Constrained Tensor Decomposition-Based Target Sensing for Sparse Non-Uniform Array-Enabled AFDM ISAC Systems
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: "Constrained Tensor Decomposition-Based Target Sensing for Sparse Non-Uniform Array-Enabled AFDM ISAC Systems".
Dev: Here's the gist of the paper. Sparse non-uniform array-enabled affine frequency division multiplexing (AFDM) is a promising candidate for integrated sensing and communication (ISAC), while its performance critically depends on accurate target parameter estimation The proposed algorithm nearly attains Cramer-Rao bound ´ (CRB) and significantly outperforms unconstrained ALS and conventional methods, validating its effectiveness for sparse nonuniform array-enabled AFDM ISAC systems System Modeling The system considers a sparse non-uniform array-enabled AFDM ISAC system that performs sensing of R surrounding targets while simultaneously supporting communication with several users in the vicinity of the base station The received signal is formulated as a high-order tensor, and subsequent analysis involves extracting factor matrices from this tensor to characterize physical parameters The received signal can be further represented as a tensor Y ∈ C Nr×N×Nt = Z + W, where W denotes the corresponding noise in the DAF domain Constrained Tensor Decomposition A manifold-constrained alternating least squares (ALS) algorithm is developed by exploiting the sparse array geometry structure, enabling robust factor matrix extraction and direct angle estimation The problem of factor matrices extraction for tensor signals is formulated as min aˆr,bˆr,cˆr Y −X R r=1 aˆr ◦ bˆ r ◦ cˆr F To overcome the drawbacks of standard ALS, the constrained alternative least square (CALS) algorithm is introduced by adding constraints Aˆ ∈ Γ1 and Cˆ ∈ Γ2 to the optimization problem The optimization of this constrained problem is achieved via the ALS framework within Niter iterations until convergence Parameter Estimation Refinement After recovering the factor matrices, the extraction of sensing parameters involves a two-stage process First, coarse estimation of the integer delay is performed using a scheme analogous to matched filtering based on pulse compression principles Second, an iterative one dimensional golden section search method is employed to refine the normalized delay and Doppler shift parameters The objective function used in the golden section search is written as f = ΛgΠlS H yr Performance Validation Simulation results demonstrate that the proposed algorithm nearly achieved the CRB and consistently outperformed baseline algorithms The algorithm achieves the best estimation performance for angular parameters and approaches the CRB, exhibiting a more pronounced advantage particularly in the low signalto-noise ratio regime Both tensor decomposition based estimation algorithms outperform the conventional MUSIC algorithm because MUSIC fails to exploit the high-dimensional structural information inherent in the received signal The computational complexity of the proposed two-stage algorithm is given by O(Niter(3NtNr+R2 (NtN + NrN + NtNr) + RTiter(Nt + Nr))) The dominant term in the complexity of the parameter estimation stage is O(T RN log N) In this paper, a constrained tensor decomposition-based sensing framework was developed for delay, Doppler, and angle estimation in sparse non-uniform array-enabled AFDM ISAC systems The framework establishes a tensor model by jointly characterizing the time, frequency, and spatial domains The proposed method successfully incorporates the sparse array geometry into the tensor decomposition to enable robust factor matrix extraction and direct angle estimation The delay and Doppler shifts are subsequently refined through an iterative one dimensional golden section search using the decomposed factor matrices The simulation results confirm that the proposed algorithm nearly achieved the CRB and consistently outperformed baseline algorithms References
1: Z.
Rosa: Wei, H.
Dev: Qu, Y.
Rosa: Wang, X.
Title and authors: Rosa: So we're looking at this paper called "Constrained Tensor Decomposition-Based Target Sensing for Sparse Non-Uniform Array-Enabled AFDM ISAC Systems" and the authors are Linchu Chen, Zhendong Li, Lin Chen, Zhou Su, Wenjie Wang and Wen Chen. Rosa is a field roboticist who runs trials outside the lab so I'm curious if this framework works outside of a perfect lab setting for sensing targets.
