Compressed Sampling for Neutron ToF Signals Based on SAMP Algorithm

Neutron time-of-flight (ToF) measurement obtains the energy and distribution of neutrons by measuring their ToF at a fixed distance, in which accurately obtaining the flight time of neutrons is critical. However, analog integration + digital peak-seeking faces the problem of low accuracy, and waveform digitizer (WFD) encounters the issues of high sampling rates and massive transmitted data when acquiring neutron ToF signals. Neutron ToF signals have natural sparsity in the time domain, which meets the prerequisite of compressed sampling (CS) theory, and CS has the ability to undersample and reconstruct neutron ToF signals with a sampling rate lower than Nyquist’s theorem to reduce the data volume. Therefore, a new CS framework combined with discrete wavelet transform (DWT), Bernoulli measurement matrix, and sparsity adaptive matching pursuit (SAMP) reconstruction has been proposed in this work. The DWT was adopted for the optimal linear combination of atoms, the Bernoulli random matrix was applied to derive the low-dimensional observed signal from the high-dimensional linear projection, and the greedy SAMP algorithm was used to reconstruct the neutron ToF signal. The performance metrics of the percent residual difference (PRD) and correlation coefficient were employed to quantify the capability of this CS framework. The experimental results presented that the PRD of 6.7348%, the correlation coefficient of 0.9977, and the reconstruction time of 0.1108 s when the sampling rate was 20% of 2.5 Gs/s for the neutron ToF based on an electron-beam-driven photoneutron source. It indicated that the proposed CS framework can be used for the accurate reconstruction of neutron ToF signals, which alleviates the difficulty of storing and processing massive signals.