Mitigating Performance Attacks on UGVs/UAVs

Security attacks are becoming increasingly common in safety­critical and mission­critical systems. As a recent example, it was reported in July 2015 that hackers were able to take control of the Jeep Grand Cherokee via its internet­connected entertainment systems [a], leading to a safety recall of 1.4 million vehicles in the US [b]. As part of a broader effort to tackle this challenge, this project aims to conduct an experimental study of performance attacks ­­ a special class of security attacks ­­on unmanned vehicles (e.g., UGVs and UAVs) and to investigate corresponding mitigation methods based on real­-time virtualization. The project will focus on real­time performance attacks, where an adversary attempts to consume the platform resources maliciously to prevent the critical functions of the vehicles from responding in a timely manner, potentially jeopardizing their safety and/or missions. In addition, we will restrict to attacks through CPU, network, and cache resources.

The team will be working with a doctoral student (Meng Xu) and the faculty. The milestones of this project are as follows:

1. Deploy a ROS system on an Intel Atom machine to control a UAV or a UGV.
2. Install Xen and configure the system to enable outside network access for guest VMs. Deploy the ROS system on multiple VMs to collaboratively control the UAV/UGV.
3. Install RT­Xen and deploy the ROS system on multiple VMs atop RT­Xen.
4. Demonstrate successful performance attacks that cause the ROS system to crash, and measure the recovery latency. 
5. Develop mitigation approaches against the above performance attacks based on RT­Xen.
6. Demonstrate and evaluate the performance of the methods developed in #5.
7. Repeat the above steps on ARM board; the key part is deploying RT­Xen on an ARM board.

References:

[a] http://www.wired.com/2015/07/hackers­remotely­kill­jeep­highway/, Accessed on July 25, 2015

[b] http://www.bbc.com/news/technology­33650491, Accessed on July 25, 2015

Interested students should:

• submit their CVs and any relevant documents that demonstrate/describe their experiences on the relevant topics and/or skills.
• indicate how long they can work on the projects.