Doa Estimation of Nonparametric Spreading Spatial Spectrum Based on Bayesian Compressive Sensing Exploiting Intra-Task Dependency

Si Qin,Qisong Wu,Yimin D. Zhang,Moeness G. Amin

DOI: https://doi.org/10.1109/icassp.2015.7178401

2015-01-01

Abstract:For spatially distributed targets encountered in radar and sonar applications, direct application of subspace-based methods usually do not lead to an accurate estimation of the direction and angular extent of the signal arrivals. If the spatial distribution of the targets can be parameterized with a known model a priori, the direction-of-arrival (DOA) estimation problems can be simplified as parameter estimation problems. However, these methods do not apply when the targets are not parameterizable. Motivated by this fact, we propose an effective approach for the DOA estimation of nonparametric spatially extended targets. In the proposed approach, the spatially extended targets are modeled as a continuous sparse structure, which are effectively estimated using the Bayesian compressive sensing techniques based on a paired spike-and-slab prior accounting for the angular target spread. In particular, the problem is examined under a collocated multiple-input multiple-output (MIMO) radar platform. Signal transmission at multiple coprime transmit frequencies are also considered to achieve increased degrees-of-freedom. The group sparsity of the targets across different frequencies is exploited to achieve improved DOA estimation performance.