⚡ electrical engineering

Exploiting Out-of-Band Information for Millimeter-Wave MIMO Channel Estimation: Performance in Static and Dynamic Scenarios

This paper demonstrates that leveraging out-of-band sub-6 GHz information significantly enhances the spectral efficiency of fully digital mmWave MIMO channel estimation in both static and dynamic environments across various pilot configurations and propagation conditions.

Original authors: Faruk Pasic, Mariam Mussbah, Stefan Schwarz, Markus Rupp, Christoph F. Mecklenbräuker

Published 2026-04-06

📖 4 min read☕ Coffee break read

CC BY 4.0

Original authors: Faruk Pasic, Mariam Mussbah, Stefan Schwarz, Markus Rupp, Christoph F. Mecklenbräuker

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 have a conversation with a friend in a very noisy, crowded stadium. You have two ways to talk to them:

The "Whisper" (Millimeter-Wave/mmWave): This is a super-fast, high-speed connection (like a laser beam). It can carry a huge amount of data (like streaming 8K video), but it's very fragile. If a bird flies between you, or if you move your head slightly, the signal breaks. It's like trying to whisper a secret across a football field; it's fast if it works, but very hard to keep the line open.
The "Shout" (Sub-6 GHz): This is a slower, older connection (like a standard walkie-talkie). It's not as fast, but it's tough. It can punch through walls, ignore the noise, and stay connected even if you're running around.

The Problem:
Future 6G networks want to use the "Whisper" (mmWave) for everything because it's so fast. But to make that whisper work, the phone and the tower need to know exactly where to point their antennas. This is called Channel Estimation.

In the past, to figure out where to point, the tower would send out a bunch of "test signals" (pilots) and wait for the phone to reply.

The Catch: In a fast-moving car (dynamic scenario), by the time the tower gets the reply, the car has moved. The "map" they drew is now outdated. It's like trying to aim a water hose at a moving target while blindfolded; by the time you aim, the target has moved.

The Solution (The "Out-of-Band" Trick):
This paper proposes a clever trick: Use the "Shout" to help the "Whisper."

Since the "Shout" (Sub-6 GHz) is so strong and reliable, the tower can use it to get a rough idea of where the car is and how it's moving. It's like using a loud, clear shout to locate your friend in the stadium, and then using that location to aim your whisper.

The researchers tested two main ideas:

The Old Way: Just use the mmWave "Whisper" to find the path.
The New Way (OOBA-MRC): Use the reliable "Shout" to help guide the "Whisper."

The Experiments (The "Pilot" Patterns):
To find the path, the tower sends out "test signals" (pilots). The paper tested three different ways to send these tests:

The "Flash" (1-symbol): Send all the tests instantly in one split second. Pros: Fast. Cons: Not enough data if the signal is weak.
The "Strobe" (2-symbol): Send tests over two seconds. Pros: Better data. Cons: Takes a bit longer.
The "Marathon" (4-symbol): Send tests over four seconds. Pros: Very accurate data. Cons: Takes a long time.

The Big Findings:

When you are standing still (Static Scenario):
- If the "Whisper" is weak (lots of obstacles), taking more time to send more "test signals" (the Marathon approach) helps a lot. You get a better map.
- If the "Whisper" is strong (clear line of sight), it doesn't matter how many tests you send; the signal is already good.
When you are moving fast (Dynamic Scenario):
- The Trap: If you are driving fast, the "Marathon" approach (sending tests over 4 seconds) is actually bad. By the time you finish the tests, you've moved so far that the map is useless. This is called "Channel Aging."
- The Winner: The "Flash" approach (1-second test) is best for fast movement because the map is fresh.
- The Magic of the New Method: Here is the cool part. When using the "Shout to help the Whisper" method (OOBA-MRC), it doesn't matter if you are moving fast or slow. Because the "Shout" is so reliable, it keeps the "Whisper" on target even when the car is speeding at 200 km/h. The new method stays strong while the old method falls apart.

The Bottom Line:
This paper proves that by listening to the "loud, slow voice" (Sub-6 GHz) to help aim the "fast, fragile voice" (mmWave), we can keep high-speed internet working perfectly, even when you are driving a car at highway speeds. It's like having a GPS that never loses its signal, ensuring your video call never drops, no matter how fast you're going.

1. Problem Statement

Future wireless systems aim to utilize Millimeter-Wave (mmWave) frequencies to support high data rates and low-latency applications. To overcome severe path loss at these frequencies, systems employ fully digital beamforming with large-scale Multiple-Input Multiple-Output (MIMO) arrays.

However, reliable link establishment in mmWave systems is challenging due to:

Low pre-beamforming Signal-to-Noise Ratio (SNR): Making accurate channel estimation difficult.
High Overhead: Traditional channel estimation requires significant pilot resources, which is problematic for latency-critical applications.
Dynamic Environments: Existing research on exploiting out-of-band information (using sub-6 GHz data to assist mmWave estimation) has largely focused on static scenarios. The impact of user mobility (channel aging) and varying pilot configurations in dynamic scenarios remains largely unexplored.

The core problem addressed is how to optimize channel estimation for digital mmWave MIMO systems by leveraging reliable sub-6 GHz information, specifically analyzing the trade-offs between pilot density, channel aging, and propagation conditions (LOS vs. NLOS) in both static and dynamic environments.

