Masters Thesis Defense “Overwhelming the SAA System of Delivery UAVs by Drone Swarming” By Barry Pfaff

Committee:  Drs. Yong Pei, Advisor, Mateen Rizki, and Mr. Vance Saunders

ABSTRACT:

As the Internet continues to replace the brick-and-mortar store as the main place for purchasing goods, web-based companies are looking for ways to reduce the cost of delivering those goods. The use of Unmanned Aerial Vehicles, or UAVs, is one delivery method that is being used increasingly. These UAVs can be programmed with delivery routes and destinations and complete the job while only requiring limited intervention from human controllers. Sense and Avoidance (SAA) systems have recently been incorporated into these UAVs, such that they can detect objects in their flight path, reroute the UAV accordingly and operate in an even more autonomous fashion such that the interaction of a controller is minimized. This is vital for delivery UAVs that are used within urban areas where the threat of objects along the flight path is the greatest. As a result, due to its low-cost and low-labor nature, UAV delivery service is expected to become a critical part of emerging smart city infrastructure.

However, these SAA-capable UAVs may create another potential attack surface for cyber threats, which may significantly hinder the broader adoption of UAV delivery services if not properly and sufficiently addressed. This research intends to provide a systematic study of this potential cyber challenge by uncovering the limitations of both UAV's SAA systems and the operational procedure of UAV delivery services that could be exploited in an attempt to disrupt orderly delivery services.

This research starts with examining how an SAA system works and investigating its limitations. We then propose a tessellation-based analysis approach, using a typical urban scenario, to define a safe-operation zone for UAVs which help determine the optimal number of UAVs that can operate safely in an urban area in order to efficiently complete the potentially contentious delivery tasks. We will also discuss the inherent weaknesses of web-based logistic management systems that are used to program the routes and destinations into the UAVs. Next, we look at how an SAA system can be overwhelmed such that the UAV can effectively be "hacked" using a physical form of Distributed Denial of Service (DDoS). Finally, we will examine ways that this form of DDoS threat can be mitigated in order to support larger scale adoption of UAV delivery services.