A novel bearing-time record estimation method based on α-stabledistribution modeling

Abstract

Bearing-time record (BTR) is widely used in the field of passive sonar information processing for target tracking. Most existing BTR estimation methods of ship radiated noise are generally based on energy-varying under the assumption of Gaussian modeling in the time domain. However, some weak trajectories may exist in BTR for the decentralized energy distribution and unstable goodness of fit with Gaussian modeling caused by the spatiotemporal instability of the marine environment. In this paper, a novel ship trajectory enhancement method which is based on α-stable distribution modeling theory is presented. In order to make the statistical characteristic of ship radiated noise better reflected, the proposed method utilizes the α-stable distribution modeling in the real part of ship radiated noise Discrete Fourier Transform (DFT) coefficient instead of the Gaussian modeling in the time domain. The scale parameter γ of α-stable distribution is substituted for variance and enhanced BTR is achieved. According to the experimental results, the real part of the DFT coefficient of ship radiated noise could be fitted by α-stable distribution at the significant level of 0.05. Compared with traditional BTR estimation method, the proposed method could gain 3.1dB enhancement on peak signal-to-noise ratio (PSNR).