Geometry-Aware Channel Aging in MIMO Systems: When Is Kronecker Separability Enough?
Listen
Radio episode about this paper
Transcript
Introduction to the show: ident: Electrical Engineering. The day in electrical engineering: signals, systems and the hardware that carries them.
Rosa: We're looking at the paper: "Geometry-Aware Channel Aging in MIMO Systems".
Dev: Here's the gist of the paper. This letter investigates when Kronecker-separable spatial correlation approximates the covariance of multiple-input multiple-output (MIMO) channels and proposes a geometry-aware channel model that links spatial eigenmode coupling and temporal correlation through the same propagation environment.
Rosa: Channel Modeling Framework The authors propose a geometry-aware MIMO channel model where the same propagation environment determines both the Weischelberger spatial coupling matrix and the temporal correlation coefficient.
Dev: This propagation environment follows a 3GPP-inspired clustered channel model.
Rosa: The construction of the Weischelberger coupling matrix is achieved by projecting cluster-level covariance contributions onto the BS and UE eigenmodes.
Title and authors: Rosa: So we're looking at this paper now titled Geometry-Aware Channel Aging in MIMO Systems: When Is Kronecker Separability Enough? It looks like they are trying to figure out exactly when the simple Kronecker separable model for MIMO channel covariance actually works and when you need something more detailed.
Dev: Right, it's about finding the boundary between when that simple product of matrices is a good approximation and when you have to look at the actual coupling between different spatial modes. It’s all tied to how much the environment—the propagation geometry—affects both how those spatial modes couple and how time correlation builds up because of user equipment mobility.
Taro: When they introduce this geometry-aware model, it seems like they want one single propagation description to dictate both the Weischelberger spatial coupling matrix and that temporal correlation coefficient. That’s a big step because usually you treat those things as separate things you calculate independently.
Rosa: Exactly. They use a 3GPP-inspired clustered channel model for this environment, which lets them define the coupling matrix by projecting cluster-level covariance contributions onto the base station and user equipment eigenmodes <ref:2611.08007#pg1,a 3GPP-inspired clustered channel model>. Then they derive a cluster-dependent temporal correlation coefficient based on Doppler shift weighted by power and angle information from both sides of the link.
Dev: It’s interesting how they handle that Doppler spread for the temporal correlation, specifically defining it as the cluster Doppler spread seen by the user equipment, denoted as DS k c. That model suggests that channel aging isn't just about how fast you are moving overall, but specifically where you are relative to those scattering clusters.
Taro: And they show this linkage means that channel aging is determined by the UE motion relative to the scattering geometry rather than just its velocity magnitude alone. That implies the physics of decorrelation is tied directly to spatial arrangement.
Rosa: What’s really interesting for me, and what I'm looking at from a field robotics side, is their analysis of Kronecker separability itself. They show that when the coupling matrix is perfectly separable—meaning the base station and user equipment eigenmodes are completely uncoupled—then the Kronecker approximation holds.
Dev: But they also measure this mismatch, which they call d Omega. And the numerical results show that diffuse propagation actually causes stronger temporal decorrelation and larger signal error degradation compared to when you have concentrated scattering. That’s a key finding for system design because it tells you where the simple model breaks down most severely.
Taro: So, if we look at performance drops, diffuse propagation is worse than concentrated scattering in terms of temporal decorrelation and error degradation, which is counter-intuitive sometimes when you think about signal strength alone.
Rosa: That’s right. Concentrated scattering creates smaller Doppler spreads and makes the temporal correlation coefficient less sensitive to UE mobility. But diffuse propagation induces greater diversity in the Doppler domain, leading to stronger channel aging effects at high user equipment velocities.
Dev: This really reinforces their point about how geometry dictates the physics of aging, not just speed. They also identify specific regimes where Kronecker separability is a good fit and where you definitely need that geometry-aware coupling matrix.
Taro: From an autonomy research standpoint, knowing these regimes helps us decide which model to run in our simulations for mobility-sensitive MIMO systems or ISAC setups. It’s about picking the right tool for the job based on how spread out those scattering clusters are.
Rosa: And they even evaluate symbol-level aging impact in a Time Division Duplex MIMO system where they explicitly use uplink and downlink pilots, which gives them a good way to measure the spectral efficiencies in that imperfect channel state information scenario.
Dev: It's valuable that they look at those specific pilots because it grounds the analysis in how we actually transmit data, not just some abstract theoretical covariance calculation. The overall implication is that for mobility-sensitive systems, you need models that respect the geometry to accurately predict performance drops caused by channel aging.
