Detection of underwater moving object based on the compressed sensing

Abstract

This paper proposes method of detecting the motion state of underwater targets based on compression sensing. A Linear frequency modulation signal is influenced by the moving state of the target under test, and its echo parameters such as the initial frequency, frequency modulation rate, and phase, will change according to the moving state of the target. Firstly, this method uses the characteristics of the high order LFM Chirplet Transform matrix, which has the bending effect, and energy accumulation in the time-frequency domain, in order to sparse the linear frequency modulated echo signal. Secondly, based on compression sensing, the characteristic parameters of an echo signal, such as the initial frequency and frequency modulation rate, have been reconstructed. At the same time, the interference by background noise in the underwater acoustic channel is eliminated. As a result, we can determine the motion state of an underwater object according to the physical characteristics of the linear frequency modulation signal echo. Simulations and experiments show that the higher order Chirplet Transform has very high resolution without cross-term inference, and is suitable for analyzing non-stationary underwater acoustic signals. After obtaining the characteristics of the time-frequency of an echo signal, the main characteristics of the data are extracted by compressed sensing based on the Noiselets matrix, and the noise interference from the underwater acoustic channel is eliminated. This technique can improve measurement of the physical parameters of underwater moving targets, and has a high detection probability under low SNR, so the validity of the theoretical analysis has been proved.