A novel wavelets method for cancelling time-varying interference in NQR signal detection

Abstract

Interference cancelation is a very important aspect of Nuclear Quadrupole Resonance (NQR) signal detection, and can become really difficult when the interference is considerably time-varying. We propose a novel wavelets method to effectively remove (or reduce) time-varying interference in the data and facilitate a valid detection of the NQR signal. The proposed algorithm uses an extended Gabor-Morlet wavelets basis to approximate interference with complicated time-varying properties. The proposed algorithm utilizes our well designed cost function to extract the interference components out from NQR data strategically.Mathematical derivations and numerical results from both simulated and measured data demonstrate that the proposed algorithm can precisely cancel strong time-varying interference without distorting signal of interest improving NQR detection, even when interference and signal of interest are severely overlapped.The proposed algorithm is beyond normal wavelets methods such as standard wavelets denoising methods, and exhibits better performance than normal Fourier analysis and related frequency selective methods and adaptive filtering methods.

Detection of extremely weak NQR signals using stochastic resonance and neural network theories

“…Alternatively, it could be trained with experimental data in controlled laboratory conditions before being tested in realistic practical applications where low SNRs pose a great detection challenge, such as humanitarian demining. Interference may be a severe problem in real-life NQR detection [9,10]. It is therefore necessary to train a network which can deal with strong interference in addition to low SNRs, adding another level of complexity relative to detecting "NQR signal + noise" cases.…”

“…These parameter estimation algorithms proposed in recent years are to some extent robust to low SNR cases, but suffer from the effect of strong radio frequency (RF) interference [7,8]. Our recent work on this topic has proposed significant improvements to enhance detection in cases of strong RF interference [9,10].…”