3D Orientation Estimation with Multiple 5G mmWave Base Stations

TL;DR

This paper presents a method to estimate a user's 3D orientation using signals from multiple 5G mmWave base stations, leveraging maximum likelihood estimation on rotation matrices and geometric least squares for initialization.

Contribution

It introduces a novel approach combining geometric least squares and maximum likelihood estimation for 3D orientation using multiple mmWave base stations, with performance close to theoretical bounds.

Findings

Orientation can be accurately estimated with signals from at least two base stations.

The proposed estimators perform near the theoretical lower error bound.

Numerical results validate the effectiveness of the method.

Abstract

We consider the problem of estimating the 3D orientation of a user, using the downlink mmWave signals received from multiple base stations. We show that the received signals from several base stations, having known positions, can be used to estimate the unknown orientation of the user. We formulate the estimation problem as a maximum likelihood estimation problem in the the manifold of rotation matrices. In order to provide an initial estimate to solve the non-linear non-convex optimization problem, we resort to a least squares estimation problem that exploits the underlying geometry. Our numerical results show that the problem of orientation estimation can be solved when the signals from at least two base stations are received. We also provide the orientation lower error bound, showing a narrow gap between the performance of the proposed estimators and the bound.