Taro: I agree. If you only use velocity magnitude, you miss this crucial dependency on the scattering environment layout. That’s what makes this paper relevant for understanding real-world mobility challenges in wireless communication.
Rosa: So, to wrap up the implications of Geometry-Aware Channel Aging in MIMO Systems: When Is Kronecker Separability Enough?, it seems like we need to move away from just checking if Kronecker separability holds and instead use the geometry to tell us when that approximation is valid.
Dev: The paper shows that diffuse propagation causes worse performance than concentrated scattering because it generates more temporal decorrelation, and they provide a concrete way to model this using the coupling matrix based on cluster contributions.
Taro: It’s about linking spatial eigenmode coupling directly to temporal correlation through the same propagation environment, which gives us a much richer picture of channel aging dynamics under mobility.
Rosa: So, for anyone designing MIMO systems that care about user equipment movement or ISAC, this paper suggests using a geometry-aware model instead of just relying on simple Kronecker separability when you suspect diffuse scattering might be present.
The paper's summary: Rosa: So we're looking at this paper now titled "Geometry-Aware Channel Aging in MIMO Systems: When Is Kronecker Separability Enough?" and it’s basically figuring out when that simple Kronecker separable model for MIMO channel covariance actually works and when you need something more detailed.
Dev: Right, they’re trying to find the boundary between when that simple product of matrices is a good approximation and when you have to look at the actual coupling between different spatial modes. It’s all tied to how much the environment—the propagation geometry—affects both how those spatial modes couple and how time correlation builds up because of user equipment mobility.
Rosa: When they introduce this geometry-aware model, it seems like one single propagation description dictates both the Weischelberger spatial coupling matrix and that temporal correlation coefficient. That’s a big step because usually you treat those things as separate things you calculate independently.
Dev: Exactly. They use a 3GPP-inspired clustered channel model for this environment, which lets them define the coupling matrix by projecting cluster-level covariance contributions onto the base station and user equipment eigenmodes <ref:2611.08007#pg1,a 3GPP-inspired clustered channel model>. Then they derive a cluster-dependent temporal correlation coefficient based on Doppler shift weighted by power and angle information from both sides of the link.
Rosa: It’s interesting how they handle that Doppler spread for the temporal correlation, specifically defining it as the cluster Doppler spread seen by the user equipment, denoted as DS k c. That model suggests that channel aging isn't just about how fast you are moving overall, but specifically where you are relative to those scattering clusters.
Dev: And they show this linkage means that channel aging is determined by the UE motion relative to the scattering geometry rather than just its velocity magnitude alone. That implies the physics of decorrelation is tied directly to spatial arrangement.
Rosa: What’s really interesting for me, and what I'm looking at from a field robotics side, is their analysis of Kronecker separability itself. They show that when the coupling matrix is perfectly separable—meaning the base station and user equipment eigenmodes are completely uncoupled—then the Kronecker approximation holds.
Dev: But they also measure this mismatch, which they call d Omega. And the numerical results show that diffuse propagation actually causes stronger temporal decorrelation and larger signal error degradation compared to when you have concentrated scattering. That’s a key finding for system design because it tells you where the simple model breaks down most severely.
Rosa: That’s right. Concentrated scattering creates smaller Doppler spreads and makes the temporal correlation coefficient less sensitive to UE mobility. But diffuse propagation induces greater diversity in the Doppler domain, leading to stronger channel aging effects at high user equipment velocities.
Dev: This really reinforces their point about how geometry dictates the physics of aging, not just speed. They also identify specific regimes where Kronecker separability is a good fit and where you definitely need that geometry-aware coupling matrix.
Rosa: So, for anyone designing MIMO systems that care about user equipment movement or ISAC, this paper suggests using a geometry-aware model instead of just relying on simple Kronecker separability when you suspect diffuse scattering might be present.
Dev: That's the core idea—not just checking if the Kronecker approximation holds, but using the geometry to tell us exactly when that approximation is valid. It gives a much richer picture of channel aging dynamics under mobility because it links the spatial coupling directly to temporal correlation through one environment.
Rosa: And they show that in highly concentrated regimes, Kronecker separability is adequate, but it misses relevant eigenmode coupling in diffuse ones. That means if you have a lot of scattered energy spread out randomly, the simple product model gives you a bad picture of how the channel actually degrades over time as the user moves.
Dev: They provide a concrete way to model this using cluster contributions for both things—the spatial coupling and that temporal correlation coefficient—which is important because it’s not just some abstract correction; it’s grounded in how clusters are positioned relative to you.
