Particle Filter Algorithm for Underwater Acoustic Source DOA Tracking With Co-Prime Array

In this paper, the problem of direction of arrival (DOA) tracking based on the co-prime array is studied. Owing to low signal-to-noise ratio (SNR) in underwater environment, making estimating DOA of a moving source become a challenging problem. In this work, a framework of particle filter (PF) with sequential importance resampling (SIR) is developed to address this problem. The involved particles are generated from the uniform motion model, and then weighted by their measurement likelihood function. By vectorizing the autocorrelation matrix of the output data from the co-prime array, a virtual uniform linear array with extended array aperture can be obtained. Applying spatial smoothing technique to the au-correlation matrix of the virtual uniform linear array, the MUSIC pseudo-spectrum can be formulated and then used as the likelihood function. Due to the enlarged array aperture and the super resolution property of MUSIC, the involved particles can be updated more appropriately. Simulation studies are presented to verify the validity of the proposed method.