Weak signal sensing using empirical mode decomposition and stochastic data reordering

Arnab Roy, John F. Doherty

Research output: Chapter in Book/Report/Conference proceedingConference contribution

4 Scopus citations

Abstract

A weak signal sensing technique that exploits the dyadic filter-bank property of the empirical mode decomposition (EMD) technique for noise-dominated signals is presented in this paper. The EMD procedure decomposes wideband noise into a series of constituents with linearly decreasing mean energy and mean frequency on the logarithmic scale, and the energy of the appropriate modes, corresponding to frequency subbands, constitutes our signal feature. A weak stochastic signal contributes to the energy of certain modes that correspond to the frequency content of the signal, which can be used to detect their presence. The effectiveness of this technique is further enhanced via local stochastic reordering of the original data samples that generates new noise realizations while affecting the stochastic part negligibly. Averaging the signal features obtained using the nonlinear decomposition over multiple reordering realizations improves signal detection reliability. This novel application of the EMD procedure is described in this paper and its performance compared against standard signal detection techniques.

Original languageEnglish (US)
Title of host publication2010 Military Communications Conference, MILCOM 2010
Pages37-41
Number of pages5
DOIs
StatePublished - Dec 1 2011
Event2011 IEEE Military Communications Conference, MILCOM 2011 - Baltimore, MD, United States
Duration: Nov 7 2011Nov 10 2011

Publication series

NameProceedings - IEEE Military Communications Conference MILCOM

Other

Other2011 IEEE Military Communications Conference, MILCOM 2011
CountryUnited States
CityBaltimore, MD
Period11/7/1111/10/11

All Science Journal Classification (ASJC) codes

  • Electrical and Electronic Engineering

Fingerprint Dive into the research topics of 'Weak signal sensing using empirical mode decomposition and stochastic data reordering'. Together they form a unique fingerprint.

Cite this