Skip to Main content Skip to Navigation

Presentation

The research activities of the “Signal & Communications” group are currently divided into two main areas:

  • Statistical signal & image processing: analysis and representation, inverse problems and computational imaging, machine learning
  • Signal for communications: waveform design, digital receivers, communication system modeling and performance.

Statistical signal & image processing

Prominent research activities in statistical signal & image processing conducted by the SC group are currently dedicated to the following research topics:

  • Signal & image analysis and representation
    • non-uniform sampling and reconstruction
    • spectral analysis: empirical mode decomposition, AM/FM decomposition
    • time-frequency analysis: reassignment and synchrosqueezing
    • multiscale, multi-resolution and multifractal analysis: scale invariance, self-similarity, local regularity, texture analysis
  • Inverse problems and computational imaging
    • restoration: denoising, deconvolution
    • single-image super-resolution
    • fusion of multi-band images (e.g., multi- and hyperspectral pansharpening)
  • Machine learning for signal & image processing
    • blind source separation: independent component analysis (ICA), nonnegative matrix factorization (NMF)
    • representation learning: dictionary learning, subspace learning, adaptive and fast transforms
    • structured modeling: sparse, anti-sparse and low-rank models

These different problems are tackled using various methodological and theoretical tools such as:

  • Bayesian modeling and computation
    • (hierarchical) Bayesian inference
    • latent variable models, hidden Markov models, Markov random fields
    • Bayesian nonparametrics: Beta, Dirichlet and Gaussian processes
    • Markov chain Monte Carlo (MCMC) algorithms: Metropolis-Hastings, (partially collapsed) Gibbs samplers, reversible jump algorithms, Hamiltonian Monte Carlo, proximal-MALA
    • approximate Bayesian computations (ABC), variational Bayesian methods
  • Deterministic and stochastic optimization
    • splitting and proximal algorithms
    • augmented Lagrangian methods
    • majorization-minimization algorithms
    • Monte Carlo integrations and approximations

These research activities are carried out in various applicative fields, namely:

  • Geoscience and remote sensing
    • hyperspectral imaging: spectral unmixing, nonlinearity and variability detections, segmentation/classification, fusion (e.g., pansharpening)
    • multi-temporal image analysis: change detection, unmixing
    • conventional and delay/Doppler altimetry: modeling and parameter estimation
  • Biomedical imaging and engineering
    • segmentation and unmixing of 3D+t TEP images
    • ultrasound imaging: fast acquisition and compressed sensing, segmentation/classification
    • P- and T-waves detection and delineation in ECG
    • source localization in EEG
    • hemodynamic brain parcellation in fMRI
    • multiscale and multifractal analysis of biomedical signals (fMRI, ECG)
    • gene expression factor analysis: identification of disease signature (e.g., influenza)
  • Diagnosis and condition monitoring
    • detection and diagnosis of mechanical and bearing faults in electrical machines
    • processing of anemometric and inertial data in flight control systems
    • aging modeling for prognosis in electrical engineering
  • Audio and music processing
    • music signal decomposition
    • multichannel source separation
    • audio restoration
    • music chord recognition
  • Microscopy and spectroscopy
    • EELS spectrum-image analysis
    • spectrochemical analysis
    • magnetic resonance force microscopy (MRFM) imagery
  • Art investigation and cultural heritage
    • audio-guided video synthesis for modern art
    • texture analysis and classification of photographic prints

Signal for digital communications

Research activities in digital communications and systems addressed by the SC group are focused on signal processing, modeling and coding for satellite and wireless communication systems.

These methods are applied to waveform desgin, coding, multi-user communications/multiple access/cooperative communications, synchronization and equalization.

  • Modulation and coding
    • block-based circular faster-than-Nyquist waveform design
    • faster-than-Nyquist modulation in the non linear regime
    • single carrier spatial modulation schemes in the frequency domain for a PAPR/performance trade-off
    • sparse graph-based coded continuous phase modulations (CPM) in coherent and non coherent channels
    • BICM schemes for nonlinear code-shift keying modulation
    • Structured coding schemes for non-ergodic channels
    • compressed sensing approach for indoor and outdoor localization of users over multipath large scale MIMO channels
    • Spatial coupling for serially concatenated systems
    • Waveform design and coding for IoT (BPSK, code-shift keying modulation, Chirp Spreaded modulation, FSK based…)
    • MIMO Radar
  • Multiple access and random access
    • Methods for Random Access: MARSALA, MUSCA and SPOTiT schemes.
    • IoT by satellite
  • Identification, equalization and detection
    • iterative schemes for sparse graph-based coded continuous phase modulation (CPM)
    • iterative detection of bit-interleaved code modulation system over frequency selectif channels
    • low-complexity frequency domain detection and synchronization methods
    • unrolled deep expectation-propagation (EP) detection
    • expectation/Belief-propagation (EP/BP) based detection schemes
    • detection in nonlinear channels using mutual information optimized neural networks (NN)
    • detection in nonlinear channels using linear low complexity MAP based or MMSE based detector for Nyquist and Faster than Nyquist
    • classification of modulations
  • Multicarrier systems and filterbanks
    • orthogonal frequency-division multiplexing (OFDM)
    • multicarrier code division multiple access (CDMA)
  • Cooperative communications
    • stochastic geometry for interference modeling

The applications, in the general frameworks of wireless and satellite communications, are mainly related to:

  • Satellite communications
    • telemetry coded waveforms for the next Ariane spacecraft generation
    • digital broadcasting
    • IoT by satellite using Low Earth Orbit (LEO) constellations
  • Wireless communications
    • unmanned aerial vehicle (UAV)
    • UAV-based mobile targets tracking
    • generation and management of adaptive UAV trajectories
    • internet-of-things (IoT)
    • ad-hoc networks
    • vehicular networks
    • Navigation
    • global navigation satellite system (GNSS)