Identifying different types of stochastic processes with the same spectra Conference Paper uri icon

abstract

  • We propose a new way of pattern recognition which can distinguish different stochastic processes even if they have the same power density spectrum. Known crosscorrelation techniques recognize only the same realizations of a stochastic process in the two signal channels. However, crosscorrelation techniques do not work for recognizing independent realizations of the same stochastic process because their crosscorrelation function and cross spectrum are zero. A method able to do that would have the potential to revolutionize identification and pattern recognition, techniques, including sensing and security applications. The new method we are proposing is able to identify independent realizations of the same process, and at the same time, does not give false alarm for different processes which are very similar in nature. We demonstrate the method by using different realizations of two different types of random telegram signals, which are indistinguishable with respect to power density spectra (PDS). We call this method bispectrum correlation coefficient (BCC) technique.

name of conference

  • Noise and Information in Nanoelectronics, Sensors, and Standards III

published proceedings

  • Noise and Information in Nanoelectronics, Sensors, and Standards III

author list (cited authors)

  • Kim, J. U., Kish, L. B., & Schmera, G.

citation count

  • 0

complete list of authors

  • Kim, JU||Kish, LB||Schmera, G

editor list (cited editors)

  • Bergou, J. A., Smulko, J. M., Dykman, M. I., & Wang, L.

publication date

  • January 2005