An Optimal Data Fusion Algorithm Based on the Triple Integration of PPP-GNSS, INS and Terrestrial Ranging System

Abstract

This paper describes the integration of Locata, GNSS and INS technologies within a loosely-coupled triple integration algorithm. The conventional methods for multi-sensor integration can be classified as either centralised filtering or decentralised filtering. Centralised Kalman filtering (CKF) provides globally optimal state estimation by directly combining measurement data. However CKF system has some disadvantages such as a comparatively large computational burden and poor fault detection and isolation ability. Decentralised Kalman filtering (DKF) addresses such defects while aiming to achieve the same accuracy as a centralised filter. On the other hand global optimal filtering (GOF) can achieve a higher accuracy than the traditional CKF because it utilises more information resources than the CKF. In the information space, the information resources that can be used for estimation include the measurements, the local predictions, and the global predictions. In order to evaluate the system performance, a field experiment was conducted on a vehicle with different kinds of maneuvers, including circular motion and accelerated motion. The results indicate that: (1) GOF-based PPP-GNSS/Locata/INS integration system can provide better positioning accuracy compared with CFK and federated Kalman filtering; (2) covariance analysis shows that the GOF improves the system estimation covariance; and (3) a comparison of GOF with local filters confirms the superiority of a GOF-based triple integration system.