DOA Estimation Fast Algorithm for Short Sampling Wideband Sources Based on Metropolis-Hastings Sampling

Approximated Maximum Likelihood estimator (AML) produces quite accurate DOA estimation of wideband sources with short time sampling. However, this solution is costly due to its high computational complexity. For this reason, Markov Monte Carlo method is combined with Approximated Maximum Likelihood DOA estimator to introduce a computationally efficient algorithm. A novel Approximated Maximum Likelihood DOA Estimator based on Metropolis-Hasting Sampling (MHAML) is proposed. In this method, the power of the AML spectrum function is regarded as the target distribution up to a constant proportionality, and Metropolis-Hasting sampler is used to sample from it. Simulations show that MHAML reduces computational burden remarkably without compromising the excellent performance of AML.