Marius Necsoiu
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Marius Necsoiu has a total of 46 co-authors. Here are the top co-authors by number of shared publications:
- Alexandru Onaca
Department of Geography, West University of Timisoara, 300223 Timisoara, Romania
2 shared publications - Deniz Erdogmus
Northeastern University, Boston, MA, USA
2 shared publications - Gunar Schirner
Department of Electrical and Computer Engineering, Northeastern University, Boston, USA
2 shared publications - Michael Potter
Northeastern University, Boston, MA, USA
2 shared publications - Murat Akcakaya
University of Pittsburgh
2 shared publications Sarah Wigginton
2 shared publications
Latest Publications
- 6 February 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Cognitive Communications and Networking
- Vol. 12, 6466-6477
TACM-MR: Topographically Augmented Channel Model Multi-Receiver Dataset
Electromagnetic spectrum congestion requires new methods to recognize and understand other transmitters for effective sharing. Existing Automatic Modulation Recognition (AMR) research utilizes datasets for single channel (SISO) or collocated systems (MIMO) with simplistic propagation models, lacking realistic propagation channels and the spatial diversity of distributed multi-receiver scenarios. We propose the Topographically Augmented Channel Model 2026.1 (TACM2026.1), a spatially-separated but contemporaneous multi-receiver dataset using propagation channels derived from simulated topographies. We modify six contemporary AMR Machine Learning (ML) models to process multiple concurrent receivers. To accommodate variations in signal occupied bandwidth, we identify the$E_{s}/N_{0}$ metric for signal quality and the$\sum E_{s}/N_{0}$ metric to quantify aggregate signal quality. Our systematic evaluation demonstrates that classification performance depends on aggregate received power ($\sum E_{s}/N_{0}$ ) rather than its distribution across receivers. The Iso-Power configuration, equalizing power across receivers, demonstrates models genuinely aggregate spatial diversity rather than simply selecting dominant receivers. This learned diversity combining is comparable (1.72% difference at$E_{s}/N_{0}=50\text {dB}$ with eight receivers) to the optimal diversity combination (MRC)) but requires no channel state information. Practically, eight-receiver systems achieve 75% accuracy at$1.26\times $ the range of single-receiver systems (4km$\rightarrow ~5$ km), enabling distributed spectrum sensing without complex infrastructure and channel information.- 6 February 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Cognitive Communications and Networking
- Vol. 12, 6466-6477
TACM-MR: Topographically Augmented Channel Model Multi-Receiver Dataset
Electromagnetic spectrum congestion requires new methods to recognize and understand other transmitters for effective sharing. Existing Automatic Modulation Recognition (AMR) research utilizes datasets for single channel (SISO) or collocated systems (MIMO) with simplistic propagation models, lacking realistic propagation channels and the spatial diversity of distributed multi-receiver scenarios. We propose the Topographically Augmented Channel Model 2026.1 (TACM2026.1), a spatially-separated but contemporaneous multi-receiver dataset using propagation channels derived from simulated topographies. We modify six contemporary AMR Machine Learning (ML) models to process multiple concurrent receivers. To accommodate variations in signal occupied bandwidth, we identify the$E_{s}/N_{0}$ metric for signal quality and the$\sum E_{s}/N_{0}$ metric to quantify aggregate signal quality. Our systematic evaluation demonstrates that classification performance depends on aggregate received power ($\sum E_{s}/N_{0}$ ) rather than its distribution across receivers. The Iso-Power configuration, equalizing power across receivers, demonstrates models genuinely aggregate spatial diversity rather than simply selecting dominant receivers. This learned diversity combining is comparable (1.72% difference at$E_{s}/N_{0}=50\text {dB}$ with eight receivers) to the optimal diversity combination (MRC)) but requires no channel state information. Practically, eight-receiver systems achieve 75% accuracy at$1.26\times $ the range of single-receiver systems (4km$\rightarrow ~5$ km), enabling distributed spectrum sensing without complex infrastructure and channel information.- 29 May 2025
- Published by SPIE-Intl Soc Optical Eng
- Vol. 13460, 42-60
Quantifying narrowband interference using deep learning models with explainable AI Integration
Efficient sharing of narrowband communication channels among radar and communication signals is a critical challenge in modern systems. As spectrum congestion intensifies, interference between these signals can severely degrade system performance. In this work, we explore the application of deep learning methods, specifically CNNs1–3 and ViTs,4, 5 for classifying and measuring interference on narrowband channels. We also propose an XAI approach that integrates Visual Large Language Models (VLLMs) with saliency maps to provide interactive explanations of model decisions.6, 7 Our results demonstrate that ViTs deliver superior classification accuracy across varying SNR conditions, reflecting their ability to capture complex time-frequency patterns.8–10 Furthermore, the addition of explainable AI enhances human interpretability by highlighting critical features in spectrograms and generating coherent textual summaries.11–13 This multimodal explainability framework shows promise for real-time interference mitigation, improved transparency, and accelerated decision-making in high-stakes applications, such as defense, aviation, and emerging 5G/6G systems.
