This paper firstly identifies and discuses the main challenges in exploiting antenna arrays to track the position of unmanned aerial vehicles (UAVs). Then, a specific radio localization method based on the multiple differential phase-of-arrival (D-PoA) and time-of-arrival (TOA) measures at a 3-axial uniform linear array (3A-ULA) is presented to estimate the positron of an UAV with respect to a reference point. The D-PoA and TOA measures are coupled with a dynamic motion model of the UAV to enable the usage of non-linear Bayesian estimation methods such as the particle filter (PF) and the cuabture Kalman filter (CKF) to improve the positioning accuracy. Furthermore, a comparison in terms of accuracy and complexity of the PF and the CKF for the considered application is presented. To assess the estimation accuracy of these methods, a confined area random aerial trajectory emulator algorithm is used to generate actual paths of the flying UAVs.