Dev: That's fair question Rosa. The title itself tells us this paper is focused on how to sense targets using sparse, non-uniform arrays with affine frequency division multiplexing or AFDM, and it mentions that accurate target parameter estimation is the critical part here.
Rosa: Yeah, so what they are proposing is this constrained tensor decomposition based sensing framework for estimating delay and Doppler shifts and angles in these systems. It sounds like they're trying to get a better handle on the target parameters when you have those complex non-uniform arrays involved.
Taro: From my side, I'm interested in how they handle the actual structure of the array geometry because that’s where things often break down when you move from simulation to real-world scenarios.
Dev: Exactly Taro. The paper suggests they develop a manifold constrained alternating least squares algorithm specifically by exploiting that sparse array geometry structure to get robust factor matrix extraction and direct angle estimation. That seems like a big step because standard methods usually struggle with irregular sampling patterns, right?
Rosa: It does seem like it is. They formulate the received signal as a high-order tensor, which is just this complex three-dimensional thing Y equals Z plus W where W is the noise in the DAF domain. Then they use this constrained algorithm to extract delay, Doppler and angular parameters independently from those different factor matrices.
Taro: So instead of just having one big estimation problem that gets messy, they are separating the estimation process based on the dimensions of these factor matrices, which is a smart way to tackle coupled parameters like delay and Doppler shift.
Dev: Right. And after they get those factor matrices, they don't stop there; they use an iterative one dimensional golden section search method to refine the normalized delay and Doppler shift parameters using an objective function that looks like f equals the squared norm of y r.
Rosa: That sounds like a pretty robust way to fine-tune those specific shifts once you have the main structure from the decomposition part. But how does this approach compare when we think about real-world deployment, like when the signal quality is really poor?
Taro: The paper shows that simulation results demonstrate that this proposed algorithm nearly attains the Cramer-Rao bound or CRB, which means it's achieving the best possible estimation performance for angular parameters and it does this particularly well in low signal to noise ratio regimes.
Title and authors: Dev: That's significant because when you're dealing with low signal to noise ratio, conventional methods like unconstrained alternating least squares or conventional MUSIC often fall apart, but this tensor decomposition approach outperforms them. It validates the method for sparse non-uniform array-enabled AFDM ISAC systems.
Rosa: So it sounds like the main thing they've done is use the manifold knowledge of those sparse non-uniform arrays within the alternating least squares algorithm to get that robust factor matrix extraction and direct angle estimation, which is tough when you don't have a regular grid.
Taro: And what they are showing us with this work is that by incorporating that manifold knowledge into the ALS framework, they can achieve robust factor matrix extraction and direct angle estimation even when existing tensor models aren't applicable to irregular spatial sampling.
Dev: That addresses a key weakness in applying these types of methods to real hardware where the array isn't perfectly uniform, which is why I think it’s important for the controls engineers like myself to see this kind of structural exploitation.
Rosa: And then they take those decomposed factor matrices and use the golden section search for refining delay and Doppler shift, which gives them a clean way to determine the integer components of normalized delay and Doppler shift information unambiguously from that process.
Taro: It's interesting how they combine the tensor decomposition for structure with this iterative refinement stage, because it allows them to extract these coupled parameters in a structured manner before doing the final fine-tuning.
Dev: The computational complexity they state is O of Niter times three Nt n r plus R squared times Nt n plus R N n plus Nt n r plus RTiter times nt plus nr, and the dominant term in that estimation stage is O of T R N log N. That gives us a sense of how much real-time processing power this thing will need to handle in practice.
Rosa: So yeah, it sounds like they’ve put together a framework that jointly characterizes the time, frequency, and spatial domains to enable delay, Doppler, and angle estimation robustly. This whole approach validates its effectiveness for sparse non-uniform array-enabled AFDM ISAC systems because it nearly attains the CRB in simulations.
Taro: I'm just thinking about what this means for future work; they focused on delay, Doppler and angle estimation using this framework, so the next step would be testing how gracefully this system recovers when the world misbehaves or when you have unexpected noise patterns.
