Direction of Arrival Estimation via Approximate Message Passing
Chunli Guo, James D. B. Nelson · 2015
In this paper, the conventional direction-of-arrival (DOA) problem is revisited by considering the recently proposed expectation maximization approximate message passing (EM-AMP) algorithm. The EM-AMP approximates the sparse DOA prior with the Bernoulli-Gaussian distribution to explore both the signal sparsity and the pdf information for the bearing estimation. Tests are performed on both synthetic and real acoustic experiment data with a partially oversampled discrete Fourier transform sensing matrix. Simulations with EM-AMP demonstrate practicable performance while circumventing the regularizor selection problem suffered by the Iι minimization approach, which makes it a useful tool for practical DOA estimation.