Indoor localization is a hot research topic yet to find a shared agreement on the key methods and technologies enabling a satisfactory solution to the problem. Among the plethora of techniques currently being studied, “Angle of Arrival” (AoA) has recently gained momentum as the Bluetooth Special Interest Group introduced it in the new Bluetooth 5.1 standard as the release’s major enhancement. In Bluetooth-based AoA, a set of locator nodes are able to compute the angle between themselves and a to-be-located device sending an ad-hoc crafted Bluetooth packet. Given the angles and the positions of the locators, a data fusion algorithm can compute the approximate relative position of the observed device. In the right conditions, AoA can reach sub-meter accuracy. However only the lowest levels of the stack are already provided, and the implementation of the data fusion layer is left to the developers. Correctly engineering this layer is not an easy task, as it would require a complex tuning process through trial and error, involving cumbersome real world tests. The goal of this thesis is to build a tool aimed at supporting the engineering of AoA-based localization applications, both in-silico and in practice. Along this line, we develop an actor-based software framework that  is agnostic with respect to the specific technology used to realize AoA, and (ii) can work with both simulated and real AoA-based sensors. The proposed framework is developed in the context of the European project “PRYSTINE”, and aims at achieving a full-stack localization system based on Bluetooth-based AoA as the physical medium, and particle filtering as the data- fusion algorithm. The proposed framework, however, is general enough to support any other technology capable of calculating AoA, and any other filtering algorithm for data fusion.