Not yet reviewed by Pith; the record is open.

We propose a novel top-k service composition framework for drone services under a dynamic environment. We develop a system model for formal modelling of drone services in a skyway network. The composition process is accomplished in two phases, i.e., computing top-k compositions and extending and ranking top-k compositions using probabilistic wait and recharge times under congestion conditions. We propose a top-k composition algorithm to compute the best service composition plan meeting user's requirements. A set of experiments with a real dataset is conducted to demonstrate the effectiveness of the proposed approach.