As more private data is stored and shared digitally, researchers are investigating new methods to protect data from bad actors. Current silicon technology uses microscopic differences between computing components to generate secure keys, but artificial intelligence (AI) techniques can be used to predict these keys and gain data access. Penn State researchers have devised a method to make the encrypted keys more difficult to crack.
The researchers, led by Saptarshi Das, assistant professor of engineering science and mechanics, used graphene – a one-atom-thick layer of carbon – to create a novel low-power, scalable, reconfigurable hardware security device with significant resilience to AI attacks. They published their findings today in Nature Electronics.
“There has been an increase in the breach of private data recently,” Das said. “We created a new hardware security device that could be used to protect these data across industries and sectors in the future.”
As more private data is stored and shared digitally, researchers are exploring new ways to protect data against attacks from bad actors. Current silicon technology exploits microscopic differences between computing components to create secure keys, but artificial intelligence (AI) techniques can be used to predict these keys and gain access to data. Now, researchers have designed a way to make the encrypted keys harder to crack.
According to the researchers, the device, known as a physically unclonable function (PUF), is the first demonstration of a graphene-based PUF. The novel PUF is more energy-efficient, scalable, and secure against AI attacks that threaten silicon PUFs due to graphene’s physical and electrical properties, as well as the fabrication process.
The team started by making nearly 2,000 identical graphene transistors, which are used to switch current on and off in a circuit. Despite their structural similarity, the electrical conductivity of the transistors varied due to the inherent randomness of the manufacturing process. While such variation is typically a disadvantage for electronic devices, it is a desirable quality for a PUF that silicon-based devices do not have.
After incorporating graphene transistors into PUFs, the researchers modeled their properties to create a simulation of 64 million graphene-based PUFs. Das and his team used machine learning to test the security of the PUFs, a method that allows AI to study a system and discover new patterns. The researchers trained the AI using graphene PUF simulation data, attempting to see if the AI could use this training to make predictions about the encrypted data and reveal system flaws.
“Neural networks are very good at developing a model from a huge amount of data,” Das said. “We found that AI could not develop a model, and it was not possible for the encryption process to be learned.”
According to Das, the PUF’s resistance to machine learning attacks makes it more secure because potential hackers cannot use breached data to reverse engineer a device for future exploitation. Even if the key could be predicted, the graphene PUF could generate a new key via a reconfiguration process that did not require any additional hardware or component replacement.
“Normally, once a system’s security is compromised, it is permanently compromised,” said Akhil Dodda, a graduate student in engineering science and mechanics working under Das’ supervision. “We devised a scheme in which a compromised system could be reconfigured and used again, with tamper resistance added as an additional security feature.”
With these characteristics, as well as the ability to operate at a wide range of temperatures, the graphene-based PUF could be used in a variety of applications. Further research could pave the way for its use in flexible and printable electronics, household devices, and other applications.
Penn State’s Dodda, Shiva Subbulakshmi Radhakrishnan, Thomas Schranghamer, and Drew Buzzell, as well as Purdue University’s Parijat Sengupta, are paper co-authors. Das is also affiliated with the Penn State Department of Materials Science and Engineering and the Materials Research Institute.
