Statistical and Deep Learning Schemes for Maritime RADAR Detection and Surveillance

Context
Coastal radars aim to control and monitor the maritime surface. By matching transmitted and received electromagnetic waves, radar is able to detect boats whose back-scattered a sufficiently strong signal relative to the sea clutter. The detection performance generally depends on the target’s signature and the clutter noise power, which can be summarized by the Signal to Noise Ratio (SNR). In a rough sea state (e.g., 5 on the Douglas sea scale), the detection performance of small boats hidden by strong sea clutter (Bragg clutter) can deteriorate drastically. The postdoc aims to innovate and improve theoretical detection methods previously developed in SONDRA and L2S laboratories in this context.

Research
The postdoc will first investigate robust detection methods such as Adaptive Normalized Matched Filters (ANMF) [1], which require estimating the covariance matrix of secondary data in a robust manner [2]. The covariance matrix estimation step could include prior information on the structure using either Riemannian geometry [3] or optimization under persymmetric [4], Toeplitz [5], Kronecker constraints [6], etc. This step could be crucial for improving detection and mitigating the probability of false alarms. The second direction investigates deep learning approaches to handle a detection, segmentation or/and generation scheme, either end-to-end or in an unrolling way [7, 8]. The latter approach can be less data-hungry and easier to interpret. Since the available data are complex-valued, an architecture based on Complex-Valued Neural Networks (CVNN) [9] can be exploited to learn data phase information. Meta-learning methods can be investigated to improve detection performance. The developed algorithms will be tested on CSIR database.

Requirements

This position is funded by ANR NEPTUNE 3. We seek a highly motivated postdoctoral fellow to investigate statistical and deep learning methods for detection. The ideal candidate should possess the following qualifications:
• A robust background in machine learning, signal processing, or applied mathematics (statistics, optimization, etc.).
• Strong programming abilities in either Matlab or Python

Supervision team contacts:
Jean-Philippe Ovarlez, SONDRA, CentraleSupelec, jeanphilippe.ovarlez@centralesupelec.fr
Frédéric Pascal, L2S, CentraleSupelec, frederic.pascal@centralesupelec.fr
Mohammed Nabil El Korso, L2S, CentraleSupelec, mohammed.el-korso@universite-paris-saclay.fr
Chengfang Ren, SONDRA, CentraleSupelec, chengfang.ren@centralesupelec.fr

References
[1] J.-P. Ovarlez, F. Pascal, and A. Breloy, “Asymptotic detection performance analysis of the robust adaptive normalized matched filter,” in 2015 IEEE 6th International Workshop on Computational Advances in Multi-Sensor Adaptive Processing (CAMSAP), pp. 137–140, 2015.

[2] F. Pascal, Y. Chitour, J.-P. Ovarlez, P. Forster, and P. Larzabal, “Covariance structure maximum-likelihood estimates in compound gaussian noise: Existence and algorithm analysis,” IEEE Transactions on Signal Processing, vol. 56, no. 1, pp. 34–48, 2008.
[3] A. Collas, Geometrie riemannienne pour l’estimation et l’apprentissage statistiques : application a la teledetection. Theses, Universit´e Paris-Saclay, Nov. 2022.
[4] G. Pailloux, P. Forster, J.-P. Ovarlez, and F. Pascal, “Persymmetric adaptive radar detectors,” IEEE Transactions on Aerospace and Electronic Systems, vol. 47, no. 4, pp. 2376–2390, 2011.
[5] B. Meriaux, C. Ren, M. N. El Korso, A. Breloy, and P. Forster, “Robust-comet for covariance estimation in convex structures: Algorithm and statistical properties,” in 2017 IEEE 7th Inter national Workshop on Computational Advances in Multi-Sensor Adaptive Processing (CAMSAP), pp. 1–5, 2017.
[6] B. M´eriaux, C. Ren, A. Breloy, M. N. E. Korso, and P. Forster, “Matched and mismatched estimation of kronecker product of linearly structured scatter matrices under elliptical distributions,” IEEE Transactions on Signal Processing, vol. 69, pp. 603–616, 2021.
[7] V. Monga, Y. Li, and Y. C. Eldar, “Algorithm unrolling: Interpretable, efficient deep learning for signal and image processing,” IEEE Signal Processing Magazine, vol. 38, no. 2, pp. 18–44, 2021.
[8] N. Arab, Y. Mhiri, I. Vin, M. N. El Korso and P. Larzabal "Unrolled Expectation Maximization for Sparse Radio Interferometric Imaging", EUSIPCO, 2024.
[9] J. A. Barrachina, Complex-valued neural networks for radar applications. Theses, Universite Paris-Saclay, Dec. 2022.