Rosa: So, what this means practically is that if you're designing an autonomous system or a massive MIMO setup, you need to know the scattering geometry first. Then, you can decide whether the simple Kronecker model is enough for your loop rate or if you have to run the more complex Weischelberger coupling matrix model to predict how fast your signal quality will drop when things get diffuse.
The paper's improvements: Taro: So we've seen how they tie the propagation environment to both spatial coupling and temporal correlation, but now we’re looking at what they suggest changing about this model itself to make it better.
Rosa: They propose identifying clear regimes where you can use that simple Kronecker separability approximation versus when you absolutely need that more detailed geometry-aware Weischelberger coupling matrix. It’s not a one-size-fits-all approach for modeling channel aging, which is good because it lets us pick the right math for the right propagation condition.
Dev: That distinction between regimes is important because it changes how we calculate spectral efficiency and how we model those performance drops in real time loops. If we can identify when to switch models, our latency stays lower and our safety case becomes much tighter.
Taro: They also suggest a way to predict the actual magnitude of the channel aging effect based on the geometry itself, instead of just relying on your current velocity measurement. This means you can anticipate performance drops by looking at where those scattering clusters are positioned in relation to your equipment’s movement.
Rosa: That prediction is tied to defining that cluster Doppler shift seen by the user equipment, which involves a specific formula based on the angle between the UE's velocity direction and the cluster directions. It moves us away from just looking at speed magnitude as you might think.
Dev: I see that linkage directly impacting how we calculate channel aging rate, which is huge for massive MIMO systems where mobility is constant. The safety case relies on knowing if that rate of change stays within acceptable bounds under different scattering conditions.
Taro: If the environment gets diffuse, they show it induces greater temporal decorrelation and larger signal error degradation compared to concentrated scattering. That’s a warning sign for high-mobility scenarios where you might think a strong link is going to hold up.
Rosa: So, the main improvement here is shifting from just saying "this model works" to being able to actively choose the best model based on your real-world propagation conditions and then predicting exactly how fast things will degrade.
Dev: It’s about building a more robust system that can gracefully handle unexpected environmental shifts by knowing which mathematical tool to deploy at any given moment. That level of adaptability is what makes this work for real-time control systems.
Taro: This moves the research from just describing the channel to actually predicting how it will behave under movement, which is critical for autonomy where failure modes need anticipating.
Rosa: So, they're giving us a better framework for designing these systems by showing us precisely when our simple assumptions break down and how we can use geometry to predict those failures.
Conclusion: Rosa: So we’re wrapping up this look at "Geometry-Aware Channel Aging in MIMO Systems: When Is Kronecker Separability Enough?". Basically, they’ve shown how to use the physical environment to decide if you need a simple model or a complex one for tracking channel changes.
Dev: They establish that diffuse propagation causes worse performance than concentrated scattering because it generates more temporal decorrelation and larger signal error degradation, which is a major finding for system design.
Taro: It really highlights that speed alone isn't the only thing dictating how fast a channel ages; the physical arrangement matters just as much.
Rosa: They also give us that way to predict aging magnitude based on geometry—specifically by calculating that cluster Doppler shift seen by the user equipment, which uses the angle between velocity and cluster directions.
Dev: That means we can anticipate performance drops in high-mobility scenarios more accurately than if we just used a single velocity-based model for channel aging. The safety case gets much stronger when you can predict those failures.
Taro: I think that ability to predict the effect of diffuse scattering by looking at cluster positions is what makes this useful for designing autonomous systems where the world misbehaves in complex ways.
Rosa: For anyone building a massive MIMO or ISAC system, this paper suggests using that geometry-aware approach instead of just checking if the simple Kronecker separability approximation holds true.
Dev: It’s about making sure we use the right mathematical tool based on whether we are in a concentrated or diffuse regime to keep those real-time loops stable and efficient.
Taro: It gives us a better way to think about channel aging that respects the physical layout of the scattering clusters, which is crucial for testing how systems recover when things get messy.
Rosa: So, this paper shows us how to move beyond just checking separability and start using geometry to truly understand and predict channel performance under movement.
Dev: It sets a clear direction for future work in mobility-sensitive MIMO research by providing that concrete model linking spatial coupling directly to temporal correlation through the propagation environment.
Taro: Next up, we have some papers on sensing-assisted OFDM with index modulation, so we'll look at how those new waveforms handle localization and communication together.
