OPTIMIZING NETWORK CONNECTION SCHEDULING FOR DRONE VIDEO STREAMING

Abstract

Drones are becoming increasingly popular today, most notably for their ability to provide a remote mobile video stream. Most commercial drones rely on Wi-Fi connections to send video packets to the client, but this limits the range and environments in which drone video streaming is reasonably possible. While work has been done to make drones capable of connecting to networks with greater range, most notably LTE, there has been little investigation into how a drone might switch between different network connections throughout a single flight session to maximize the data it can transfer throughout that flight. The goal of this thesis is to construct a scheduling optimization algorithm that can model the selection of the optimal sequence of connections over a flight for maximum data transfer while accounting for the overhead involved in switching from one connection to another. Using this algorithm, we observe that scheduling optimization yields significantly larger increases in data transferred than a naïve approach as the speed of the drone increases and the density of Wi-Fi APs increases.