Marius Pesavento
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Marius Pesavento has a total of 74 co-authors. Here are the top co-authors by number of shared publications:
- Yang Yang
Competence Center for High Performance Computing, Fraunhofer Institute for Industrial Mathematics, Kaiserslautern, Germany
5 shared publications - Bjorn Ottersten
Interdisciplinary Centre for Security, Reliability and Trust (SnT), University of Luxembourg, Luxembourg City, Luxembourg
4 shared publications - Gianni Allevato
Department of Electrical Engineering and Information Technology, Technische Universität Darmstadt, Germany
4 shared publications - Lukas Schynol
Technical University of Darmstadt, Darmstadt, Germany
4 shared publications - Mario Kupnik
Measurement and Sensor Technology Group, Technical University of Darmstadt, Darmstadt, Germany
4 shared publications - Mats Viberg
Blekinge Institute of Technology, Karlskrona, Sweden
4 shared publications
Latest Publications
- 27 May 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Signal Processing
- Vol. 74 (1053587X) , 2503-2518
Maximum a Posteriori Direction-of-Arrival Estimation via Mixed-Integer Semidefinite Programming
We propose a joint sparse maximum a posteriori (MAP) estimator for DOA estimation from multiple snapshots, reformulated as a mixed-integer semidefinite program (MISDP). This enables efficient computation of globally optimal solutions using off-the-shelf MISDP solvers based on the branch-and-bound method. Unlike other nonconvex approaches for joint sparse recovery, such as the greedy methods and sparse Bayesian learning techniques, it provides a solution with an optimality assessment even with early termination. Additionally, we present a more scalable approximate solution approach for the MISDP problem based on randomized rounding. Numerical simulations demonstrate the improved threshold behavior, resolution, and robustness of our proposed method against popular DOA estimation methods. In particular, the proposed method applied with the randomized rounding algorithm exhibits a superior estimation performance at a significantly reduced running time, compared to the deterministic maximum likelihood (DML) estimator.
- 27 May 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Signal Processing
- Vol. 74 (1053587X) , 2503-2518
Maximum a Posteriori Direction-of-Arrival Estimation via Mixed-Integer Semidefinite Programming
We propose a joint sparse maximum a posteriori (MAP) estimator for DOA estimation from multiple snapshots, reformulated as a mixed-integer semidefinite program (MISDP). This enables efficient computation of globally optimal solutions using off-the-shelf MISDP solvers based on the branch-and-bound method. Unlike other nonconvex approaches for joint sparse recovery, such as the greedy methods and sparse Bayesian learning techniques, it provides a solution with an optimality assessment even with early termination. Additionally, we present a more scalable approximate solution approach for the MISDP problem based on randomized rounding. Numerical simulations demonstrate the improved threshold behavior, resolution, and robustness of our proposed method against popular DOA estimation methods. In particular, the proposed method applied with the randomized rounding algorithm exhibits a superior estimation performance at a significantly reduced running time, compared to the deterministic maximum likelihood (DML) estimator.
- Open Access
- 5 March 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Open Journal of the Communications Society
- Vol. 7, 2559-2576
Hybrid Downlink Beamforming With Outage Constraints Under Imperfect CSI Using Model-Driven Deep Learning
We consider energy-efficient multi-user hybrid downlink beamforming (BF) and power allocation under imperfect channel state information (CSI) and probabilistic outage constraints. In this domain, classical optimization methods resort to computationally costly conic optimization problems. Meanwhile, generic deep network (DN) architectures lack interpretability and require large training data sets to generalize well. In this paper, we therefore propose a lightweight model-aided deep learning architecture based on a greedy selection algorithm for analog beam codewords. The architecture relies on an instance-adaptive augmentation of the signal model to estimate the impact of the CSI error. To learn the DN parameters, we derive a novel and efficient implicit representation of the nested constrained BF problem and prove sufficient conditions for the existence of the corresponding gradient. In the loss function, we utilize an annealing-based approximation of the outage compared to conventional quantile-based loss terms. This approximation adaptively anneals towards the exact probabilistic constraint depending on the current level of quality of service (QoS) violation. Simulations validate that the proposed DN can achieve the nominal outage level under CSI error due to channel estimation and channel compression, while allocating less power than benchmarks. Thereby, a single trained model generalizes to different numbers of users, QoS requirements and levels of CSI quality. We further show that the adaptive annealing-based loss function can accelerate the training and yield a better power-outage trade-off.
