Higher-order statistics to detect and characterise termite emissions
Juan José González de la Rosa, I. Lloret, Carlos G. Puntonet, J. M. Górriz · Electronics Letters · 2004
An independent components analysis algorithm is applied to extract vibratory signals generated by termites from background noise. Signals from a microphone in uniform and Gaussian noise, were taken as sensor outputs. Bispectrum is proposed as a higher-order statistic to characterise time series. Experiments prove that detection and characterisation can be performed successfully even with low signal-to-noise ratio signals.