Steven Rivetti, Gabor Fodor, Emil Bjornson, Mikael Skoglund
KTH Royal Institute of Technology · Ericsson Research
eess.SP
Submitted: 2026-11-05
Updated: 2026-11-05
Comments: 5 pages, 3 figures. Submitted to IEEE Wireless Communication Letters
License: http://creativecommons.org/licenses/by-sa/4.0/
Importance score: 83/100
The gist: This letter investigates when Kronecker-separable spatial correlation approximates the covariance of multiple-input multiple-output (MIMO) channels and proposes a geometry-aware channel model that…
Key concepts
- Geometry-Aware Channel Model
- This model uses a clustered channel approach where the same physical environment dictates both the spatial coupling matrix (how signals mix in space) and the temporal correlation coefficient (how signals change over time). It projects cluster contributions onto BS and UE eigenmodes to build these matrices.
- Kronecker Separability
- This is a simplified model where the MIMO channel covariance is treated as a simple product of two separate matrices. It assumes the BS and UE eigenmodes are perfectly uncoupled. The paper tests when this simple approximation is accurate compared to more complex, jointly-correlated models.
- Cluster Doppler Spread ($\varsigma D_{k,c}$)
- This measures the uncertainty in the Doppler domain for a specific scattering cluster seen by a user. It is calculated based on the propagation geometry—specifically, the UE's position relative to a cluster's center and its velocity. This allows prediction of channel aging based on motion relative to scattering geometry, not just raw velocity.
- Propagation Geometry Definition
- This defines the physical setup by specifying a set of C scattering clusters. Each cluster has a center location ($\mathbf{p}_c$), is viewed from the Base Station at angle $\psi_c$, and from the User at angle $\phi_{k,c}$. The size of each cluster is defined by its effective radius $r_c$.
Terminology
Summary
This letter investigates when Kronecker-separable spatial correlation approximates the covariance of multiple-input multiple-output (MIMO) channels and proposes a geometry-aware channel model that links spatial eigenmode coupling and temporal correlation through the same propagation environment. p. 0 Steven Rivetti, Student Member, IEEE, Gabor Fodor, Fellow, IEEE
Emil Bjornson, ¨ Fellow, IEEE, Mikael Skoglund, Fellow, IEEE
Abstract—This letter investigates when Kronecker-separable
spatial correlation approximates the covariance of multiple-inputmultiple-output (MIMO) channels. We propose a geometry-aware
channel model in which the same propagation environment determines both the Weischelberger spatial coupling matrix and the
temporal correlation coefficient.Where the paper says it
Channel Modeling Framework
The authors propose a geometry-aware MIMO channel model where the same propagation environment determines both the Weischelberger spatial coupling matrix and the temporal correlation coefficient. This propagation environment follows a 3GPP-inspired clustered channel model. p. 0 We propose a geometry-aware
channel model in which the same propagation environment determines both the Weischelberger spatial coupling matrix and the
temporal correlation coefficient. The propagation environment
follows a 3GPP-inspired clustered channel model. The propagation geometry is inspired by the 3GPP-standardized cluster
delay line channel model. The propagation environment is
indeed made of propagation clusters, whose positions are
randomly generated, even though it is worth mentioning that
the same framework could be driven by ray tracing in practical
deployments We construct a Weischelberger coupling matrix
by projecting the cluster-level covariance contributions onto
the BS and UE eigenmodes. The propagation environment is
indeed made of propagation clusters, whose positions are
randomly generated, even though it is worth mentioning that
the same framework could be driven by ray tracing in practical
deployments We construct a Weischelberger coupling matrix
by projecting the cluster-level covariance contributions onto
the BS and UE eigenmodes. A cluster-dependent temporal
correlation coefficient is then derived in which each cluster
Doppler shift is weighted by its power, UE-side angle, and
Doppler spread.Where the paper says it
Where the paper says it
Where the paper says it
Kronecker Separability Analysis