- 29 May 2025
- Published by SPIE-Intl Soc Optical Eng
- Vol. 13460, 42-60
Quantifying narrowband interference using deep learning models with explainable AI Integration
Efficient sharing of narrowband communication channels among radar and communication signals is a critical challenge in modern systems. As spectrum congestion intensifies, interference between these signals can severely degrade system performance. In this work, we explore the application of deep learning methods, specifically CNNs1–3 and ViTs,4, 5 for classifying and measuring interference on narrowband channels. We also propose an XAI approach that integrates Visual Large Language Models (VLLMs) with saliency maps to provide interactive explanations of model decisions.6, 7 Our results demonstrate that ViTs deliver superior classification accuracy across varying SNR conditions, reflecting their ability to capture complex time-frequency patterns.8–10 Furthermore, the addition of explainable AI enhances human interpretability by highlighting critical features in spectrograms and generating coherent textual summaries.11–13 This multimodal explainability framework shows promise for real-time interference mitigation, improved transparency, and accelerated decision-making in high-stakes applications, such as defense, aviation, and emerging 5G/6G systems.
- 13 January 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Aerospace and Electronic Systems
- Vol. 61 (3) , 6236-6251
Continuously Optimizing Radar Placement With Model-Predictive Path Integrals
Continuously optimizing sensor placement is essential for precise target localization in various military and civilian applications. While information theory has shown promise in optimizing sensor placement, many studies oversimplify sensor measurement models or neglect dynamic constraints of mobile sensors. To address these challenges, we employ a range measurement model that incorporates radar parameters and radar-target distance, coupled with Model Predictive Path Integral (MPPI) control to manage complex environmental obstacles and dynamic constraints. We compare the proposed approach against stationary radars or simplified range measurement models based on the Root Mean Squared Error (RMSE) of the Cubature Kalman Filter (CKF) estimator for the targets' state. Additionally, we visualize the evolving geometry of radars and targets over time, highlighting areas of highest measurement information gain, demonstrating the strengths of the approach. The proposed strategy outperforms stationary radars and simplified range measurement models in target localization, achieving a 38-74% reduction in mean Root Mean Squared Error (RMSE) and a 33-79% reduction in the upper tail of the 90% Highest Density Interval (HDI) over 500 Monte Carlo (MC) trials across all time steps. Code available at https://github.com/mlpotter/OptimalTrackingAndControl .
- 13 January 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Aerospace and Electronic Systems
- Vol. 61 (3) , 6236-6251
Continuously Optimizing Radar Placement With Model-Predictive Path Integrals
Continuously optimizing sensor placement is essential for precise target localization in various military and civilian applications. While information theory has shown promise in optimizing sensor placement, many studies oversimplify sensor measurement models or neglect dynamic constraints of mobile sensors. To address these challenges, we employ a range measurement model that incorporates radar parameters and radar-target distance, coupled with Model Predictive Path Integral (MPPI) control to manage complex environmental obstacles and dynamic constraints. We compare the proposed approach against stationary radars or simplified range measurement models based on the Root Mean Squared Error (RMSE) of the Cubature Kalman Filter (CKF) estimator for the targets' state. Additionally, we visualize the evolving geometry of radars and targets over time, highlighting areas of highest measurement information gain, demonstrating the strengths of the approach. The proposed strategy outperforms stationary radars and simplified range measurement models in target localization, achieving a 38-74% reduction in mean Root Mean Squared Error (RMSE) and a 33-79% reduction in the upper tail of the 90% Highest Density Interval (HDI) over 500 Monte Carlo (MC) trials across all time steps. Code available at https://github.com/mlpotter/OptimalTrackingAndControl .
