Overview

Independent Component Analysis (ICA) is a technique to project a set of signals into another set, or components, as statistically independent as possible. When the original signals or observations are known to be a mixture of independent sources, ICA can be applied to separate them, recovering the sources. We say that ICA can be used to solve the blind source separation (BSS) problem.

Our group has been working on the development of novel approaches to solve the ICA problem. In particular, we focused on the solving of the minimization of the entropy at the output, as ICA criterium. We approximated it by using moments up to fourth order. Then proved that, under mild conditions, in the two dimensional case the criterium reduces to a sinusoid, easily optimize. We also focused on the extension to the higher dimensional problem.

We found that ICA applied to BSS could be used in several communications problems, such as CDMA and OFDM, proposing successful solutions.

Finally, we first proposed ICA as a watermarking algorithm and applied it to images. The algorithm was based on the use of ICA to find a valid projection of an image. This projection acts as a key of the algorithm. The image to be watermarked is projected using this basis, and the watermark embedded into the components with larger energy.

Involved Members

Juan José Murillo Fuentes, Rafael Boloix Tortosa, Francisco J. Simois-Tirado

Code

Matlab Code for the SICA algorithm. It includes also the minimization of the General Weighted Estimator (GWE). Optimized Jacobi Optimization is used. Real case is considered here.

Function: ogwe.m

Demo: demo.m

Read me file: readme.txt

Publications

Journals

  • F.J. Simois, J.J. Murillo-Fuentes, R. Boloix-Tortosa, L. Salamanca (2012), ”Near the Cramér-Rao Bound Precoding Algorithms for OFDM Blind Channel Estimation”, IEEE Transactions on Vehicular Technology, Vol. 61, N. 2, Pp. 651 – 661, Feb. 2012. paper

  • J. J. Murillo-Fuentes and R. Boloix-Tortosa (2010). ”Strict Separability and Identifiability of a Class of ICA Models”. IEEE Signal Processing Letters, Vol. 17, N. 3, Pp. 285 – 288. paper (the version in ieeexplore has important typos)

  • Juan. J. Murillo-Fuentes (2007). ”Independent component analisis in the blind watermarking of digital image”, Neurocomputing Vol. 70, Pp. 2881-2890, 2007. paper

  • Vicente Zarzoso, Juan J. Murillo-Fuentes, Rafael Boloix-Tortosa, Asoke K. Nandi (2006). ”Optimal pairwise fourth-order independent component analysis”. IEEE Transactions on Signal Processing, vol. 54, pp. 3049-3063. 2006. paper (See code section)

  • J.J. Murillo-Fuentes (2005). ”Robust Blind Image Watermarking with independent component analysis: an embedding algorithm”. Lectures notes on computer science, vol. 3512, pp. 1100-1107, 2005. paper

  • J.J. Murillo-Fuentes (2004). ”Independent component analysis in the watermarking of digital images”Lectures Notes on Computer Science, vol. 3195, pp. 938-945, 2004. paper

  • R. Boloix-Tortosa, J.J. Murillo-Fuentes (2004). ”Blind Source Separation in the Adaptive Reduction of Inter-Channel Interference for OFDM”. Lectures notes on computer science, vol. 3195, Pp. 1142-1149, 2004. paper

  • Juan J. Murillo-Fuentes and Francisco J. González-Serrano (2004).’’A sinusoidal contrast function for the blind separation of statistically independent sources’’, IEEE Transactions on Signal Processing. vol. 52, n 12, pp. 3459-3463, 2004. paper

  • Antonio J. Caamaño, Rafael Boloix-Tortosa, Javier Ramos and Juan J. Murillo-Fuentes (2004). ‘’Hybrid Higher Order Statistics Learning in Multiuser Detection’’. IEEE Transactions on System, Man and Cybernetics, Part C. vol. 34, n. 4, pp. 417-423, 2004. paper

