Abstract
In this paper, improved Bernoulli filtering methods are developed to deal with the problem of joint passive detection and tracking of an underwater acoustic target with multiple arrays. Three different likelihood calculation methods based on local beamforming results are proposed for the Bernoulli filter updating. Firstly, multiple peaks, including both mainlobe and sidelobe peaks, are selected to form the direction-of-arrival (DOA) measurement set, and then the Bernoulli filter is used to extract the target track. Secondly, to make full use of the informations in the beamforming output, not only the DOAs but also their intensities, the beam powers are used as the input measurement sets of the filter, and an approach based on Pearson correlation coefficient (PCC) is developed for distinguishing between signal and noise. Lastly, a hybrid method of the former two is proposed in the case of fewer then three arrays. The tracking performances of the three methods are compared in simulations and experiment. The simulations with three distributed arrays show that, compared with the DOA-based method, the beam-based method and the hybrid method can both improve the target tracking accuracy. The processing results of the shallow water experimental data collected by two arrays show that the hybrid method can achieve a better tracking performance.
Highlights
In many target detection and tracking applications, detection and estimation are usually carried out separately under the Bayes framework [1]
The horizontal line arrays (HLA) are drawn as black dots in the figure
We studied the passive joint detection and tracking problem of underwater targets using multiple arrays
Summary
In many target detection and tracking applications, detection and estimation are usually carried out separately under the Bayes framework [1]. Hypothesis tests based on the Neyman–Pearson or Bayes criterion [2] are applied to detect the existence of the target. Sequential probability ratio test (SPRT) which makes full use of all collected data based on the continuity of the target state [3]. The estimators start working after the target has been declared to be detected. Examples of this type of sequential estimators include the Kalman filter (KF) [4], the extended Kalman filter (EKF) [5], the unscented Kalman filter (UKF) [6], and the particle filter (PF) [7]. The PF is a sequential Monte Carlo (SMC) method that uses a random particle system (states and weights) to approximate the relevant probability distribution. Gaussian sum particle filter is a special kind of particle filter easier to implement where the relevant probability distribution is assumed as the sum of Gaussian distributions [11]
Sign-in/Register to view highlights & summary Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
