Fast Algorithm for Bayesian DOA Estimator Based on Metropolis-Hastings Sampling

摘要

Bayesian estimator is known to have the best performance in DOA estimation of narrowband sources. However, it is computationally intensive. In order to reduce its computational complexity, Metropolis-Hasting Sampler, one of the most popular of Markov Monte Carlo methods, is employed to combine with it to propose a novel method called Bayesian DOA Estimator based on Metropolis-Hasting Sampling (MHB). In this method the power of the MHB spectrum function is viewed as the target distribution up to a constant proportionality, which is sampled by Metropolis-Hasting Sampler. Simulations show that MHB not only keeps the excellent performance of Bayesian estimator but also reduces computational cost remarkably.