section{Conclusion} could serve as a blueprint for developing

The related work done in the field of drone security to explore the vulnerabilities in drones, helps us understand the imminent attacks that could be carried out on commercial drones in the future. In order to gain complete knowledge about the mechanism of such attacks, we proposed a honeypot solution which would help gather valuable information about the attacks. This information could be used to analyze the attacks and develop security solutions against them.
par We analyzed the attacks that were carried out on drones, the protocols that are most commonly used by drones, and the hardware and software used by the drones. Based on this analysis, we developed an architecture that could serve as a blueprint for developing honeypot software for drones. Then a prototype implementation of the proposed architecture has been implemented followed by an evaluation of the implementation. The implementation helps understand the advantage of honeypots in a drone environment.

During the course of this thesis, we have tried to answer the research questions introduced in chapter 1. The below list discusses briefly the answers that were found as a result of our research.
item extbf{Research question – 1}: What are the different attacks that can be possible on a drone?
par Drones are susceptible to a wide range of attacks both physical and logical. Chapter 3 of this thesis sheds light on a variety of attacks possible on drones including GPS attacks, Wi-Fi based attacks, MAVLink based attacks, etc.
item extbf{Research question – 2}: Which of the attacks found from research question – 1 can be detected or mitigated through a honeypot?
par Since a honeypot is intended to run in the application layer, it can capture information only about those attacks that exploit the application layer protocols. Some of these attacks include Telnet attack, SSH based attack, etc.
item extbf{Research question – 3}: How suitable are traditional honeypots for a drone and how does a drone-specific honeypot vary from a traditional honeypot?
par Traditional honeypots are not developed with drone scenarios in mind. Therefore, they are not readily suitable to be used in drones. Table 6 reveals the requirements of drone honeypot and how the existing honeypots match these requirements. It can be seen, there is no honeypot that matches all the requirements to emulate a drone.
item extbf{Research question – 4}: How can a drone honeypot be realized in terms of technology and architecture?
par From chapter 2, we infer that the most commonly used operating system in drones is Linux and common communication link used is Wi-Fi. Based on this, our honeypot implementation runs on Linux with a Wi-Fi access point.

We Will Write a Custom Essay Specifically
For You For Only $13.90/page!

order now

par In addition to the above mentioned findings, we also found that UDP communication through port 14550 is used by most of the ac{gcs} software, that support MAVLink, to communicate with the drone. This is also the port used by companion computers to talk to the drone. When the Wi-Fi is not secure, it is easy to establish a UDP connection and pass MAVLink messages to the drone.

There are certain limitations with respect to honeypots when they emulate a drone. The drones, for example are expected to be attacked when they are airborne. Therefore, the location of the drones is expected to vary constantly. But, the honeypot is usually installed in a fixed location. Due to this, it is possible for an attacker to detect a honeypot by monitoring the strength of the Wi-Fi signal that comes from the honeypot’s access point. Another limitation is the challenge in emulating the drones that offer ac{fpv} streams to the ac{gcs} software. While it is possible to render fake streams through a honeypot, these would be easily identified if the attacker tries to issue telemetry commands to the honeypot.

subsubsection*{Future Work}
The main focus of this thesis is on drones that use Wi-Fi for communication. However, the usage of LTE has been increasing among the commercial drone solution providers. The future work could extend the idea of honeypot to LTE based drones. Another area that could be explored is the usage of honeypot as a intrusion detection system, where the collected information can be used to resort to fault handling mechanisms. In an area of high drone traffic such as in fulfillment centers of Amazon, such intrusion detection systems could alert the central controllers thereby securing the network when an anomaly is detected. Since we use the existing simulator from Ardupilot to emulate MAVLink, we have limited functionality. It would be
very effective to develop an emulator specifically for drone honeypots or extend the Ardupilot SITL to cover more honeypot functionality. More research could be done to have the honeypot run on board a real drone. This could help the limitation introduced due to the varying signal strength.


I'm Johnny!

Would you like to get a custom essay? How about receiving a customized one?

Check it out