Measurement of extreme weak signal embedded in large non-stationary background typical of earth's magnetic field

Abstract

This work presents an alternative approach for the detection of extremely weak signals embedded in large non-stationary backgrounds such as the stochastic nature of earth's field. The purpose of this work is to testify the use of signal's non-stationarity as a mean to enhance detection of weak signals that are embedded in background signal/noise <−40 dB ranges. To illustrate the technique, audio sound tracks are used as example source signals. Present results indicate that the use of non-stationarity improves separation ability by about 10 dB and successful source separations can be achieved for weak signals down to −80 dB. Unlike conventional measurement techniques, this source separation algorithm utilises no filter at the input end and hence all information will be measured and subsequently separated for further analysis. This approach will help to search for new phenomenon, as no information is lost during data requisition. This paper also hopes to provide alternative measurement techniques to the experimentalists who are involving measurements of low-level signals such as those engaging research in 2D/1D magneto-transport or magneto-optical phenomena, bio-magnetic and geo-magnetic research, etc., and to avoid using conventional filtering techniques that can create distortions and artefacts.