- 5 March 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Open Journal of the Communications Society
- Vol. 7, 2559-2576
Open AccessHybrid Downlink Beamforming With Outage Constraints Under Imperfect CSI Using Model-Driven Deep Learning
We consider energy-efficient multi-user hybrid downlink beamforming (BF) and power allocation under imperfect channel state information (CSI) and probabilistic outage constraints. In this domain, classical optimization methods resort to computationally costly conic optimization problems. Meanwhile, generic deep network (DN) architectures lack interpretability and require large training data sets to generalize well. In this paper, we therefore propose a lightweight model-aided deep learning architecture based on a greedy selection algorithm for analog beam codewords. The architecture relies on an instance-adaptive augmentation of the signal model to estimate the impact of the CSI error. To learn the DN parameters, we derive a novel and efficient implicit representation of the nested constrained BF problem and prove sufficient conditions for the existence of the corresponding gradient. In the loss function, we utilize an annealing-based approximation of the outage compared to conventional quantile-based loss terms. This approximation adaptively anneals towards the exact probabilistic constraint depending on the current level of quality of service (QoS) violation. Simulations validate that the proposed DN can achieve the nominal outage level under CSI error due to channel estimation and channel compression, while allocating less power than benchmarks. Thereby, a single trained model generalizes to different numbers of users, QoS requirements and levels of CSI quality. We further show that the adaptive annealing-based loss function can accelerate the training and yield a better power-outage trade-off.
- Open Access
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Open Journal of Signal Processing
- p. 1-20
Ambiguity Analysis and Design of Sparse Arrays Via Generalized Vandermonde Rank Conditions
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Open Journal of Signal Processing
- p. 1-20
Open AccessAmbiguity Analysis and Design of Sparse Arrays Via Generalized Vandermonde Rank Conditions
- 23 October 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Signal and Information Processing over Networks
- Vol. 11, 1422-1436
Decentralized Singular Value Decomposition for Feature-Wise Partitioned Data in Large-Scale Distributed Sensor Networks
This article studies the problem of decentralized Singular Value Decomposition (d-SVD), which is fundamental in various signal processing applications. Two scenarios are considered depending on the availability of the data matrix under consideration. In the first scenario, the matrix of interest is row wisely available in each local node in the network. In the second scenario, the matrix of interest implicitly forms an outer product from two different series of measurements. By combining the lightweight local rational function approximation approach with parallel averaging consensus algorithms, two d-SVD algorithms are proposed to cope with the two aforementioned scenarios. We evaluate the proposed algorithms using two application examples: decentralized sensor localization via low-rank matrix completion and decentralized passive radar detection. Moreover, a novel and non-trivial truncation technique, which employs a representative vector that is orthonormal to the principal signal subspace, is proposed to further reduce the communication cost associated with the d-SVD algorithms. Simulation results show that the proposed d-SVD algorithms converge to the centralized solution with reduced communication cost compared to those facilitated with the state-of-the-art decentralized power method.
- 23 October 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Signal and Information Processing over Networks
- Vol. 11, 1422-1436
Decentralized Singular Value Decomposition for Feature-Wise Partitioned Data in Large-Scale Distributed Sensor Networks
This article studies the problem of decentralized Singular Value Decomposition (d-SVD), which is fundamental in various signal processing applications. Two scenarios are considered depending on the availability of the data matrix under consideration. In the first scenario, the matrix of interest is row wisely available in each local node in the network. In the second scenario, the matrix of interest implicitly forms an outer product from two different series of measurements. By combining the lightweight local rational function approximation approach with parallel averaging consensus algorithms, two d-SVD algorithms are proposed to cope with the two aforementioned scenarios. We evaluate the proposed algorithms using two application examples: decentralized sensor localization via low-rank matrix completion and decentralized passive radar detection. Moreover, a novel and non-trivial truncation technique, which employs a representative vector that is orthonormal to the principal signal subspace, is proposed to further reduce the communication cost associated with the d-SVD algorithms. Simulation results show that the proposed d-SVD algorithms converge to the centralized solution with reduced communication cost compared to those facilitated with the state-of-the-art decentralized power method.