  • J.J. Murillo-Fuentes and F. J. González-Serrano (2000). ”Improving stability in blind source separation with the stochastic median gradien”. IEE Electronic Letters, V. 36, Nº 19, pp 1662-1663, September 2000. paper

  • Juan. J. Murillo-Fuentes and Francisco J. González-Serrano, ”Median equivariant adaptive separation via independence: application to communications” Neurocomputing. Vol 49, issue
    1-4, pp 389-409. 2002. paper

International Congress

  • J J. Murillo-Fuentes; R Boloix-Tortosa; S. Hornillo-Mellado, V. Zarzoso (2004). ”Independent Component Analysis Based on Marginal Entrophy Approximations”, In Proc. of the ISIAC International Symposium on Intelligent Automation and Control (WAC) 2004. paper

  • Juan J. Murillo-Fuentes, Rafael Boloix,Francisco J. González Serrano (2003), ”Inititalized Jacobi Optimization in Independent Component Analysis”. In Proc. of the Congress on Independent Component Analysis (ICA). Nara (Japan), Pp. 1053-1058, Apr 2003. paper

  • Juan J. Murillo-Fuentes, Rafael Boloix, Francisco J. González Serrano (2003), ”Adaptive initialized jacobi optimization in independent component analysis”. In Proc. of the Congress on Independent Component Analysis (ICA). Nara (Japan), Pp. 1065-1070, Apr 2003. paper

  • Antonio Caamaño and Juan J. Murillo-Fuentes and Francisco J. González-Serrano and Javier Ramos, ”Natural Gradient Based Multiuser Detection”. In Proc of the IEEE The 13
    th IEEE International Symposium on Personal, Indoor and Mobile Radio Communications PIMRC 2002    . September 15-18, 2002 Lisbon, Portugal. paper

  • Antonio J. Caamaño, Rafael Boloix, Tortosa y Juan J. Murillo Fuentes (2002), ”Centralized blind multiuser detection using SICA”, In Proc. of Advances in Multimedia Communication, Information Processing and Education (Learning). Leganés (Spain). paper

  • J.J. Murillo-Fuentes and F. J. González-Serrano (2000). ”Higher Order Moments Algorithms for Blind Signal Separation”, In Proc. of the Congress on Independent Component Analysis (ICA), pp 345-350, Helsinki, Jun 2000.  paper

  • J.J. Murillo-Fuentes, F.J. González-Serrano (2001). ”Independent component analysis with sinusoidal fourth-order contrasts”, In Proc. of the IEEE Int. Conf. on Acoustics, Speech and Signal Processing (ICASSP). Pp. 2785-2788. IEEE Press.IEEE. Salt Lake City (USA), 2001. paper

  • F.J. González-Serrano, H.Y. Molina-Bulla, J.J. Murillo-Fuentes (2001). ”Independent component analysis applied to digital image watermarking”. In Proc. of the IEEE Int. Conf. on Acoustics, Speech and Signal Processing (ICASSP). IEEE Press. Pp. 1997-2000. Salt Lake City (USA), 2001. paper

  • J.J. Murillo-Fuentes, M. Sánchez-Fernández, A. Caamaño-Fernández and F.J. González-Serrano (2001). ”Apdative blind joint source-phase separation in digital communications”. IEEE International Conference on Communications (ICC). IEEE. Helsinki, Finland. Proc. IEEE International Conference on Communications. Pp. 930-934, 2001. paper

Acknowledgements

These results were possible thanks to public funding. The Universidad Carlos III de Madrid and the Universidad de Sevilla trusted us to carry out this research as professor. The Spanish Government and the European Union (FEDER) also founded this research through the projects MEC.CICYT.TIC 2003-03781 and TEC2006-13514-C02-2/TCM.

USfeder

Copyright Notice

The material above is presented here to ensure dissemination of scholar and technical work by the author. Copyright and all rights are retained by authors and other copyright holders. Accordingly, all persons using this material are expected to adhere to the terms and constraints invoked by each holder’s copyright. Notice that, in most cases, these works may not be reposted without the explicit permission of the copyright holder.