Phase Unwrapping by Sequential Bayesian Filters with Circular Statics and Trend

Hiroaki UMEHARA  Masato OKADA  Yasushi NARUSE  

A - Abstracts of IEICE TRANSACTIONS on Fundamentals of Electronics, Communications and Computer Sciences (Japanese Edition)   Vol.J103-A   No.8   pp.156-164
Publication Date: 2020/08/01
Online ISSN: 1881-0195
Type of Manuscript: PAPER
ranging,  positioning,  integer ambiguity,  angular statistics,  maximizer of posterior marginal,  

Full Text(in Japanese): PDF(677.6KB)>>
Buy this Article

The inference model of phase unwrapping from the carrier wave is improved for realization of precise positioning even with unidirectional movement for a certain period of time. A sequential Bayesian filter is constituted by maximum posterior marginal inference with a trend component model for the prior probability distribution and a circular statistics for the likelihood function. The marginal posterior distribution is formulated by the multiple dimensional distribution with the number which is equal to the degree of order for the assumed trend component model, and later the degree is determined by the maximum likelihood estimation. For the numerical experiments by using the artificial data, the Monte-Carlo filter (the particle filter) is set up for the marginalizing computation of the probability variables. The assumed degree of order is properly selected and the precision of estimation with the proper selection is statistically higher than the ones with the assumption of the other degrees. The presented model provides a flexible phase unwrapping framework with noise reduction which is applicable to the stochastic drift.