- 19 September 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Selected Topics in Signal Processing
- Vol. 19 (6) , 896-908
Crámer-Rao Bound Analysis for Mixed-ADC Direction of Arrival Estimation in Two-Dimensional Arrays
Digital modulation schemes, such as phase-modulated continuous wave (PMCW) and orthogonal frequency-division multiplexing (OFDM), require high-sampling analog to-digital converters (ADCs), which are power-consuming and costly. As the trend indicates an increase in the number of antennas employed, the issue becomes more critical. Mixed-analog-to-digital converter (ADC) arrays, combining high- and low-resolution ADCs, have recently been investigated to address this issue. In high-resolution radar systems, accurately estimating target positions requires capturing range, Doppler, azimuth, and elevation information. To achieve reliable azimuth and elevation estimation, two-dimensional antenna layouts are essential. Previous studies have focused solely on one-dimensional mixed-ADC assignments, such as uniform linear arrays (ULAs) and sparse linear arrays (SLAs). This paper addresses this problem by providing an analytical derivation of the multivariate Crámer-Rao bound (CRB) for estimating azimuth and elevation jointly in mixed-ADC arrays. Simulation results indicate that optimizing mixed-ADC assignment, i.e., identifying the optimal locations of high- and low-resolution ADCs, enhances the direction of arrival (DoA) estimation accuracy.
- 19 September 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Selected Topics in Signal Processing
- Vol. 19 (6) , 896-908
Crámer-Rao Bound Analysis for Mixed-ADC Direction of Arrival Estimation in Two-Dimensional Arrays
Digital modulation schemes, such as phase-modulated continuous wave (PMCW) and orthogonal frequency-division multiplexing (OFDM), require high-sampling analog to-digital converters (ADCs), which are power-consuming and costly. As the trend indicates an increase in the number of antennas employed, the issue becomes more critical. Mixed-analog-to-digital converter (ADC) arrays, combining high- and low-resolution ADCs, have recently been investigated to address this issue. In high-resolution radar systems, accurately estimating target positions requires capturing range, Doppler, azimuth, and elevation information. To achieve reliable azimuth and elevation estimation, two-dimensional antenna layouts are essential. Previous studies have focused solely on one-dimensional mixed-ADC assignments, such as uniform linear arrays (ULAs) and sparse linear arrays (SLAs). This paper addresses this problem by providing an analytical derivation of the multivariate Crámer-Rao bound (CRB) for estimating azimuth and elevation jointly in mixed-ADC arrays. Simulation results indicate that optimizing mixed-ADC assignment, i.e., identifying the optimal locations of high- and low-resolution ADCs, enhances the direction of arrival (DoA) estimation accuracy.
- Open Access
- 31 May 2025
- Published by Elsevier in Signal Processing
- Vol. 238
Sequential maximum-likelihood estimation of wideband polynomial-phase signals on sensor array
- 31 May 2025
- Published by Elsevier in Signal Processing
- Vol. 238
Open AccessSequential maximum-likelihood estimation of wideband polynomial-phase signals on sensor array
- Open Access
- 10 March 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Open Journal of Signal Processing
- Vol. 6, 417-433
Adaptive Anomaly Detection in Network Flows With Low-Rank Tensor Decompositions and Deep Unrolling
Anomaly detection (AD) is increasingly recognized as a key component for ensuring the resilience of future communication systems. While deep learning has shown state-of-the-art AD performance, its application in critical systems is hindered by concerns regarding training data efficiency, domain adaptation and interpretability. This work considers AD in network flows using incomplete measurements, leveraging a robust tensor decomposition approach and deep unrolling techniques to address these challenges. We first propose a novel block-successive convex approximation algorithm based on a regularized model-fitting objective where the normal flows are modeled as low-rank tensors and anomalies as sparse. An augmentation of the objective is introduced to decrease the computational cost. We apply deep unrolling to derive a novel deep network architecture based on our proposed algorithm, treating the regularization parameters as learnable weights. Inspired by Bayesian approaches, we extend the model architecture to perform online adaptation to per-flow and per-time-step statistics, improving AD performance while maintaining a low parameter count and preserving the problem's permutation equivariances. To optimize the deep network weights for detection performance, we employ a homotopy optimization approach based on an efficient approximation of the area under the receiver operating characteristic curve. Extensive experiments on synthetic and real-world data demonstrate that our proposed deep network architecture exhibits a high training data efficiency, outperforms reference methods, and adapts seamlessly to varying network topologies.