The Kronecker separable model represents the MIMO channel covariance as the product of one-sided base station and UE correlation matrices. The Kronecker approximation is obtained when the coupling matrix is defined as Kronecker separable, where the BS and UE eigenmodes are perfectly uncoupled. p. 1 Weischelberger correlation model
We here take the Weichselberger Rayleigh fading model
[5] as a reference and use it to study when the Kroneckerseparable approximation is accurate. This model introduces
an eigenmode coupling matrix, here denoted by Ωk ∈
R
MBS×MUE
+ , specifying how the channel power is distributed
across BS–UE eigenmode pairs, while the Kronecker model
is obtained as a separable special case, where the BS and
UE eigenmodes are perfectly uncoupled. It should be noted that the coupling matrix Ω
kron
k =
1
MBSMUE
λBS,kλ
⊤
UE,k yields a Kronecker-separable covariance
model. The covariance mismatch between the Kroneckerseparable model and the jointly-correlated one is measured
as
dΩ = max
k
Ωk − Ω
kron
k
2
F
∥Ωk∥
2
F
. Finally, we evaluate the
symbol-level aging impact in a time-division duplexing (TDD)
MIMO system with explicit uplink (UL) and downlink (DL)
pilots, covariance-based channel estimation, and imperfectCSI SE expressions. Numerical results identify regimes where
the Kronecker approximation is accurate and regimes where
geometry-aware coupling is needed, showing that diffuse propagation induces stronger temporal decorrelation and larger SE
losses than concentrated scattering.Where the paper says it
Where the paper says it
Where the paper says it
Propagation Geometry Definition
The propagation environment consists of a set of C scattering clusters, where each cluster is centered at pc and seen by the BS at the angle ψc and at the angle φk,c by UE k. The size of each cluster is described by its effective radius rc. p. 1 The angles of the
rays scattered by each cluster follow a normal distribution with
side-dependent angle spreads, denoted by ςBS,c and ςUE,k,c. Assume that the size of each cluster is described by its
effective radius rc. (6)
The rationale behind this model is that closer clusters scatter
rays within a wider angular interval than farther ones. The
marginal covariances CBS,k and CUE,k are computed according to the model in [8]: under the assumption of small angle
spreads, we can write CBS =
PC
c=1 ρ
BS
c CBS,c, where
[CBS,c]
l,m≈ e
−jπ(l−m) sin(ψc)
e
− π2
2
((l−m) cos(ψc)ςBS,c)
2
. We propose the following coupling matrix
Proposition 2. Let uBS,l and uUE,k,m denote the l-th largest
eigenvector of CBS and the m-th largest eigenvector of CUE,k,
then the geometry-based coupling matrix
[Ωk]
l,m=
X
C
c=1
ρ
J
c,k νc,l υk,c,m (9)
νc,l =
u
H
BS,l CBS,cuBS,l
/MBS (10)
υk,c,m =
u
H
UE,k,m CUE,k,cuUE,k,m
/MUE (11)
This coupling matrix maps the propagation environment to
eigenmode coupling: [Ωk]
l,m is large when both eigenmodes
are excited by one or more clusters with high power.Where the paper says it
Where the paper says it
Where the paper says it
Temporal Correlation Dynamics
The temporal correlation coefficient ζk(i) is modeled using a Doppler-spread model that directly links the uncertainty in the angular domain to that in the Doppler domain. The cluster Doppler spread as seen by UE k, denoted by ςD,k,c, is defined based on propagation geometry. p. 2 Let us consider a cluster seen by UE k at an angle
φk,c with an angle spread of ςUE,k,c. Assuming a Gaussian
intra-cluster angular deviation, the cluster Doppler spread as
seen by UE k, denoted by ςD,k,c, is defined as
ς
2
D,k,c=f
2
D,kh
0. CONCLUSIONS
We proposed a geometry-aware channel model where the
same clustered propagation environment determines both the
Weischelberger eigen-mode coupling and the temporal correlation decay. Channel aging is now determined by the UE motion
relative to the scattering geometry, rather than its velocity magnitude alone.Where the paper says it
Where the paper says it
Conclusion and Findings
The results show that Kronecker separability is an adequate approximation in highly concentrated propagation regimes, but can miss relevant eigenmode coupling in diffuse environments. Diffuse propagation induces stronger Doppler diversity, causing stronger channel aging and larger SE degradation at high UE velocities. p. 4 The results show
that Kronecker separability is an adequate approximation in
highly concentrated propagation regimes, but can miss relevant
eigenmode coupling in diffuse environments. Moreover, diffuse propagation induces stronger Doppler diversity, causing
