Abstract
This article presents a framework for opportunistic unmanned aerial vehicle (UAV) navigation by exploiting carrier phase measurements from ambient cellular signals of opportunity. In the proposed framework, the cellular base transceiver stations (BTSs) are not assumed to be synchronous. A complete framework that employs an extended Kalman filter (EKF) is presented, including filter initialization and process and measurement noise covariance selection. The EKF estimates the position and velocity of the UAV, as well as the differences between the UAV-mounted receiver and each of the BTSs’ clock bias and clock drift. The observability of the estimation framework is analyzed, and the boundedness of the EKF’s errors is studied. It is shown that the system is observable given a class of vehicle and receiver clock dynamics. A lower bound for the EKF estimation error covariance is derived, and it is shown that the covariance remains bounded. Monte Carlo simulations are conducted to study the effect of the number of BTSs, the initial UAV speed, and the receiver’s oscillator quality, on the estimation performance. Two sets of experimental results are presented demonstrating UAVs navigating exclusively with cellular carrier phase measurements via the developed framework, achieving a total position root-mean-squared error of 2.94 and 5.99 m for UAV trajectories of 2.6 and 2.9 km, respectively.
