Application of the reversible jump Markov chain Monte Carlo method to real-life propagation measurements

Abstract

This paper shows the application of the reversible jump Markov chain Monte Carlo (RJMCMC) method to a joint detection-estimation angle-of-arrival (AOA) problem, with simulations and processing of real-life propagation measurements. A minimum number of assumptions are made as the number of sources impinging the array, their AOA and amplitude, as well as the noise variance, are considered random variables. The algorithm jointly estimates all these parameters by sampling the posterior distribution that is made up of the union of disjoint subspaces of different dimensions. We apply the RJMCMC technique to simulations and to real data to jointly detect and estimate the parameters and the noise characteristics. We also present an approach, using space projection, which avoids the manipulation of the nuisance parameters.