Underwater multi-target passive acoustic localization based on multi-sensor collaboration

In the problem of multi-sensor passive acoustic localization, for a specific target, the target signals received from different sensors all originate from the same target position and thus are intrinsically correlated. Based on this physical foundation, a particle filtering-based passive acoustic localization technique is proposed to effectively integrate data from multiple sensors and thereby improve localization performance. The proposed method defines the likelihood function of the particle filter as the product of the output of the cross-correlation between the signal of different sensors conditioned on the state of the particle. This likelihood function is designed to ensure that the processing gain of multi-sensors can be fully obtained. Moreover, since the proposed method is free from the traditional localization paradigm, it can circumvent the measurement-to-track associated problem faced by the traditional localization paradigm. The lake experiment indicates that under the condition of strong interference, the average localization error of the traditional localization method is 7.2 m, while the proposed method performs better with the average localization error being 1.2 m.