Dev: That makes sense. The paper itself flags that it's a framework for sensing parameters in sparse non-uniform array-enabled AFDM ISAC systems, and the authors validated its effectiveness through simulation results against unconstrained ALS and conventional methods.
Rosa: So to wrap up this section, this constrained tensor decomposition based sensing framework is about joint characterization of time, frequency, and spatial domains for delay Doppler angle estimation in these systems. It successfully incorporates the sparse array geometry into the tensor decomposition to enable robust factor matrix extraction and direct angle estimation.
Title and authors: Taro: I think what’s really important is that they've shown how to get high-resolution target parameter estimation by nearly attaining the CRB and significantly outperforming unconstrained ALS and conventional methods, especially in low signal to noise ratio regimes.
Dev: And for me as someone focused on real-time loops, the fact that they have a defined complexity O of Niter times three Nt n r plus R squared times Nt n plus R N n plus Nt n r plus RTiter times nt plus nr is something I need to keep in mind when designing the actual hardware implementation.
Rosa: So we've seen this paper, "Constrained Tensor Decomposition-Based Target Sensing for Sparse Non-Uniform Array-Enabled AFDM ISAC Systems", and it gives us a solid constrained tensor decomposition framework for delay, Doppler and angle estimation in sparse non-uniform array enabled AFDM ISAC systems.
Taro: I just want to emphasize that the constrained tensor decomposition approach avoids the loss of structural information caused by dimensionality flattening when processing those high-dimensional received signals, leading to more precise parameter estimation.
Dev: And by exploiting the sparse array geometry structure, they enable the extraction of delay, Doppler shift and angle parameters independently from the factor matrices of their respective dimensions, which results in a robust method for estimating these coupled parameters.
Rosa: By incorporating that manifold knowledge into the alternating least squares algorithm, the system can achieve robust factor matrix extraction and direct angle estimation even when existing tensor models are not applicable to irregular spatial sampling.
Taro: And they refine delay and Doppler shifts through an iterative one dimensional golden section search method by optimizing an objective function f equals the squared norm of y r using the decomposed factor matrices, allowing for the determination of the integer components of normalized delay and Doppler shift information can be obtained unambiguously from that process.
Dev: The simulation results confirm that this proposed algorithm nearly achieved the Cramer-Rao bound and consistently outperformed baseline algorithms, exhibiting a more pronounced advantage particularly in the low signal to noise ratio regime.
Rosa: So in conclusion, this paper presents a constrained tensor decomposition based sensing framework for delay Doppler and angle estimation in sparse non-uniform array enabled AFDM ISAC systems. It establishes a tensor model by jointly characterizing the time frequency and spatial domains, successfully incorporating the sparse array geometry into the tensor decomposition to enable robust factor matrix extraction and direct angle estimation.
Taro: This entire process ultimately allows for high-resolution target parameter estimation in sparse non-uniform array enabled AFDM ISAC systems by nearly attaining the Cramer-Rao bound and significantly outperforming unconstrained ALS and conventional methods.
Dev: It's a powerful tool because it’s validating effectiveness for sparse nonuniform array enabled AFDM ISAC systems, and we have to keep an eye on how this translates to real hardware.
Rosa: That's all for this paper. Next up, we'll look at some other papers on arXiv that are really pushing the boundaries in wireless communication and sensing.
The paper's summary: Rosa: So, this paper is about building a framework to sense targets using sparse, non-uniform arrays combined with affine frequency division multiplexing, or AFDM.
Dev: Right. It tackles how to get accurate estimates for delay and Doppler shifts and angles in those kinds of systems where the antenna spacing isn't regular.
Rosa: The authors developed a constrained tensor decomposition method to handle the high-dimensional signals from these arrays. They use this structure to extract the factor matrices that contain all that physical information about the targets.
Dev: And they don't just stop at extracting those matrices; they use them in a two-stage process, starting with a coarse estimate for delay based on pulse compression principles, and then refining the normalized delay and Doppler shifts using an iterative one dimensional golden section search method.