- 10 March 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Open Journal of Signal Processing
- Vol. 6, 417-433
Open AccessAdaptive Anomaly Detection in Network Flows With Low-Rank Tensor Decompositions and Deep Unrolling
Anomaly detection (AD) is increasingly recognized as a key component for ensuring the resilience of future communication systems. While deep learning has shown state-of-the-art AD performance, its application in critical systems is hindered by concerns regarding training data efficiency, domain adaptation and interpretability. This work considers AD in network flows using incomplete measurements, leveraging a robust tensor decomposition approach and deep unrolling techniques to address these challenges. We first propose a novel block-successive convex approximation algorithm based on a regularized model-fitting objective where the normal flows are modeled as low-rank tensors and anomalies as sparse. An augmentation of the objective is introduced to decrease the computational cost. We apply deep unrolling to derive a novel deep network architecture based on our proposed algorithm, treating the regularization parameters as learnable weights. Inspired by Bayesian approaches, we extend the model architecture to perform online adaptation to per-flow and per-time-step statistics, improving AD performance while maintaining a low parameter count and preserving the problem's permutation equivariances. To optimize the deep network weights for detection performance, we employ a homotopy optimization approach based on an efficient approximation of the area under the receiver operating characteristic curve. Extensive experiments on synthetic and real-world data demonstrate that our proposed deep network architecture exhibits a high training data efficiency, outperforms reference methods, and adapts seamlessly to varying network topologies.
- Open Access
- 14 January 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Open Journal of Signal Processing
- Vol. 6, 184-193
An Accelerated Successive Convex Approximation Scheme With Exact Step Sizes for L1-Regression
We consider the minimization of $\ell _{1}$-regularized least-squares problems. A recent optimization approach uses successive convex approximations with an exact line search, which is highly competitive, especially in sparse problem instances. This work proposes an acceleration scheme for the successive convex approximation technique with a negligible additional computational cost. We demonstrate this scheme by devising three related accelerated algorithms with provable convergence. The first introduces an additional descent step along the past optimization trajectory in the variable update, that is inspired by Nesterov's accelerated gradient method and uses a closed-form step size. The second performs a simultaneous descent step along both the best response and the past trajectory, thereby finding a two-dimensional step size, also in closed-form. The third algorithm combines the previous two approaches. All algorithms are hyperparameter-free. Empirical results confirm that the acceleration approaches improve the convergence rate compared to benchmark algorithms, and that they retain the benefits of successive convex approximation also in non-sparse instances.
- 14 January 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Open Journal of Signal Processing
- Vol. 6, 184-193
Open AccessAn Accelerated Successive Convex Approximation Scheme With Exact Step Sizes for L1-Regression
We consider the minimization of $\ell _{1}$-regularized least-squares problems. A recent optimization approach uses successive convex approximations with an exact line search, which is highly competitive, especially in sparse problem instances. This work proposes an acceleration scheme for the successive convex approximation technique with a negligible additional computational cost. We demonstrate this scheme by devising three related accelerated algorithms with provable convergence. The first introduces an additional descent step along the past optimization trajectory in the variable update, that is inspired by Nesterov's accelerated gradient method and uses a closed-form step size. The second performs a simultaneous descent step along both the best response and the past trajectory, thereby finding a two-dimensional step size, also in closed-form. The third algorithm combines the previous two approaches. All algorithms are hyperparameter-free. Empirical results confirm that the acceleration approaches improve the convergence rate compared to benchmark algorithms, and that they retain the benefits of successive convex approximation also in non-sparse instances.
