Autonomous Passive Detection For Underwater Unmanned Vehicle

In this paper,we consider the problem of autonomous passive detection based on the underwater unmanned vehicle(UUV)sonar,in which the prior information of the target is absent and model mismatch always exists.Since the acoustic radiation of underwater moving target usually contains two components,i.e.,the wideband noise and the linear component,our work focus on these two components.In order to detect the wideband noise component,a steering angle self-correcting based robust adaptive beamformer is proposed.In the proposed algorithm,we first construct a subspace according to the pre-defined sector of interest,and use the subspace to separate the mismatch vector into two orthogonal components,where one is the projection component in the subspace,and the other is orthonormal to the subspace.Then,we use the subspace constraint and the mainlobe distortion minimization constraint to estimate these orthogonal components,respectively.Finally,the corrected steering vector is applied to calculate the beamformer weights.Theoretical analysis and simulation results show that the steering angle of the proposed beamformer can be self-corrected to the signal direction without prior information.Hence,its performance is almost always close to the optimal value across the whole sector of interest.Furthermore,in order to reduce the computational complexity,the closed-form solution of the proposed algorithm is also derived.For the linear component,a steering-angle and working-band self-correcting-based time-domain broadband beamformer is proposed.The basic idea of the proposed beamformer is to extract the characteristic components of the signal of interest(SOI)from the received array data by an adaptive beamspace transformation matrix first,then employ these characteristic components to reconstruct the signal subspace,and finally construct a set of linearly constrained minimum variance constraints to protect the SOI components.Theoretical analysis and simulation results show that,the steering-angle and working-band of the proposed beamformer are effective to match the SOI without prior information.Hence,its performance is almost always close to the optimal value across the whole region of interest.In addition,the computational complexity of the proposed beamformer is equivalent to the conventional Frost beamformer.When the proposed beamformer is applied to the UUV system,it is expected to significantly improve the detection performance of passive sonar in the absence of target information.For the problem of multiple source detection in a subregion of interest,an efficient algorithm based on subspace matrix filtering is proposed.The basic idea of the proposed algorithm is to project the received array data onto the subspace domain first,then extract the characteristic components of the signals of interest from the estimated signal-plus-interference subspace by a matrix filter,and finally these characteristic components are applied to detect the signal of interest.Compared with some other spatial filtering-based approaches,it has two significant advantages.First,since the power of each source in the signal-plus-interference subspace is normalized,the proposed algorithm is effective to filter out interferences regardless of their strengths.Second,the matrix filter would not reduce the dimension of SOIs,thus,the proposed algorithm is able to distinguish multiple SOIs from the output of the matrix filter.Furthermore,in the proposed algorithm,the matrix filter,the detection threshold,etc.can be pre-designed,and do not require to be adjusted online,these features make the algorithm be very suitable for the UUV sonar system,and be able to work autonomously.To verify the performance of the proposed algorithms,lake experiments were carried out for the case of single and multiple targets,respectively.The results show that both the proposed beamforming algorithms and the multiple source detection algorithm are able to detect the target signal effectively and reliably.