Chen, Wenxin (University of Hawaii), Duan, Zhenhai (Florida State University), Dong, Yingfei (University of Hawaii)

False Data Injection on EKF-Based Navigation Control

Scheduled for presentation during the "GNC - UAS" (FrB4), Friday, June 16, 2017, 15:05−15:25, Lummus Island

2017 International Conference on Unmanned Aircraft Systems, June 13-16, 2017, Miami Marriott Biscayne Bay, Miami, FL,

This information is tentative and subject to change. Compiled on April 26, 2024

Abstract

Small consumer unmanned aircraft systems (sUASs) have been involved in many security incidents recently. Current defense methods have serious limitations because they mostly try to physically capture an sUAS or jam its communications and its GPS signals to trigger its fail-safe features. While physical methods are usually unscalable and often slow in response, the default fail-safe features can be easily changed to other options such as continuing its pre-configured mission, regardless of jamming. Therefore, new defense solutions must be developed to address these issues. To our best knowledge, very little research has been published in this area.

In this paper, we propose a false data injection attack to exploit the limitation of sensoring and processing capabilities on an sUAS and manipulate its on-board sensors, in order to deviate it from its (pre-configured) flight plan. We have analyzed the popular ArduPilot flight control system, identified potential vulnerabilities in its sensor data fusion process, and developed a method to manipulate on-board magnetometer readings while avoiding being detected by its bad-data detection schemes. Our initial results show that the proposed method is able to misguide the navigation system and deviate an sUAS from its target destination far away. While we present our main ideas in this paper, we are further investigating related theoretical issues and improving the proposed method with practical experimental tests.