- Open Access
- 28 November 2024
- Published by Elsevier in Signal Processing
- Vol. 230
A tensor model for the calibration of air-coupled ultrasonic sensor arrays in 3D imaging
- 28 November 2024
- Published by Elsevier in Signal Processing
- Vol. 230
Open AccessA tensor model for the calibration of air-coupled ultrasonic sensor arrays in 3D imaging
- Open Access
- 7 August 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Open Journal of Ultrasonics, Ferroelectrics, and Frequency Control
- Vol. 3, 113-127
Two-Scale Sparse Spiral Array Design for 3D Ultrasound Imaging in Air
Sparse array designs are a promising approach to improve the beam pattern and imaging quality, especially for applications, where hardware resources are severely limited. In particular, spiral sunflower arrays become increasingly popular due to their excellent point-spread-function (PSF) characteristics and their simple, deterministic and scalable design. Therefore, several sunflower modifications for further improvement have been investigated, e.g. density tapering based on window functions adapted from apodization techniques. In this article, we introduce a two-scale spiral array design concept, which exploits the specific PSF structure of the sunflower geometry, instead of relying on window functions. The modification proposed combines two nested sunflower sub-arrays featuring two different spatial element densities such that the locations of their respective main, side and grating lobe zones differ, resulting in a balanced and improved composite one-way PSF in terms of main lobe width (MLW) and maximum side lobe level (MSLL) under far-field and narrow-band conditions. First, we provide an analysis of the unmodified classic sunflower geometry, describe its PSF zones and show how their locations in the PSF can be estimated based on the array design parameters, which finally leads to the two-scale concept. Second, we examine a specific well-matching combination of nested sub-arrays to discuss the advantages and limitations of the resulting PSF. Third, we benchmark the respective optimum arrays of the classic sunflower and density tapering strategies with the two-scale method, where the latter shows an improved performance of the one-way PSF in terms of MLW and MSLL. Fourth, the two-scale design strategy is validated using a real-world 64-element prototype for narrow-band ultrasound imaging in air. We conduct two experiments to analyze the resulting PSF and angular resolution. Overall, the results demonstrate that the proposed flexible four-parameter concept is particularly valuable for high frame rate imaging as well as for transmit-only and receive-only applications.
- 7 August 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Open Journal of Ultrasonics, Ferroelectrics, and Frequency Control
- Vol. 3, 113-127
Open AccessTwo-Scale Sparse Spiral Array Design for 3D Ultrasound Imaging in Air
Sparse array designs are a promising approach to improve the beam pattern and imaging quality, especially for applications, where hardware resources are severely limited. In particular, spiral sunflower arrays become increasingly popular due to their excellent point-spread-function (PSF) characteristics and their simple, deterministic and scalable design. Therefore, several sunflower modifications for further improvement have been investigated, e.g. density tapering based on window functions adapted from apodization techniques. In this article, we introduce a two-scale spiral array design concept, which exploits the specific PSF structure of the sunflower geometry, instead of relying on window functions. The modification proposed combines two nested sunflower sub-arrays featuring two different spatial element densities such that the locations of their respective main, side and grating lobe zones differ, resulting in a balanced and improved composite one-way PSF in terms of main lobe width (MLW) and maximum side lobe level (MSLL) under far-field and narrow-band conditions. First, we provide an analysis of the unmodified classic sunflower geometry, describe its PSF zones and show how their locations in the PSF can be estimated based on the array design parameters, which finally leads to the two-scale concept. Second, we examine a specific well-matching combination of nested sub-arrays to discuss the advantages and limitations of the resulting PSF. Third, we benchmark the respective optimum arrays of the classic sunflower and density tapering strategies with the two-scale method, where the latter shows an improved performance of the one-way PSF in terms of MLW and MSLL. Fourth, the two-scale design strategy is validated using a real-world 64-element prototype for narrow-band ultrasound imaging in air. We conduct two experiments to analyze the resulting PSF and angular resolution. Overall, the results demonstrate that the proposed flexible four-parameter concept is particularly valuable for high frame rate imaging as well as for transmit-only and receive-only applications.