Tesi etd-09162025-110404
Link copiato negli appunti
Tipo di tesi
Corso di Dottorato (D.M.226/2021)
Autore
AHMAD, FAWAD
URN
etd-09162025-110404
Titolo
Photonic-Assisted Distributed MIMO Radar: Advance Signal Processing for High Resolution Detection and Imaging
Settore scientifico disciplinare
ING-INF/03
Corso di studi
Istituto di Tecnologie della Comunicazione, dell'Informazione e della Percezione - Ph.D. in Emerging Digital Technologies
Relatori
relatore BOGONI, ANTONELLA
Parole chiave
- Distributed MIMO Radar
- Radar on UAVs
- Target Detection
- Radar Imaging
- SAR Imaging
- Enhancement Techniques
- Machine Learning
- Radar Cross Section; Signal Processing
- Optimization
Data inizio appello
13/02/2026;
Disponibilità
parziale
Riassunto analitico
This thesis presents a comprehensive investigation into advanced radar imaging techniques, focusing on the design, simulation, and experimental validation of a photonics-assisted distributed Multiple-Input Multiple-Output (D-MIMO) radar system. The research is motivated by the growing demand for high-resolution, robust, and adaptive radar-based surveillance solutions capable of operating in complex environments, including ground-based, airborne, and spaceborne platforms.
The study begins by exploring the limitations of conventional monostatic and bistatic radar systems, particularly in terms of spatial resolution, target detectability, and operational flexibility. To face these challenges, a D-MIMO radar architecture mounted on a swarm of drones is proposed, leveraging spatial diversity and photonics technologies to achieve phase coherent signal processing across widely separated radar heads (RHs). The integration of microwave photonics solutions enables low-noise, high-bandwidth signal generation and distribution, addressing synchronization issues inherent in RF-based systems.
A MATLAB-based simulation framework is developed to model the D-MIMO radar system under both ideal and non-ideal operational conditions. The simulator supports both coherent and non-coherent processing as well as multi-band operation. Key performance indicators (KPIs) such as range resolution, cross-range resolution, peak-to-sidelobe ratios, and localization error are defined and evaluated across various system configurations. The simulation results are validated through indoor experimental testbeds using a 2\ \times2 photonics assisted by MIMO radar demonstrator. To enhance radar system performance, the thesis introduces both hardware-based and algorithmic optimization techniques. Hardware strategies include varying swarm geometry, baseline length, and operating bandwidth, while algorithmic approaches employ metaheuristic optimization (Particle Swarm Optimization and Genetic algorithm) and adaptive post-processing using Z-score based Kaiser windowing. A hybrid framework based on RH placement optimization and sidelobe suppression is proposed, achieving simultaneous improvement in spatial resolution reaching about 2\ \times2 cm in range and cross range with target detectability of more than 90% using Cell-Averaging Constant false alarm rate (CA-CFAR) and without increasing hardware complexity. This enhancement framework is applicable to any D-MIMO radar system where resource efficiency is a priority.
The ability of radar system to image an extended target is a key requirement of next generation radar system mounted on drones. Typically, extended targets have been simplified as point targets, assumed to be encapsulated within a single large resolution cell. This thesis overcomes that limitation by considering truly extended targets that span multiple resolution cells. As a foundational step for imaging, radar cross section (RCS) estimation is a crucial step in radar processing. Here, RCS modelling of a 40m long yacht is done using both physical option (POFACETS) and machine learning (ML) as an initial step for imaging. A supervised ensemble ML model is trained to predict RCS values with high accuracy and low computational cost, enabling real-time estimation of complex targets such as yachts and tanks. These RCS values are integrated into the D-MIMO radar imaging simulator to generate high-resolution images of extended targets. This D-MIMO imaging framework is validated through an indoor experiment by considering a multi-scatterer and multi-resolution cell triangular shaped target. Finally, the thesis extends the radar imaging framework from airborne to spaceborne applications by investigating D-MIMO Synthetic Aperture Radar (SAR) systems. A signal model for D-MIMO SAR is developed, and the limitations of conventional frequency-domain algorithms such as Range-Doppler Algorithm (RDA) are addressed through the implementation of a back-projection algorithm (BPA). Comparative analysis demonstrates the superiority of BPA in handling non-linear trajectories and distributed geometries, laying the foundation for future satellite-based MIMO SAR imaging systems.
