Enable breadcrumbs token at /includes/pageheader.html.twig

Sandia Labs Deploys Brain-Inspired Cybersecurity System

The processor detects threats much faster and more efficiently than traditional systems.

Researchers from Sandia and Lewis Rhodes labs examine the Neuromorphic Cyber Microscope, a small processor that can replace multiple racks of traditional systems used for cybersecurity. Photo by Randy Montoya

Researchers at Sandia National Laboratories helped develop a potentially game-changing cybersecurity system that mimics the human brain’s ability to analyze data, and they are now testing the technology on the labs’ highly targeted networks.

The Neuromorphic Cyber Microscope, designed by Lewis Rhodes Labs in partnership with Sandia, has proved 100 times faster and 1,000 times more energy-efficient at finding threats than conventional state-of-the-art systems. Furthermore, as threat patterns become become complex and sophisticated conventional systems slow down, the brain-inspired system keeps chugging along at the same highly efficient rate.

That’s because traditional systems sequentially analyze data in small chunks as they search for the known patterns of cyber intruders. The Neuromorphic Cyber Microscope, on the other hand, looks at streaming data in parallel, much the same way the human brain analyzes vast amounts of information being delivered through the sense of vision. The neural-inspired architecture can do on a single chip, a field-programmable gate array, what would normally require multiple racks of computers. Compared with conventional systems, performance gains of more than 10,000 times could be attainable.

“As you look around, your eyes collect a tremendous amount of imagery, and at very low power, it’s able to identify what’s around you. If you think of the millions of things the brain is comparing against in real time, and the fact that it’s doing this with just a couple of watts of power, it’s really a remarkable capability,” says David Follett, co-founder and CEO of Lewis Rhodes Labs. “The product is based on this single chip that’s programmed to view the world the way that the human brain views the world.”

John Naegle, a computer systems expert who led the Sandia team on the project, adds that the technology upends the traditional method of processing data. “This architecture ... turns the model on its head as to how the data moves through the processor. It’s taking the world that it sees—the cyber information—and streaming it through the device in a completely different way so that it burns a whole lot less energy,” he explains.

Sandia not only helped develop the technology but also is testing the Neuromorphic Cyber Microscope on its own networks. In six months, it likely will be fully integrated with them.

“For us, any kind of additional capability gives us a much better tool to look at the vast amounts of data that we have that streams through our networking systems,” Naegle says. “Sandia is a very high-value target, and we’re in the business of making sure that we protect that data.”

Follett reports that the system also is deployed with three different Fortune 100 companies that, he suggests, may not want to be named.

Naegle offers that because it shows such great promise, the system could be improved over a number of years as researchers learn to better harness its capabilities. Follett agrees: “Because it’s based on concepts from the brain, it has the ability to do interesting behavioral algorithms. We’re just in the early stages of learning to take advantage of all of these unique capabilities that it has. The technology is very disruptive.”

Like the human brain, the system may prove adept at recognizing or even predicting behavioral changes, which will be beneficial in battling cyber adversaries. “If an attacker changes their techniques or process in some way, there are some very interesting ways of using this device to alert the analyst,” Follett says.

The device also might have practical applications outside of the cyber realm. For instance, Sandia researchers have explored a type of machine learning used for audio and image processing and efficiently sorting numbers. “The device itself will analyze any stream of data. In the financial industry, you could use it for doing fraud detection. For social media, you could use it for detecting opportunities for ad placement,” Follett says. “We do expect to generalize it out into the commercial space into a number of noncyber applications.”

The processor in the Neuromorphic Cyber Microscope is based on research conducted by Follett’s wife, Pamela, a co-founder of Lewis Rhodes Labs. The pediatric neurologist and neuroscientist studies developmental diseases such as cerebral palsy. Her research has compared healthy brains to brains with the movement disorder. It offered key insights that led to the cyber microscope.

The researchers say that calling the piece of computer hardware a “microscope” is, in this case, appropriate because of the fidelity it brings to the cyberthreat. Analysts, they say, talk about cybersecurity having a fidelity problem because, along with small bits of useful data, the sensors provide a great deal of information that is somewhat less than useful. “Building on this concept of fidelity, we thought microscopes have extremely good fidelity and high resolution, and so it just morphed into being a cyber microscope,” Follett explains.