Rosa: The results show this works really well. They nearly hit the Cramer-Rao bound, which means it’s getting as accurate as possible for angular parameters. And it actually outperforms standard methods like unconstrained alternating least squares, especially when the signal quality is low, which is pretty important for real deployment.
Dev: It's significant because conventional MUSIC algorithms struggle with irregular array structures because they don't exploit all the structural information in the received signals. This tensor approach seems to do that by jointly characterizing time, frequency, and spatial domains.
Rosa: So what this means for us is that we can get high-resolution target parameter estimation in these complex systems without needing perfect uniformity in our array geometry. It’s a way to make sensing robust even when the world misbehaves or the signal conditions are messy.
Dev: I agree with that part about robustness. But we have to look at the complexity too, which is stated as being proportional to Niter times three Nt n r plus R squared times Nt n plus R N n plus Nt n r and Titer times nt plus nr. That tells you how much processing power this thing will need when you put it into a real-time loop.
Rosa: Exactly. So, it’s a framework that successfully incorporates the geometry of sparse non-uniform arrays into the decomposition to get robust factor matrix extraction and direct angle estimation. It sets a high bar for how we can estimate those parameters in these systems.
Dev: And that high bar is really important because if we can do this reliably, it opens up possibilities for integrated sensing and communication systems that need to work together over shared resources.
Rosa: So the next thing we're looking at is seeing how this framework handles real-world failures, like unexpected noise patterns or when the array itself isn't perfectly placed in a lab setting.
The paper's improvements: Tom: So, we're looking at what else these authors suggest for improving this constrained tensor decomposition method for sensing targets.
Rosa: They focus on how to make the extraction of delay, Doppler shift, and angle parameters even more independent by leveraging the structure of the factor matrices in each dimension. It’s about getting those coupled parameters separated cleanly.
Dev: That makes sense because coupling between delay and Doppler shifts can make estimation really tricky for real-time systems. So they're trying to make that extraction process more robust against noise and structural imperfections in the array setup.
Rosa: They also emphasize the role of using that manifold knowledge of the sparse non-uniform array geometry directly inside the alternating least squares algorithm. This means even when existing tensor models don't fit a perfectly irregular spatial sampling pattern, they can still get robust factor matrix extraction.
Dev: I see what you mean. So it’s about making sure that if we move this out of a clean simulation and into something more rugged, the core estimation part doesn't completely break down because of those geometric constraints.
Rosa: Exactly. And they also detail how that iterative golden section search refines the delay and Doppler shifts by optimizing that objective function, which is basically just making sure we get those integer components of the normalized shift information as clearly as possible from the decomposed matrices.
Dev: That refinement step seems like a solid addition to make sure the final parameter estimates are high quality, not just good enough. It’s about precision on top of accuracy.
Rosa: And they also mention that their complexity analysis shows that while it’s computationally intensive, it stays within a manageable bound given the hardware constraints we discussed earlier. The dominant term in the estimation stage is O of T R N log N.
Dev: That scaling is important for my work on real-time loops; I need to know if that O of T R N log N term will actually keep up with our required loop rate when we’re dealing with high-order tensors.
Rosa: So the implication here is that they’ve created a method that is theoretically very precise, nearly hitting the Cramer-Rao bound, while also being structured enough to handle the physical reality of sparse non-uniform arrays. It validates this specific tensor model for integrated sensing and communication systems.
Dev: It means we have a more reliable tool for target parameter estimation in these complex environments than we had with simpler methods like conventional MUSIC or unconstrained ALS.
Rosa: So next, we need to see how this framework performs when the world gets really messy, like when the signal-to-noise ratio drops way down.
Conclusion: Rosa: So, to wrap up this paper on "Constrained Tensor Decomposition-Based Target Sensing for Sparse Non-Uniform Array-Enabled AFDM ISAC Systems," it’s about a framework that uses tensor decomposition to get high-resolution delay, Doppler, and angle estimates in those complex sparse array systems.