- 19 August 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Aerospace and Electronic Systems
- Vol. 61 (1) , 151-161
Multistatic-Radar RCS-Signature Recognition of Aerial Vehicles: A Bayesian Fusion Approach
Radar Automated Target Recognition (RATR) for Unmanned Aerial Vehicles (UAVs) involves transmitting Electromagnetic Waves (EMWs) and performing target type recognition on the received radar echo, which has important applications in defense and aerospace. Previous work has demonstrated the benefits of employing multistatic radar configurations in RATR compared to monostatic radar configurations. However, multistatic radar configurations commonly use fusion methods which combine the classification vectors of multiple individual radars suboptimally from a probabilistic perspective. To address this issue, this work leverages Bayesian analysis to provide a fully Bayesian RATR framework for UAV type classification. Specifically, we employ an Optimal Bayesian Fusion (OBF) method, from the Bayesian perspective of expected 0-1 loss, to formulate a posterior distribution that aggregates the classification probability vectors from multiple individual radar observations at a given time step. This OBF method is used to update a separate Recursive Bayesian Classification (RBC) posterior distribution on the target UAV type. The RBC posterior is conditioned on all historical observations made from multiple radars across multiple time steps. To evaluate the proposed approach, we simulate random walk trajectories for seven drones and correspond the target's aspect angles to Radar Cross Section (RCS) measurements acquired in an anechoic chamber. We then compare the performance of single radar Automated Target Recognition (ATR) system and suboptimal fusion methods against the OBF method. We empirically show that the OBF method, integrated with RBC, significantly outperforms other fusion methods and single radar configuration in terms of classification accuracy.
- 19 August 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Aerospace and Electronic Systems
- Vol. 61 (1) , 151-161
Multistatic-Radar RCS-Signature Recognition of Aerial Vehicles: A Bayesian Fusion Approach
Radar Automated Target Recognition (RATR) for Unmanned Aerial Vehicles (UAVs) involves transmitting Electromagnetic Waves (EMWs) and performing target type recognition on the received radar echo, which has important applications in defense and aerospace. Previous work has demonstrated the benefits of employing multistatic radar configurations in RATR compared to monostatic radar configurations. However, multistatic radar configurations commonly use fusion methods which combine the classification vectors of multiple individual radars suboptimally from a probabilistic perspective. To address this issue, this work leverages Bayesian analysis to provide a fully Bayesian RATR framework for UAV type classification. Specifically, we employ an Optimal Bayesian Fusion (OBF) method, from the Bayesian perspective of expected 0-1 loss, to formulate a posterior distribution that aggregates the classification probability vectors from multiple individual radar observations at a given time step. This OBF method is used to update a separate Recursive Bayesian Classification (RBC) posterior distribution on the target UAV type. The RBC posterior is conditioned on all historical observations made from multiple radars across multiple time steps. To evaluate the proposed approach, we simulate random walk trajectories for seven drones and correspond the target's aspect angles to Radar Cross Section (RCS) measurements acquired in an anechoic chamber. We then compare the performance of single radar Automated Target Recognition (ATR) system and suboptimal fusion methods against the OBF method. We empirically show that the OBF method, integrated with RBC, significantly outperforms other fusion methods and single radar configuration in terms of classification accuracy.