stronger channel aging and larger SE degradation at high
UE velocities. Moreover, diffuse propagation induces stronger Doppler diversity, causing
stronger channel aging and larger SE degradation at high
UE velocities. This motivates eigenmode-coupled models for
mobility-sensitive MIMO and ISAC systems. To this end, we propose a Doppler-spread
model that directly links the uncertainty in the angular domain
to that in the Doppler domain. Finally, we evaluate the
symbol-level aging impact in a time-division duplexing (TDD)
MIMO system with explicit uplink (UL) and downlink (DL)
pilots, covariance-based channel estimation, and imperfectCSI SE expressions.Where the paper says it
Where the paper says it
Where the paper says it
References
[1] S. Fodor, G. Fodor, D. Gurg ¨ uno ¨ glu, and M. p. 4 REFERENCES
[1] S. Gurg ¨ uno ¨ glu, and M. Gurg ¨ uno ¨ glu, and M. Telek, “Optimizing pilot ˘
spacing in MU-MIMO systems operating over aging channels,” IEEE
Transactions on Communications, vol.Where the paper says it
Where the paper says it
[2] K. T. Truong and R. p. 4 3708–3720, 2023. Truong and R. Truong and R. Heath, “Effects of channel aging in massive
MIMO systems,” Journal of Communications and Networks, vol.Where the paper says it
Where the paper says it
[3] C.-X. Wang and M. p. 4 338–351, 2013. Wang and M. Wang and M. Patzold, “Efficient simulation of multiple crosscorrelated Rayleigh fading channels,” in 14th IEEE Proceedings on
Personal, Indoor and Mobile Radio Communications, 2003. Patzold, “Efficient simulation of multiple crosscorrelated Rayleigh fading channels,” in 14th IEEE Proceedings on
Personal, Indoor and Mobile Radio Communications, 2003. PIMRC
2003. Wang and M. Patzold, “Efficient simulation of multiple crosscorrelated Rayleigh fading channels,” in 14th IEEE Proceedings on
Personal, Indoor and Mobile Radio Communications, 2003.Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
[4] M. Bengtsson and P. p. 4 1526–1530. Bengtsson and P.Where the paper says it
[5] W. Weichselberger, M. p. 1 Lastly, ni,v,k ∼ CN (0, IMUE σ
2
k
) is UE k receiver noise. Weischelberger correlation model
We here take the Weichselberger Rayleigh fading model
[5] as a reference and use it to study when the Kroneckerseparable approximation is accurate. Weichselberger, M. Ozcelik, and E. Ozcelik, and E. Bonek, “A stochastic
mimo channel model with joint correlation of both link ends,” IEEE
Transactions on wireless Communications, vol.Where the paper says it
Where the paper says it
Where the paper says it
[6] Z. Wang, J. Zhang, B. Ai, C. Yuen, and M. p. 4 We obtain the reported curves by
averaging 20 Monte Carlo realizations of the propagation environment; for each environment, we average the SE over 500 channel
and precoder realizations. Yuen, and M. Yuen, and M. Debbah, “Uplink performance of cell-free massive MIMO with multi-antenna users over jointlycorrelated Rayleigh fading channels,” IEEE Transactions on Wireless
Communications, vol.Where the paper says it
Where the paper says it
[7] K. Dovelos, M. p. 4 7391–7406, 2022. Dovelos, M. Dovelos, M. Matthaiou, H. Matthaiou, H. Ngo, and B. Ngo, and B. Bellalta, “Massive mimo
with multi-antenna users under jointly correlated ricean fading,” in ICC
2020-2020 IEEE International Conference on Communications (ICC). Ngo, and B. Bellalta, “Massive mimo
with multi-antenna users under jointly correlated ricean fading,” in ICC
2020-2020 IEEE International Conference on Communications (ICC).Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
[8] O. T. Demir, E. Bj ¨ ornson, and L. p. 4 IEEE, 2020, pp. Bj ¨ ornson, and L. Bj ¨ ornson, and L. Sanguinetti, “Channel modeling and ¨
channel estimation for holographic massive MIMO with planar arrays,”
IEEE Wireless Communications Letters, vol.Where the paper says it
Where the paper says it
[9] M. Series, “Guidelines for evaluation of radio interface technologies for imt-2020,” Report ITU, vol. p. 4 997–1001,
2022. Series, “Guidelines for evaluation of radio interface technologies for
imt-2020,” Report ITU, vol.Where the paper says it
[10] R. Sinkhorn and P. p. 2 , uBS,k,l). A
Sinkhorn-like scaling is applied to Ω [10] to match the row
and column sums to the BS and UE marginal eigenvalues, thus
ensuring a fair comparison with Ω
kron
. Sinkhorn and P. Knopp, “Concerning nonnegative matrices and
doubly stochastic matrices,” Pacific Journal of Mathematics, vol.Where the paper says it
Where the paper says it
[11] S. Rivetti, G. p. 0 Geometry-Aware Channel Aging in MIMO