Dev: Right. The main point is that by incorporating the geometry of the sparse non-uniform arrays into the decomposition, they achieve robust factor matrix extraction and direct angle estimation even when standard tensor models don't fit perfectly.
Rosa: It really shows how you can jointly characterize time, frequency, and spatial domains to get a solid picture of what’s happening with those targets. It’s a big step for making sensing work reliably in non-ideal settings where the array isn't perfectly uniform.
Dev: The simulation results confirm that this approach nearly achieves the Cramer-Rao bound, which means it hits the best possible performance, especially when the signal to noise ratio is low. That level of accuracy is what we need for reliable real-time sensing applications.
Taro: I just think it’s interesting how they handle those failure modes when things get messy; they show that this framework can still deliver high-resolution estimation even under challenging conditions, which gives me confidence about its real-world applicability.
Rosa: It does, and it really proves that we can build these systems to perform well outside of a perfect lab setting for sensing targets. It validates the entire approach for sparse non-uniform array enabled AFDM ISAC systems.
Dev: For my work on real-time loops, I just need to keep an eye on that complexity, which is O of Niter times three Nt n r plus R squared times Nt n plus R N n plus Nt n r and Titer times nt plus nr. That’s the number we have to manage when we put this into actual hardware.
Taro: I’m curious if they explore what happens when the noise patterns get truly unexpected, like sudden interference or severe signal fading; that’s where you really test how gracefully it recovers from those real-world messes.
Rosa: Well, for now, this paper lays a solid foundation by establishing this constrained tensor decomposition framework for delay Doppler and angle estimation in those sparse non-uniform array enabled AFDM ISAC systems. It sets a high bar for how we can estimate target parameters in these systems.
Dev: That's the summary of the core contribution. It’s a powerful tool because it validates effectiveness for sparse nonuniform array enabled AFDM ISAC systems, and we have to keep an eye on how this translates to real hardware latency.
Taro: I think what’s really important is that they've shown how to get high-resolution target parameter estimation by nearly attaining the Cramer-Rao bound and significantly outperforming unconstrained ALS and conventional methods.
Rosa: So in conclusion, this paper presents a constrained tensor decomposition based sensing framework for delay Doppler and angle estimation in sparse non-uniform array enabled AFDM ISAC systems. It establishes a tensor model by jointly characterizing the time frequency and spatial domains, successfully incorporating the sparse array geometry into the tensor decomposition to enable robust factor matrix extraction and direct angle estimation.
Dev: It’s a powerful tool because it’s validating effectiveness for sparse nonuniform array enabled AFDM ISAC systems, and we have to keep an eye on how this translates to real hardware.
Taro: I think the next thing we need to see is testing how this system recovers when the world misbehaves or when you have unexpected noise patterns.
Rosa: That’s all for this paper on "Constrained Tensor Decomposition-Based Target Sensing for Sparse Non-Uniform Array-Enabled AFDM ISAC Systems". Next up, we'll look at some other papers on arXiv that are really pushing the boundaries in wireless communication and sensing.
School of Information and Communication Engineering, Xi’an Jiaotong University · Department of Electrical and Computer Engineering, Stevens Institute of Technology Department of Electronic Engineering, Shanghai Jiao Tong University
eess.SP
Submitted: 2026-07-20
Updated: 2026-11-05
Comments: 5 pages,3 figures
License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
Importance score: 91/100
The gist: Sparse non-uniform array-enabled affine frequency division multiplexing (AFDM) is a promising candidate for integrated sensing and communication (ISAC)…
Key concepts
- Sparse Non-Uniform Array
- This refers to an antenna setup where the physical arrangement of sensors is not a simple, regular grid. This irregularity is exploited in the method because it allows for more accurate characterization of the physical parameters of targets being sensed, which is crucial for high-resolution estimation.
- Tensor Decomposition
- The received signal from the array is modeled as a high-order tensor. Tensor decomposition breaks this complex signal into simpler factor matrices that represent different physical aspects like time, frequency, and spatial information. This allows the system to extract detailed characteristics of the sensed targets.
