Abstract:
We consider multipath mitigation for multi-antenna GNSS receivers using the maximum likelihood (ML) principle and Newton's method (NM). If good initial estimates for the ...Show MoreMetadata
Abstract:
We consider multipath mitigation for multi-antenna GNSS receivers using the maximum likelihood (ML) principle and Newton's method (NM). If good initial estimates for the parameters of all multipath components are available, NM is a very effective tool to find the global optimum of the ML cost function and, in particular, NM yields a very accurate estimate for the delay of the line-of-sight path. By introducing a finite grid for the parameters, algorithms from the field of sparse recovery, e.g., the orthogonal matching pursuit (OMP) algorithm, can be used as initialization schemes. In this paper, we propose two extensions of the OMP algorithm that use grid-less refinement steps, which are themselves based on NM applied to marginals of the ML objective. As numerical simulations show, initializing NM using OMP with grid-less refinement steps improves the estimation performance in the high signal-to-noise power ratio (SNR) regime if compared to, e.g., initialization using standard grid-based OMP or the SAGE algorithm.
Published in: 2017 IEEE 7th International Workshop on Computational Advances in Multi-Sensor Adaptive Processing (CAMSAP)
Date of Conference: 10-13 December 2017
Date Added to IEEE Xplore: 12 March 2018
ISBN Information: