Signal processing

The group has a solid experience in techniques for statistical signal processing and time series analysis. Apart from our background with classic tools such as ARMA models, array beamforming, or adaptive filters, we have actively developed in the recent years new algorithms based on state-of-the-art kernel and Bayesian learning.  Among other applications, we have successfully applied our designs in acoustics, communications, forecasting, and wireless networks.

Specific research lines of the group in this direction include:

  • Improved Adaptive Filtering via Adaptive Combination
    • Experimental and theoretical characterization of algorithms for combining adaptive filters
    • Alpha-stable robust signal filtering
    • Self-adjustable echo cancelers robust to SNR and non-linearities
    • Blind equalization
  • Distributed estimation in wireless networks
  • Kernel-based signal processing tools
    • Non-linear on-line novelty detection
    • Kernel ARMA modelling
    • Kernel Adaptive filtering
    • Kernel array processing
  • Bayesian signal filtering