- Constrained Alternating Least Squares (CALS)
- This is an optimization technique used to find the best factor matrices for the tensor decomposition. It adds specific constraints based on the known sparse array geometry into a standard least squares problem, making it more robust and effective than unconstrained methods when dealing with irregular sensor placements.
Terminology
Summary
Sparse non-uniform array-enabled affine frequency division multiplexing (AFDM) is a promising candidate for integrated sensing and communication (ISAC), while its performance critically depends on accurate target parameter estimation The proposed algorithm nearly attains Cramer-Rao bound ´ (CRB) and significantly outperforms unconstrained ALS and conventional methods, validating its effectiveness for sparse nonuniform array-enabled AFDM ISAC systems
System Modeling
The system considers a sparse non-uniform array-enabled AFDM ISAC system that performs sensing of R surrounding targets while simultaneously supporting communication with several users in the vicinity of the base station The received signal is formulated as a high-order tensor, and subsequent analysis involves extracting factor matrices from this tensor to characterize physical parameters The received signal can be further represented as a tensor Y ∈ C Nr×N×Nt = Z + W, where W denotes the corresponding noise in the DAF domain
Constrained Tensor Decomposition
A manifold-constrained alternating least squares (ALS) algorithm is developed by exploiting the sparse array geometry structure, enabling robust factor matrix extraction and direct angle estimation The problem of factor matrices extraction for tensor signals is formulated as min aˆr,bˆr,cˆr Y −X R r=1 aˆr ◦ bˆ r ◦ cˆr F To overcome the drawbacks of standard ALS, the constrained alternative least square (CALS) algorithm is introduced by adding constraints Aˆ ∈ Γ1 and Cˆ ∈ Γ2 to the optimization problem The optimization of this constrained problem is achieved via the ALS framework within Niter iterations until convergence
Parameter Estimation Refinement
After recovering the factor matrices, the extraction of sensing parameters involves a two-stage process First, coarse estimation of the integer delay is performed using a scheme analogous to matched filtering based on pulse compression principles Second, an iterative one dimensional golden section search method is employed to refine the normalized delay and Doppler shift parameters The objective function used in the golden section search is written as f = ΛgΠlS H yr
Performance Validation
Simulation results demonstrate that the proposed algorithm nearly achieved the CRB and consistently outperformed baseline algorithms The algorithm achieves the best estimation performance for angular parameters and approaches the CRB, exhibiting a more pronounced advantage particularly in the low signalto-noise ratio regime Both tensor decomposition based estimation algorithms outperform the conventional MUSIC algorithm because MUSIC fails to exploit the high-dimensional structural information inherent in the received signal The computational complexity of the proposed two-stage algorithm is given by O(Niter(3NtNr+R2 (NtN + NrN + NtNr) + RTiter(Nt + Nr))) The dominant term in the complexity of the parameter estimation stage is O(T RN log N)
Conclusion
In this paper, a constrained tensor decomposition-based sensing framework was developed for delay, Doppler, and angle estimation in sparse non-uniform array-enabled AFDM ISAC systems The framework establishes a tensor model by jointly characterizing the time, frequency, and spatial domains The proposed method successfully incorporates the sparse array geometry into the tensor decomposition to enable robust factor matrix extraction and direct angle estimation The delay and Doppler shifts are subsequently refined through an iterative one dimensional golden section search using the decomposed factor matrices The simulation results confirm that the proposed algorithm nearly achieved the CRB and consistently outperformed baseline algorithms
References
