Brigitte Bidégaray-Fesquet — Non-uniform signal processing

Partners

LJK Brigitte Bidégaray-Fesquet
TIMA Laurent Fesquet, Gilles Sicard, and Taha Beyrouthy,
and formerly Fabien Aeschlimann, Saeed Mian Qaisar
Gipsa-Lab Carlos Canudas de Wit and Nicolas Marchand

Previous work

Most electronic systems have a clock which synchronises everything. A numerical electronic signal is therefore usually synchronous and corresponds to data at times which are all multiples of a unique sampling time. Classical signal processing is strongly based on this property.

The CIS team of the TIMA laboratory in Grenoble develops asynchronous (i.e. clock-less) processors for which samples are collected each time some event occurs, such as level-crossing of an analog signal. For these asynchronous processors, all the usual tools of microelectrnics have to be rebuilt, in particular signal processing chains.

For a few years there has been a first informal collaboration between the LJK and TIMA laboratory on non-uniform signal processing. For the years 2009–2010 we have the support of the M-STIC pole of the Joseph Fourier University: TATIE project (Technologie Asynchrone et Traitement des signaux Irrégulièrement Echantillonnés). We have a certain number of common results:

  • the application of usual IIR filters to non-uniform signals: [FB10], following Fabien Aeschlimann PhD [Aes06],
  • the design of asynchronous filters (sampled in frequency): [BF09], [BF11a], [BF11b],
  • the analysis of low-consumption algorithms: [FSB10],
  • a Matalab toolbox SPASS (Signal Processing for ASynchronous Systems) which implements all the algorithms we have developed.

Ongoing work

We are currently working on

  • a new more robust version of the SPASS toolbox,
  • the comparison of fully non-uniform algorithms with resampling techniques of active parts of a sporadic signal developped in Saeed Mian Qaisar's PhD [Qai09],
  • the analysis of the reconstruction of a non-uniform signal,
  • the extension of mathematical results to level-crossing sampling problems in control, in collaboration with the Gipsa-Lab in Grenoble.

References

[Aes06] Fabien Aeschlimann. Traitement du signal échantillonné non uniformément : algorithme et architecture. PhD thesis, INP Grenoble, February 2006.
[BF09] Brigitte Bidégaray-Fesquet and Laurent Fesquet. A fully non-uniform approach to FIR filtering. 8th International Conference on Sampling Theory and Applications (SampTa'09), Marseille, France, 2009.
[BF11a] Brigitte Bidégaray-Fesquet and Laurent Fesquet. Non-uniform filter interpolation in the frequency domain. To be published in Sampling Theory in Signal and Image Processing, 2011.
[BF11b] Brigitte Bidégaray-Fesquet and Laurent Fesquet. Non-uniform design of FIR filters. in preparation, 2011.
[FB10] Laurent Fesquet and Brigitte Bidégaray-Fesquet. IIR Digital Filtering of Non-uniformly Sampled Signals via State Representation. Signal Processing, 90, 2811–2821 (2010).
[FSB10] Laurent Fesquet, Gilles Sicard, and Brigitte Bidégaray-Fesquet. Targeting ultra-low power consumption with non-uniform sampling and filtering. IEEE International Symposium on Circuits and Systems (ISCAS2010), Paris, France. 3585–3588, IEEE, 2010.
[Qai09] Saeed Mian Qaisar. Échantillonnage et traitement conditionnés par le signal : une approche prometteuse pour des traitements efficaces à pas adaptatifs. PhD thesis, INP Grenoble, May 2009.