Systems: When Is Kronecker Separability Enough? Steven Rivetti, Student Member, IEEE, Gabor Fodor, Fellow, IEEE
Emil Bjornson, ¨ Fellow, IEEE, Mikael Skoglund, Fellow, IEEE
Abstract—This letter investigates when Kronecker-separable
spatial correlation approximates the covariance of multiple-inputmultiple-output (MIMO) channels. Geometry-Aware Channel Aging in MIMO
Systems: When Is Kronecker Separability Enough? Steven Rivetti, Student Member, IEEE, Gabor Fodor, Fellow, IEEE
Emil Bjornson, ¨ Fellow, IEEE, Mikael Skoglund, Fellow, IEEE
Abstract—This letter investigates when Kronecker-separable
spatial correlation approximates the covariance of multiple-inputmultiple-output (MIMO) channels. Geometry-Aware Channel Aging in MIMO
Systems: When Is Kronecker Separability Enough? Steven Rivetti, Student Member, IEEE, Gabor Fodor, Fellow, IEEE
Emil Bjornson, ¨ Fellow, IEEE, Mikael Skoglund, Fellow, IEEE
Abstract—This letter investigates when Kronecker-separable
spatial correlation approximates the covariance of multiple-inputmultiple-output (MIMO) channels. Geometry-Aware Channel Aging in MIMO
Systems: When Is Kronecker Separability Enough? Steven Rivetti, Student Member, IEEE, Gabor Fodor, Fellow, IEEE
Emil Bjornson, ¨ Fellow, IEEE, Mikael Skoglund, Fellow, IEEE
Abstract—This letter investigates when Kronecker-separable
spatial correlation approximates the covariance of multiple-inputmultiple-output (MIMO) channels. Bjornson, and M. Skoglund, “On the impact of ¨
channel aging and doppler-affected clutter on OFDM ISAC systems,”
arXiv preprint arXiv:2601.Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
[12] E. B. Kama, J. Kim, and E. p. 4 05032, 2026. Kim, and E. Kim, and E. Bjornson, “Downlink pilots are essential ¨
for cell-free massive MIMO with multi-antenna users,” in 2024 IEEE
Wireless Communications and Networking Conference (WCNC). Kim, and E. Bjornson, “Downlink pilots are essential ¨
for cell-free massive MIMO with multi-antenna users,” in 2024 IEEE
Wireless Communications and Networking Conference (WCNC).Where the paper says it
Where the paper says it
Where the paper says it
[13] S. Rivetti, G. Fodor, E. Bjornson, and M. Skoglund, “Cram’er-rao bound analysis of bistatic isac under partial symbol knowledge and clutter,” arXiv preprint arXiv:2604. p. 4 Bjornson, and M. Skoglund, “Cram\’er-rao
bound analysis of bistatic isac under partial symbol knowledge and
clutter,” arXiv preprint arXiv:2604.Where the paper says it
[14] O. T. Demir, E. Bj ¨ ornson, L. p. 4 IEEE, 2020, pp. Bj ¨ ornson, and L. Bj ¨ ornson, and L. Sanguinetti, “Channel modeling and ¨
channel estimation for holographic massive MIMO with planar arrays,”
IEEE Wireless Communications Letters, vol. Sanguinetti ¨ et al. , “Foundations of usercentric cell-free massive MIMO,” Foundations and Trends® in Signal
Processing, vol.Where the paper says it
Where the paper says it
Where the paper says it
[15] European Telecommunications Standards Institute (ETSI), “Study on channel model for frequencies from 0.5 to 100 GHz: 3GPP TR 38. p. 4 [15] European Telecommunications Standards Institute (ETSI), “Study on
channel model for frequencies from 0. 5 to 100 GHz: 3GPP TR 38. 5 to 100 GHz: 3GPP TR 38. 901
version 15.Where the paper says it
Where the paper says it
[26] This paper is arXiv:2611.08007v1 [eess. p. 0 The BS performs orthogonal frequency division multiplexing
arXiv:2611. 08007v1 [eess.Where the paper says it
Page 3
Fig. 1: Representation of the pilot and data symbol transmission structure in the time-frequency domain
Page 4
Fig. 2 shows the empirical Cumulative distribution function (CDF) of the covariance mismatch between the Kronecker separable model and the Jointly correlated one
Page 5
Fig. 3 shows the average UL and DL SEs evolution for increasing UE mobility in diffusive (i. p. 4 We now move to investigating the impact
of a Kronecker-separable approximation on the achievable SE:
Fig. 3 shows the average UL and DL SEs evolution for increasing UE mobility in diffusive (i. We schedule both uplink
and downlink pilots, thereby deriving the respective spectral
efficiencies (SEs) in the presence of imperfect channel state
information. Numerical results show that channel aging’s magnitude is strongly geometry-dependent, with diffuse propagation
inducing greater temporal decorrelation and SE degradation
than concentrated scattering.Where the paper says it
Where the paper says it
