects of frequency scaling from hardware size variations.
Channel Sparsification Characterization: Systematically quantified the transition from rich scattering at 8 GHz to sparse channels at 15 GHz, revealing that higher frequencies suffer from reduced delay spread and concentrated angular dispersion.
Coverage vs. Capacity Trade-off Analysis: Provided empirical evidence on whether the "array gain" from increased element count can compensate for "propagation loss."
Array Topology Robustness: Investigated how different geometric layouts (1×32, 2×16, 4×8) affect performance, challenging the assumption that maximizing horizontal aperture is always necessary.

4. Key Results

A. Channel Characteristics

Sparsity: The 15 GHz channel is significantly sparser than 8 GHz. Higher-order multipath components (reflections, diffractions) are often attenuated below the noise floor at 15 GHz.
Delay Spread: RMS Delay Spread decreases at 15 GHz (mean $\mu$ drops from -7.49 to -7.78 in LoS), indicating a more temporally compact channel.
Angular Spread: Azimuth Angular Spread (ASA) decreases significantly at 15 GHz, while Elevation Angular Spread (ESA) remains largely unchanged (constrained by geometry). This indicates anisotropic sparsification.
Path Loss: The Path Loss Exponent (PLE) and shadow fading standard deviation ( $\sigma$ ) increase at 15 GHz, indicating higher attenuation and greater sensitivity to environmental blockages.

B. Coverage Performance

Theoretical vs. Practical: Theoretically, under fixed aperture and power, received power should be frequency-invariant due to coherent array gain ( $N \propto 1/\lambda^2$ ).
Non-Coherent Gap: Without beamforming, 15 GHz suffers a ~3.9 dB coverage deficit compared to 8 GHz due to penetration loss and sparsity.
Coherent Beamforming Gap: Even with optimal coherent beamforming (Dominant Eigenmode Transmission), a residual ~3.0 dB coverage deficit remains at the cell edge (15% outage probability).
- Implication: Array gain cannot fully compensate for the severe propagation losses and lack of diffraction in deep shadowing scenarios at FR3 frequencies.

C. Spectral Efficiency (SE)

Hardware-Enabled Gain: Despite the sparser channel, the 15 GHz system (128 elements) achieves higher SE (129.0 bit/s/Hz) than the 8 GHz system (32 elements, 113.8 bit/s/Hz) under equal aperture.
Intrinsic vs. Hardware: When element counts are matched (e.g., both at 128 elements), the 8 GHz system outperforms 15 GHz (154.2 vs. 129.0 bit/s/Hz) due to richer multipath.
Conclusion: The SE advantage of 15 GHz in equal-aperture scenarios is strictly a hardware dividend (more elements) compensating for poor channel richness, not an intrinsic channel property.
Topology Independence: System performance (both coverage and SE) is highly insensitive to array geometry. Compact 4×8 arrays perform nearly identically to elongated 1×32 arrays (difference < 1.5%).

5. Significance and Implications

Design Paradigm Shift: The study confirms that FR3 Massive MIMO design must rely on ultra-dense antenna integration (hardware scaling) rather than relying on intrinsic channel richness.
Deployment Flexibility: The finding that array topology has negligible impact on performance allows network operators to adopt compact, 2D panel designs (e.g., 4×8) for FR3 base stations. This simplifies site acquisition, reduces wind load, and eases installation without sacrificing capacity or coverage.
Coverage Limitations: While 15 GHz offers superior capacity, it cannot fully overcome the coverage limitations of high-frequency propagation in dense urban canyons without additional infrastructure (e.g., more cells or repeaters) to address the ~3 dB edge deficit.
Theoretical Foundation: The paper provides a rigorous, measurement-backed baseline for 6G system simulations, moving beyond idealized i.i.d. assumptions to realistic sparse channel models.