- 1 July 2019
- Published by Society of Exploration Geophysicists in The Leading Edge
- Vol. 38 (7) , 550-553
Nonlinear inversion of subsidence signatures induced by tunnels detected with surface and remote sensing measurements
The state of the art in predicting tunnel-induced subsidence settlements is based on empirical and analytical methods. Empirical methods are useful when the equations are implemented with host medium properties where tunnels have been excavated. Analytical solutions can predict tunneling-induced ground movements, with the predictions accounting for tunnel radius and depth as well as ground-loss parameters in soft soils. The drawback is that these methods require human intervention, as each model must be adjusted manually by the interpreter until the model signature fits the observed data. It would take tremendous effort to evaluate displacement anomalies detected by remote sensing methods using such forward-modeling methods. Therefore, we present a method based on an inversion algorithm that automatically inverts subsidence signatures for tunnel radius, depth, Poisson's ratio, and the gap parameter. It is an advancement over conventional methods because it does not require a first guess, and it can invert several subsidence signatures in a matter of minutes. The algorithm, coupled with remote sensing-based displacement maps, is a cost-effective solution in operational characterization of displacement anomalies. We demonstrate that observed and predicted subsidence signatures are in good agreement with existing tunnel data in uniform clay and that the inversion parameters correspond to those predicted with forward modeling alone.
- 1 July 2019
- Published by Society of Exploration Geophysicists in The Leading Edge
- Vol. 38 (7) , 550-553
Nonlinear inversion of subsidence signatures induced by tunnels detected with surface and remote sensing measurements
The state of the art in predicting tunnel-induced subsidence settlements is based on empirical and analytical methods. Empirical methods are useful when the equations are implemented with host medium properties where tunnels have been excavated. Analytical solutions can predict tunneling-induced ground movements, with the predictions accounting for tunnel radius and depth as well as ground-loss parameters in soft soils. The drawback is that these methods require human intervention, as each model must be adjusted manually by the interpreter until the model signature fits the observed data. It would take tremendous effort to evaluate displacement anomalies detected by remote sensing methods using such forward-modeling methods. Therefore, we present a method based on an inversion algorithm that automatically inverts subsidence signatures for tunnel radius, depth, Poisson's ratio, and the gap parameter. It is an advancement over conventional methods because it does not require a first guess, and it can invert several subsidence signatures in a matter of minutes. The algorithm, coupled with remote sensing-based displacement maps, is a cost-effective solution in operational characterization of displacement anomalies. We demonstrate that observed and predicted subsidence signatures are in good agreement with existing tunnel data in uniform clay and that the inversion parameters correspond to those predicted with forward modeling alone.
- 1 September 2016
- Published by Elsevier in Quaternary International
- Vol. 415, 164-174
Recent morphodynamics of alpine lakes in Southern Carpathian Mountains using high-resolution optical imagery
- 1 September 2016
- Published by Elsevier in Quaternary International
- Vol. 415, 164-174
Recent morphodynamics of alpine lakes in Southern Carpathian Mountains using high-resolution optical imagery
- 1 May 2016
- Published by Elsevier in Remote Sensing of Environment
- Vol. 177, 21-36
Rock glacier dynamics in Southern Carpathian Mountains from high-resolution optical and multi-temporal SAR satellite imagery
- 1 May 2016
- Published by Elsevier in Remote Sensing of Environment
- Vol. 177, 21-36
Rock glacier dynamics in Southern Carpathian Mountains from high-resolution optical and multi-temporal SAR satellite imagery
- Top Cited Papers
- 14 March 2016
- Published by Springer Nature in Nature Geoscience
- Vol. 9 (4) , 312-318
Pan-Arctic ice-wedge degradation in warming permafrost and its influence on tundra hydrology
- 14 March 2016
- Published by Springer Nature in Nature Geoscience
- Vol. 9 (4) , 312-318
Top Cited PapersPan-Arctic ice-wedge degradation in warming permafrost and its influence on tundra hydrology
- Open Access
- 1 October 2015
- Published by Elsevier in Engineering Geology
- Vol. 197, 267-277
Detection of uranium mill tailings settlement using satellite-based radar interferometry
- 1 October 2015
- Published by Elsevier in Engineering Geology
- Vol. 197, 267-277
Open AccessDetection of uranium mill tailings settlement using satellite-based radar interferometry