Distributed Extended Kalman Filter for Position, Velocity, Time, Estimation in Satellite Navigation Receivers

Common techniques for position-velocity-time estimation in satellite navigation, iterative least squares and the extended Kalman filter, involve matrix operations. The matrix inversion and inclusion of a matrix library pose requirements on a computational power and operating platform of the navigati...

Full description

Saved in:
Bibliographic Details
Published inRadioengineering Vol. 22; no. 3; pp. 776 - 790
Main Authors O. Jakubov, P. Kovar, P. Kacmarik, F. Vejrazka
Format Journal Article
LanguageEnglish
Published Spolecnost pro radioelektronicke inzenyrstvi 01.09.2013
Subjects
Factor graph
GNSS
Kalman filter
PVT
sum-product algorithm
Online AccessGet full text

Cover

More Information
Summary:Common techniques for position-velocity-time estimation in satellite navigation, iterative least squares and the extended Kalman filter, involve matrix operations. The matrix inversion and inclusion of a matrix library pose requirements on a computational power and operating platform of the navigation processor. In this paper, we introduce a novel distributed algorithm suitable for implementation in simple parallel processing units each for a tracked satellite. Such a unit performs only scalar sum, subtraction, multiplication, and division. The algorithm can be efficiently implemented in hardware logic. Given the fast position-velocity-time estimator, frequent estimates can foster dynamic performance of a vector tracking receiver. The algorithm has been designed from a factor graph representing the extended Kalman filter by splitting vector nodes into scalar ones resulting in a cyclic graph with few iterations needed. Monte Carlo simulations have been conducted to investigate convergence and accuracy. Simulation case studies for a vector tracking architecture and experimental measurements with a real-time software receiver developed at CTU in Prague were conducted. The algorithm offers compromises in stability, accuracy, and complexity depending on the number of iterations. In scenarios with a large number of tracked satellites, it can outperform the traditional methods at low complexity.
ISSN:1210-2512