Australian researchers at UniSA University have succeeded in preventing Man-in-the-Middle (MitM) attacks on autonomous military robots. They did so using an algorithm that taught the robot to detect MitM attacks.
In the experiment, researchers from Australian universities succeeded in preventing a MitM attack on an autonomous US Army robot. The intrusion detection algorithm trained the robot to recognize and interrupt such an attack.
The tests showed that the algorithm they used worked in 99 percent of cases. False positive rates were less than 2 percent. Thus, these figures show that the algorithm was quite successful. The researchers also indicated that the experimental algorithm works better than existing techniques for detecting cyber attacks.
Military autonomous robots are vulnerable
The researchers are also pleased that their experiment succeeded because, in general, autonomous military robots are very vulnerable to cyber attacks such as data breaches and electronic hijackings. This is because these systems are so dependent on network technology that hackers can always find an attack vector.
According to the researchers, encrypted data traffic and limited integrity check functionality also limit these types of systems for proper security measures in the underlying code.
Also interesting for Industry 4.0
Not just the military should be happy about the end result, as the entire Industry 4.0 sector also stands to benefit. The Australian experts argue that This sector is still vulnerable to cyber attacks.
The researchers will now test their algorithm on more (military) autonomous applications, such as drones.
Tip: How Anthropic is unraveling the mysteries of neural networks