Australian Algorithm Stops Cyberattacks on Unmanned Military Robots
A team of artificial intelligence experts from Charles Sturt University and the University of South Australia have developed an algorithm that can quickly prevent man-in-the-middle cyberattacks on unmanned military robots. The algorithm, created through deep learning neural networks, was trained to recognize the signature of these attacks, where hackers interrupt ongoing conversations or data transfers.
An Effective Defense Against Cyberattacks
Using a replica of a United States army combat ground vehicle, the algorithm was tested in real time and successfully prevented 99% of malicious attacks. With false positive rates of less than 2%, the system proves its effectiveness. These results have been published in the IEEE Transactions on Dependable and Secure Computing.
Better Than Other Techniques
Professor Anthony Finn, an autonomous systems researcher at UniSA, believes that this algorithm outperforms other recognition techniques used globally to detect cyberattacks. Collaborating with the US Army Futures Command, Professor Finn and Dr. Fendy Santoso from Charles Sturt Artificial Intelligence and Cyber Futures Institute effectively replicated a man-in-the-middle cyberattack on a ground vehicle and trained its operating system to identify the attack.
“The reliance on highly networked robot operating systems makes them extremely vulnerable to cyberattacks,” says Professor Finn.
The Need for Sophisticated AI Algorithms
Industry 4, characterized by the evolution in robotics, automation, and the Internet of Things, requires robots to work collaboratively and exchange information via cloud services, leaving them susceptible to cyberattacks. However, the speed of computing is rapidly increasing, allowing for the development and implementation of advanced AI algorithms to protect systems against digital attacks.
According to Dr. Santoso, the robot operating system lacks security measures due to encrypted network traffic data and limited integrity-checking capability. However, their intrusion detection framework, backed by deep learning, proves to be highly accurate and robust. The system can handle large datasets necessary to safeguard big-scale, real-time, data-driven systems like the robot operating system.
In the future, Professor Finn and Dr. Santoso plan to test their intrusion detection algorithm on various robotic platforms, such as drones, which have faster and more complex dynamics compared to ground robots.