Applying Modified Householder Transform to Kalman Filter

Kalman filter (KF) is a key operation in many engineering and scientific applications ranging from computational finance to aircraft navigation. Recently, there have been proposals in the literature for acceleration of KF using modified Faddeeva algorithm (MFA) where the classical Householder transf...

Full description

Saved in:
Bibliographic Details
Published in2019 32nd International Conference on VLSI Design and 2019 18th International Conference on Embedded Systems (VLSID) pp. 431 - 436
Main Authors Merchant, Farhad, Vatwani, Tarun, Chattopadhyay, Anupam, Raha, Soumyendu, Nandy, S K, Narayan, Ranjani, Leupers, Rainer
Format Conference Proceeding
LanguageEnglish
Published IEEE 01.01.2019
Subjects
Acceleration
Field programmable gate arrays
instruction level parallelism
kalman filter
Kalman filters
Parallel processing
Pipelines
reconfigurable computing
Sparse matrices
state estimation
Transforms
Online AccessGet full text

Cover

More Information
Summary:Kalman filter (KF) is a key operation in many engineering and scientific applications ranging from computational finance to aircraft navigation. Recently, there have been proposals in the literature for acceleration of KF using modified Faddeeva algorithm (MFA) where the classical Householder transform (HT) is used in implementation of MFA on a custimizable platform called REDEFINE. REDEFINE is a coarse-grained reconfigurable architecture that has capabilities of recomposing data-paths at run-time and on-demand. In this paper, we present realization of KF using MFA where we implement MFA using modified Householder transform (MHT) presented in the literature. We call this as M^2 FA. It is shown that the implementation of KF using M^2 FA clearly outperforms the implementation of KF using MFA on REDEFINE and also the realization of KF on REDEFINE is scalable. Performance improvements over state-of-the-art implementations are also discussed.
ISSN:2380-6923
DOI:10.1109/VLSID.2019.00092