Cooperative/Parallel Kalman Filtering for Decentralized Network Navigation

Wenyun GAO  Xi CHEN  Dexiu HU  Haisheng XU  

IEICE TRANSACTIONS on Communications   Vol.E99-B   No.9   pp.2087-2098
Publication Date: 2016/09/01
Publicized: 2016/03/18
Online ISSN: 1745-1345
DOI: 10.1587/transcom.2016EBP3006
Type of Manuscript: PAPER
Category: Navigation, Guidance and Control Systems
network navigation,  decentralized algorithm,  cooperative algorithms,  Kalmen filter,  

Full Text: PDF>>
Buy this Article

This paper presents non-iterative cooperative/parallel Kalman filtering algorithms for decentralized network navigation, in which mobile nodes cooperate in both spatial and temporal domains to infer their positions. We begin by presenting an augmented minimum-mean-square error (MMSE) estimator for centralized navigation network, and then decouple it into a set of local sub-ones each corresponding to a mobile node; all these sub-estimators work in parallel and cooperatively — with the state estimates exchanging between neighbors — to provide results similar to those obtained by the augmented one. After that, we employ the approximation methods that adopted in the conventional nonlinear Kalman filters to calculate the second-order terms involved in these sub-estimators, and propose a decentralized cooperative/parallel Kalman filtering based network navigation framework. Finally, upon the framework, we present two cooperative/parallel Kalman filtering algorithms corresponding to the extended and unscented Kalman filters respectively, and compare them with conventional decentralized methods by simulations to show the superiority.