Overall, this work contributes novel methodologies for photonics-assisted D-MIMO radar system design, performance enhancement, RCS modelling and imaging of extended targets. It establishes a scalable and resource-efficient framework for next-generation radar surveillance across airborne and spaceborne platforms.
The study begins by exploring the limitations of conventional monostatic and bistatic radar systems, particularly in terms of spatial resolution, target detectability, and operational flexibility. To face these challenges, a D-MIMO radar architecture mounted on a swarm of drones is proposed, leveraging spatial diversity and photonics technologies to achieve phase coherent signal processing across widely separated radar heads (RHs). The integration of microwave photonics solutions enables low-noise, high-bandwidth signal generation and distribution, addressing synchronization issues inherent in RF-based systems.
A MATLAB-based simulation framework is developed to model the D-MIMO radar system under both ideal and non-ideal operational conditions. The simulator supports both coherent and non-coherent processing as well as multi-band operation. Key performance indicators (KPIs) such as range resolution, cross-range resolution, peak-to-sidelobe ratios, and localization error are defined and evaluated across various system configurations. The simulation results are validated through indoor experimental testbeds using a 2\ \times2 photonics assisted by MIMO radar demonstrator. To enhance radar system performance, the thesis introduces both hardware-based and algorithmic optimization techniques. Hardware strategies include varying swarm geometry, baseline length, and operating bandwidth, while algorithmic approaches employ metaheuristic optimization (Particle Swarm Optimization and Genetic algorithm) and adaptive post-processing using Z-score based Kaiser windowing. A hybrid framework based on RH placement optimization and sidelobe suppression is proposed, achieving simultaneous improvement in spatial resolution reaching about 2\ \times2 cm in range and cross range with target detectability of more than 90% using Cell-Averaging Constant false alarm rate (CA-CFAR) and without increasing hardware complexity. This enhancement framework is applicable to any D-MIMO radar system where resource efficiency is a priority.
The ability of radar system to image an extended target is a key requirement of next generation radar system mounted on drones. Typically, extended targets have been simplified as point targets, assumed to be encapsulated within a single large resolution cell. This thesis overcomes that limitation by considering truly extended targets that span multiple resolution cells. As a foundational step for imaging, radar cross section (RCS) estimation is a crucial step in radar processing. Here, RCS modelling of a 40m long yacht is done using both physical option (POFACETS) and machine learning (ML) as an initial step for imaging. A supervised ensemble ML model is trained to predict RCS values with high accuracy and low computational cost, enabling real-time estimation of complex targets such as yachts and tanks. These RCS values are integrated into the D-MIMO radar imaging simulator to generate high-resolution images of extended targets. This D-MIMO imaging framework is validated through an indoor experiment by considering a multi-scatterer and multi-resolution cell triangular shaped target. Finally, the thesis extends the radar imaging framework from airborne to spaceborne applications by investigating D-MIMO Synthetic Aperture Radar (SAR) systems. A signal model for D-MIMO SAR is developed, and the limitations of conventional frequency-domain algorithms such as Range-Doppler Algorithm (RDA) are addressed through the implementation of a back-projection algorithm (BPA). Comparative analysis demonstrates the superiority of BPA in handling non-linear trajectories and distributed geometries, laying the foundation for future satellite-based MIMO SAR imaging systems.
Overall, this work contributes novel methodologies for photonics-assisted D-MIMO radar system design, performance enhancement, RCS modelling and imaging of extended targets. It establishes a scalable and resource-efficient framework for next-generation radar surveillance across airborne and spaceborne platforms.
File
| Nome file | Dimensione |
|---|---|
Ci sono 1 file riservati su richiesta dell'autore. |
|