2. Methodology

System Model

Architecture: A point-to-point multi-band MIMO system where sub-6 GHz and mmWave bands operate simultaneously. Both use Uniform Linear Arrays (ULAs) with fully digital beamforming (one RF chain per antenna).
Channel Model: A time-varying, frequency-selective Rician fading model is used.
- Deterministic Part: Line-of-Sight (LOS) component modeled as free-space propagation.
- Stochastic Part: Non-Line-of-Sight (NLOS) component modeled using the 3GPP channel model.
- Correlation: The LOS angles (AoD/AoA) are assumed identical across bands, while the stochastic NLOS components are statistically independent due to the large frequency gap.
Pilot Configurations: Three pilot patterns (based on 3GPP CSI-RS concepts) are evaluated:
1. 1-symbol pattern: All pilots in one OFDM symbol (low overhead, low frequency resolution).
2. 2-symbol pattern: Pilots spread over two symbols.
3. 4-symbol pattern: Pilots spread over four symbols (high overhead, high frequency resolution).

Channel Estimation Methods

The paper compares two estimation strategies:

Conventional Estimation: Uses only in-band mmWave pilots. The channel is estimated via Least Squares (LS) at pilot subcarriers and linearly interpolated for data subcarriers.
Out-of-Band Aided MRC (OOBA-MRC): A hybrid approach proposed in prior work [12]. It combines:
- The in-band mmWave estimate ( $\hat{H}^{(m)}$ ).
- An out-of-band aided estimate ( $\tilde{H}^{(m)}$ ) derived from the sub-6 GHz channel state, specifically exploiting the LOS component.
- Combining: The two estimates are weighted using an optimal combining factor ( $b_w$ ) that adapts to the Rician K-factor and noise variance.

Performance Metric

The primary metric is Spectral Efficiency (SE) in bits/s/Hz, calculated based on the achievable rate using Singular Value Decomposition (SVD) for precoding/combining and water-filling for power allocation.

3. Key Contributions

Dynamic Scenario Analysis: Unlike previous studies limited to static environments, this work evaluates the performance of out-of-band aided estimation under time-varying channel conditions (varying receiver velocities).
Pilot Configuration Trade-offs: It systematically analyzes how different pilot densities (1, 2, and 4 symbols) impact SE in both static and dynamic settings, highlighting the "channel aging" effect.
Propagation Condition Impact: The study distinguishes between LOS (high K-factor) and NLOS (low K-factor) scenarios, demonstrating that the efficacy of out-of-band assistance is heavily dependent on the dominance of the LOS path.
Comprehensive Simulation: The evaluation covers a wide range of SNRs, receiver velocities, and Rician K-factors, providing a holistic view of system behavior.

4. Simulation Results

Static Scenarios

Low K-factor (NLOS): Increasing the number of pilot symbols improves SE because the channel is frequency-selective, and more pilots provide better interpolation accuracy.
High K-factor (LOS): The channel is dominated by the LOS path. Increasing pilot symbols yields diminishing returns as the channel becomes effectively flat.
OOBA-MRC Performance: In static scenarios, OOBA-MRC consistently matches or outperforms the conventional method. At 0 dB SNR, it offers significant gains (e.g., ~87% gain in specific NLOS cases) over conventional estimation.

Dynamic Scenarios (Mobility)

The Channel Aging Trade-off: As receiver velocity increases, the channel changes during the pilot transmission interval.
- Conventional Method: Suffers from channel aging. Using more pilot symbols (e.g., 4-symbol pattern) spreads the estimation over a longer time, causing the estimates to become outdated. Consequently, SE degrades significantly at high velocities (e.g., 200 km/h) with dense pilot patterns.
- OOBA-MRC Method: Demonstrates robustness to mobility. Because it relies heavily on the stable LOS component derived from the sub-6 GHz band, its performance is largely independent of the number of pilot symbols.
NLOS (Low K-factor) at High Speed: The conventional method benefits from more pilots at low speeds (accuracy dominates) but suffers at high speeds (aging dominates). OOBA-MRC performs similarly to the conventional method here as the LOS aid is weak.
LOS (High K-factor) at High Speed: OOBA-MRC significantly outperforms the conventional method. While the conventional method's SE drops as pilot density increases (due to aging), OOBA-MRC maintains high SE (dropping only from 5.6 to 3.9 bit/s/Hz at 200 km/h) regardless of pilot configuration.

5. Significance and Conclusion

This paper establishes that exploiting out-of-band sub-6 GHz information is a viable and robust strategy for mmWave MIMO channel estimation, particularly in dynamic, Line-of-Sight environments.

Latency vs. Accuracy: The study highlights that in dynamic LOS scenarios, the "channel aging" penalty of dense pilot patterns makes the conventional approach inefficient. OOBA-MRC allows for lower pilot overhead (fewer symbols) while maintaining high spectral efficiency, which is crucial for latency-critical applications.
Design Implications: For future 6G systems deploying mmWave, utilizing sub-6 GHz assistance can mitigate the need for excessive pilot signaling in mobile LOS scenarios, thereby improving overall system spectral efficiency and link reliability.
Limitations: The benefits of OOBA-MRC are most pronounced in LOS conditions; in pure NLOS scenarios with high mobility, the method relies more on in-band estimation, where the trade-off between pilot density and channel aging remains critical.