[1] Z. Wei, H. Qu, Y. Wang, X. Yuan, H. Wu, Y. Du, K. Han, N. Zhang, and Z. p. 4 Feng “Integrated sensing and communication signals toward 5G-A and 6G: A survey” IEEE Internet Things J. p. 4, vol 10 no 13 pp 11068–11092 Jul. p. 0 2023
[2] D. C. Nguyen, M. Ding, P. N Pathirana, A Seneviratne, J Li, D Niyato, O Dobre and H V Poor “6G Internet of things: A comprehensive survey” IEEE Internet Things J. p. 4, vol 9 no 1 pp 359–383 Jan. p. 4 2022
[3] Y Li, J Chuang and N Sollenberger “Transmitter diversity for OFDM systems and its impact on high-rate data wireless networks” IEEE J Sel Areas Commun., vol 17 no 7 pp 1233–1243 Jul. p. 4 1999
[4] Bemani, Ali and Ksairi, Nassar and Kountouris, Marios “Affine frequency division multiplexing for next generation wireless communications” IEEE Trans Wireless Commun., vol 22 no 11 pp 8214–8229 Nov. p. 4 2023
[5] A Bemani, N Ksairi, and Kountouris, Marios “AFDM: A full diversity next generation waveform for high mobility communications” in 2021 IEEE International Conference on Communications Workshops (ICC Workshops) Jun. 2021 pp 1–6
[6] Bemani, Ali, N Ksairi, and M Kountouris “Integrated sensing and communications with affine frequency division multiplexing” IEEE Wireless Commun Lett., vol 13 no 5 pp 1255–1259 May 2024
[7] Z Yang, W Wang, B Ding, B Rao and D Song “Sparse antenna array synthesis method for higher efficiency and lower cost via adaptive multipoint mutation genetic algorithm” IEEE Trans Green Commun Netw., vol 10 pp 2190–2205 Feb. p. 4 2026
[8] S Yang, W Lyu, Z Zhang and C Yuen “Enhancing near-field sensing and communications with sparse arrays: Potentials, challenges, and emerging trends” arXiv e-prints 2023 [Online] Available: https://arxiv.org/abs/2309. p. 408681
[9] R Zhang, L Cheng, S Wang, Y Lou, Y Gao, W Wu and D W K Ng “Integrated sensing and communication with massive MIMO: A unified tensor approach for channel and target parameter estimation” IEEE Trans Wireless Commun., vol 23 no 8 pp 8571–8587 Aug. p. 4 2024
[10] Z Zhou, J Fang, L Yang, H Li, Z Chen and R S Blum “Lowrank tensor decomposition-aided channel estimation for millimeter wave MIMO-OFDM systems” IEEE J Sel Areas Commun., vol 35 no 7 pp 1524–1538 Jul. p. 4 2017
[11] Y Luo, Y L Guan, Y Ge, D Gonzalez G and C Yuen “A novel angle-delay-doppler estimation scheme for AFDM-ISAC system in mixed near-field and far-field scenarios” IEEE Internet Things J., vol 12 no 13 pp 22669–22682 Jul. p. 0 2025
[12] L Chen and H Li “Tensor-based angle-range estimation in near-field ISAC systems” in 59th Asilomar Conference on Signals, Systems, and Computers Oct.
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
-
Constraint-based factor matrix extraction for tensor decomposition ensures that
the loss of structural information caused by dimensionality flattening is avoided
when processing high-dimensional received signals, leading to more precise parameter estimation. -
The constrained tensor decomposition framework enables the extraction of delay, Doppler shift, and angle parameters independently from the factor matrices of their respective dimensions by exploiting
the sparse array geometry structure,
which results in a robust method for estimating these coupled parameters. -
By incorporating the manifold knowledge of sparse non-uniform arrays into the alternating least squares algorithm, the system can achieve
robust factor matrix extraction and direct angle estimation
even when existing tensor models are not applicable to irregular spatial sampling. -
The iterative one dimensional golden section search method refines delay and Doppler shift estimates by optimizing an objective function
f = yr 2
using the decomposed factor matrices, allowing for the determination ofthe integer components of normalized delay and Doppler shift information can be obtained unambiguously from locr.
-
The proposed system can perform high-resolution target parameter estimation in sparse non-uniform array enabled AFDM ISAC systems by nearly attaining the Cramer-Rao bound and significantly outperforming unconstrained ALS and conventional methods, especially in low signalto noise ratio regimes.
Sources
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