Trvanie projektu:
2020 – 2024

Vedúca krajina projektu:
USA a Holandsko

Popis projektu v anglickom jazyku:
Members from 11 countries investigated the integration of Compressive Sensing (CS) and Machine Learning (ML), in particular Neural Networks (NN), for radar applications for the purpose of improving the speed and applicability of CS techniques using deep net architectures with learned signal priors, while supporting learning in data-starved regimes with CS based models. Throughout the final report, the RTG members identified several enabling technologies, such as complex-data representations within neural networks, algorithm unrolling, and dictionary learning. The RTG members studied the development of these CS and ML techniques for the applications of Synthetic Aperture Radar (SAR) image classification, detection of targets in non-Gaussian clutter, and Radio Frequency (RF) sensor design. Three datasets are made available to NATO countries to enhance the repeatability and comparability of techniques:

1. UAV micro-Doppler signature measurements;

2. the Synthetic and Measured Paired and Labelled Experimental (SAMPLE) dataset for SAR ground vehicle classification; and

3. simulated targets within non-Gaussian clutter and Gaussian thermal noise in the form of radar range / Doppler maps.

Riešitelia za Akadémiu ozbrojených síl generála Milana Rastislava Štefánika:
pplk. Ing. Jozef PERĎOCH, PhD.

Trvanie projektu:
2021 – trvá

Vedúca krajina projektu:
Nemecko, Taliansko a Fínsko

Popis projektu v anglickom jazyku:
Members from more than 10 countries perform theoretical investigations and experiments as well as develop new methods to demonstrate the benefits of cognition in radar systems. Specifically, the group builds on the work done in SET-227 by investigating the role of learning in cognitive radar and distributed cognitive radar systems. Members are targeting to

• analysis and assessment of different learning and optimization methods for continuously and autonomously improving performance in different radar tasks and different operational levels of radar systems,

• analysis of the feasibility and implications of learning in a cognitive radar, for example, in terms of robustness, reliability, partial observability and trustworthiness,

• learning within a cognitive radar perception-action cycle starting from signal level to mission level,

• the cognitive radar experiments

Riešitelia za Akadémiu ozbrojených síl generála Milana Rastislava Štefánika:
pplk. Ing. Stanislava GAŽOVOVÁ, PhD.

Trvanie projektu:
2025 – trvá

Vedúca krajina projektu:
USA a Slovenská republika

Popis projektu v anglickom jazyku:
Three major applications are investigated by the RTG, namely:

1. RF signal processing utilizing SR and ML algorithms.

2. RF Signal Characterization and Classification, Classification of SAR and Micro-Doppler images with an emphasis on a limited training data regime.

3. Utilizing SR and ML algorithms in Electromagnetic warfare (EW) and jamming.

Seven common datasets are made available for use within SET-350:

• Synthetic and Measured Paired and Labeled Experimental (SAMPLE) base (10 classes of targets) and extended (20 classes of targets) database;

• Simulated dataset marked as ONERA v01 and ONERA v02, including several scenarios with different numbers of targets, noise, and non-Gaussian clutter;

• RCS measurements of drone dataset;

• 3-channel HF coherent receiver signals (from OTH radars) dataset;

• Human Activity Datasets;

• Micro Doppler; and

• ESM dataset.

Riešitelia za Akadémiu ozbrojených síl generála Milana Rastislava Štefánika:

• pplk. Ing. Jozef PERĎOCH, PhD.

• pplk. Ing. Stanislava GAŽOVOVÁ, PhD.

• mjr. Ing. Miroslav PACEK