Detection and extraction of weak signals buried in noise
M. Mohideen Abdul Razak · American Journal of Physics · 2009
The autoregressive method of detection of weak signals from stationary background noise and their time of occurrence and extraction is discussed. The background noise is decomposed into an autoregressive model of finite order. The coefficients in the model are optimized using a least-squares error minimization technique. The coefficients are used to predict the present and future values of the background noise. The predicted noise is subtracted from the original signal. The signal of interest can be easily identified from the subtracted signal, which has a high signal-to-noise ratio compared to the original signal and possesses a large amplitude near the time of occurrence (onset time) of the signal of interest. An example of detection and extraction of a weak synthetic signal buried in noise is discussed.