A concentrated scattering environment is generates higher SEs: concentrated scattering geometries produce smaller Doppler spreads and their ζk(·) is less sensitive to UE mobility. The DL SE in Fig. p. 2 The lemma follows by computing
ς
2
D,k,c = E[f
2
D,k,c,n] − E[fD,k,c,n]
2
III. CHANNEL ESTIMATION AND SPECTRAL EFFICIENCY
Fig. The DL SE in
Fig. 3c is more sensitive to ∥ν∥ than the UL, as the UL SE
benefits from centralized BS-side detection, and shows a lowto-moderate Geo-Kron gap. 3c is more sensitive to ∥ν∥ than the UL, as the UL SE
benefits from centralized BS-side detection, and shows a lowto-moderate Geo-Kron gap. On the contrary, the UL SE is
less sensitive to UE mobility and exhibits a large Geo-Kron
gap. On the contrary, the UL SE is
less sensitive to UE mobility and exhibits a large Geo-Kron
gap. Interestingly, with ZF BS combining, the Geo-Kron gap
becomes smaller in concentrated propagation environments.Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Green — the paper says it, and the page is printed. Red — no passage carries it, so treat that sentence as unverified. The same check runs on your own text: check a summary against its source, $5.
Improvements for AI systems
-
Geometry-aware Channel Model Integration: The improved system can determine both
the Weischelberger spatial coupling matrix and the temporal correlation coefficient
from a single propagation environment, which is inspired by a 3GPP-standardized clustered channel model. This allows for a more accurate representation of channel aging than models relying solely on UE velocity magnitude. -
Kronecker Separability Regime Identification: The system can identify
regimes where Kronecker separability can be applied and regimes where geometry-aware Weischelberger coupling is needed.
This enables the AI to select the appropriate mathematical model for channel aging and spectral efficiency calculations based on the specific propagation conditions, as shown by comparing accuracy against aKron
baseline. -
Geometry-Dependent Aging Prediction: The system can predict channel aging magnitude based on propagation geometry, specifically noting that
diffuse propagation inducing greater temporal decorrelation and SE degradation than concentrated scattering.
This allows the AI to anticipate performance drops in high-mobility scenarios more accurately by analyzing cluster positions and spreads. -
Personalized Channel Aging Rate: Instead of a single velocity-based model, the system can calculate a cluster Doppler shift based on
Doppler shift seen by cluster c is defined as fD,k,c = (f νk /c0) cos(φk,c − θk).
This means the AI can predict how quickly a channel ages based on the specific angular relationship between the UE's velocity direction and the scattering cluster directions.
Abstract
This letter investigates when Kronecker-separable spatial correlation approximates the covariance of multiple-input-multiple-output (MIMO) channels. We propose a geometry-aware channel model in which the same propagation environment determines both the Weischelberger spatial coupling matrix and the temporal correlation coefficient. The propagation environment follows a 3GPP-inspired clustered channel model. We consider a MIMO system affected by symbol-level channel aging caused by the user equipment's mobility. We schedule both uplink and downlink pilots, thereby deriving the respective spectral efficiencies (SEs) in the presence of imperfect channel state information. Numerical results show that channel aging's magnitude is strongly geometry-dependent, with diffuse propagation inducing greater temporal decorrelation and SE degradation than concentrated scattering. The results also identify regimes where Kronecker separability can be applied and regimes where geometry-aware Weischelberger coupling is needed.
Sources
- On the Impact of Channel Aging and Doppler-Affected Clutter on OFDM ISAC Systems
- Cramer-Rao Bound Analysis of Bistatic ISAC Under Partial Symbol Knowledge and Clutter
Related papers
- Open-H-RF: A Large-Scale Ultrasound Raw Channel Dataset Initiative
- Scalable Attention for 5G NR Channel Estimation
- Low-Altitude Wireless Networks: A Comprehensive Survey
- Sensing-Assisted OFDM with Index Modulation: A Waveform for Optical Wireless Integrated Communication and Localisation
- Iteration-Free Candidate Pruning and an Index-Domain VLSI Architecture for Sort-Free MIMO Box Decoding up to 1024-QAM
- A Foundation Model for Large-Scale Wireless Network Planning, Operation and Optimization