Detecting NQR signals severely polluted by interference

Abstract

HIGHLIGHTS• The proposed method effectively cancels complicated interference which can be strong, nonstationary, and have frequencies close to that of signal of interest.• The proposed method facilitates a valid detection of the NQR signal severely polluted by interference.• The proposed method performs better than general frequency selective methods of interference cancellation. AbstractNuclear Quadrupole Resonance (NQR) signal detection can be severely obstructed by interference in real life settings, especially when the interference is strong, nonstationary, and its frequencies are close to that of the NQR signal. A novel algorithm is proposed to effectively remove (or reduce) interference components in the data and facilitate a valid detection of the NQR signal. The proposed method exhibits better performance compared to the previously proposed ETAML and FETAML algorithms, when applied to both simulated and measured data. Importantly, the present algorithm directly operates on the original primary data, without requiring any secondary data (NQR signal-free data) for acquiring prior knowledge of the interference.

An advanced beamforming approach based on two-channel echo-train system to cancel interference within an NQR signal resonance spectrum

“…In such situations of strong interference, Fourier-based spectrum analysis and frequency selective methods [8,11] can deal successfully with interference that does not overlap significantly with the NQR signal's resonant frequency spectrum. In cases of strong overlaps, algorithms which include an interference cancellation (IC) step must be applied to improve NQR detection, as in our previous work in [12]. This algorithm, however, will inevitably affect the NQR signal if the interference is centred within the NQR band (i.e., the interval containing all possible values of the NQR signal resonant frequency under given environment conditions).…”

“…This work, however, indirectly requires that the NQR signal subspace and the interference subspace should be orthogonal, which can not be satisfied when the frequencies of the NQR signal and the interference are the same or very close to each other. (For example, considering the continuous Hilbert space on t∈[0, T ], Subspace "cos( 2πmt T + φ 1 )" and Subspace "cos( 2πnt T + φ 2 )" is not orthogonal if m = n. Due to the same fact, we do not apply our IC method [12] inside the NQR band in case of cancelling the NQR signal.) Besides, this work also faces the problem of calibration errors.…”

“…Then it precisely estimates the amplitude, frequency, and phase of the interference, which is robust to the calibration errors (gain errors). Combined with our previous IC technique [12], it can cancel interference in the whole frequency domain. Subsequently, the "echo-train" approximate maximum likelihood (ETAML) algorithm [8] can be applied to ensure a valid NQR detection strategy.…”

“…Since the signal of interest is within the NQR band, applying ETAML algorithm only to frequency information within the NQR band should be enough and discards any interference outside (but without overlapping with) the NQR band [8,12]. Selecting this frequency information can be done by dividing the NQR bands in a vector of J subbands [ f s1 f s2 ... f sJ ] and then performing a DFT for Z N M and Q N M which yields…”

“…The FETAML algorithm, unfortunately, can only deal with some interference centered distant from the NQR band. For this reason, we have proposed an efficient IC method for cases with strong and even slightly time-varying interference overlapping with the NQR band [12]. This algorithm, however, do not deal with interference centered inside the (narrow) NQR band, which may occur in some real-life settings.…”