Mike Koivisto*, Mário Costa†, Aki Hakkarainen*, Kari Leppänen† and Mikko Valkama*
* Tampere University of Technology, Finland
† Huawei Technologies, Finland
It is commonly expected that future fifth generation (5G) networks will be deployed with a high spatial density of access nodes (ANs) in order to meet the envisioned capacity requirements of the networks. Densification is beneficial not only for communications but it also creates a perfect infrastructure for highly accurate user node (UN) positioning. Despite the fact that positioning will play an important role in future networks enabling huge amount of location-based applications and services, this great opportunity is not widely identified in the existing literature. Therefore, an unscented Kalman filter (UKF)-based method for estimating directions of arrival (DoAs) and times of arrival (ToA) at ANs as well as performing joint 3D positioning and network synchronization in a network-centric manner is proposed in this paper. In addition to derivation of the proposed UKF-based solution, the existing 2D extended Kalman filter (EKF)-based solution is extended to cover also realistic 3D positioning scenarios. Building on the premises of 5G ultra-dense networks (UDNs), performance of the both methods is evaluated and analysed in terms of DoA and ToA estimation as well as positioning and clock offset estimation accuracy, through a map- based approach combined with detailed ray tracing propagation modeling. Based on the comprehensive numerical evaluations, the both proposed methods can provide the envisioned one meter positioning accuracy even in the case of unsynchronized 5G network while tracking the clock offsets of network elements with extremely high accuracy.
Published in the proceedings of IEEE GLOBECOM Workshops 2016
Videos and Demonstrations
3D Drone Tracking
3D UN position tracking with the proposed DoA/ToA Pos&Sync EKF and UKF in an example drone trajectory on the METIS-based Madrid map. First, the drone takes off from the ground, and in the middle of the trajectory it performs a landing and short halt before a new take-off. In addition to the 3D positioning, the clock offsets of the phase-locked LoS-ANs are also tracked within the proposed filters, and these estimation errors are illustrated in the figure on the right.
Vehicle Tracking
3D UN position tracking with the proposed DoA/ToA Pos&Sync EKF and UKF in an example vehicle trajectory on the METIS-based Madrid map. The maximum speed of the vehicle is set to 50km/h while turning speed is set to 20km/h. In addition to the 3D positioning, the clock offsets of the phase-locked LoS-ANs are also tracked within the proposed filters, and these estimation errors are illustrated in the figure on the right.