The U.S. Department of Defense is finding what could be considered “the sacred grave of data encryption,” which would seal a loophole that allows hackers to access sensitive information while processing it.
In the present day encryption, a well-defined set of calculations, called an algorithm, scrapes data so that it is no longer readable. Those who access the data are given a set of numbers called a key, which is the code that allows you to disable that data again.
If someone wanted to use the encrypted data to do anything useful, they would first have to decrypt it back to so-called “plain text”, which makes it more likely to spy again. To help protect the now decrypted information, those working with the plain text usually do not trust computers. However, as is evident from regular headlines about data breaches at key organizations, it is becoming increasingly difficult to tell which devices are secure.
“With all the news about these sections, these malware attacks, we can’t trust our hardware or software systems,” said Tom Rondeau, program manager at Research Projects Group. Advanced Defense (DARPA), to Live Science.
Related: Flight salons to mind control: 22 proven military secrets
That’s why DARPA is trying to drive improvements in something called full homomorphic encryption (FHE). This method makes it possible to analyze computer data while it is still in encrypted format. This could allow financial crime investigators to scrape sensitive bank records without revealing customer details, for example, or allow health researchers to analyze private health data while ‘ in which they preserve patient privacy, Rondeau said. This approach could help the military keep their battlefield data more secure and make it easier to allow friends to work with recorded intelligence data.
The key to the approach is its name, which comes from the Latin words “homos,” meaning “the same,” and “morphe,” meaning “shape.” It refers to the the fact that some mathematical operations they can map data from one format to another without changing the basic structure of the data. That means that changes made to the data and in one format are saved when that data is converted back to the other one. This principle can be applied to encryption, because computers represent all data, including text, as numbers.
Here’s a pretty simple example of how this might work: Think of an encryption scheme that writes data by multiplying it by 3, so if you encrypt the number 8 you get 24. If you multiply your encrypted data with 2, you get 48. When you decrypt it again by dividing it by 3, you get 16, which is the same result you would get if you multiplied the non- you encrypt with 2.
In this example, the encryption method is very easy to work out from the output, so it is not secure. But FHE relies on something far more complex called lattice cryptography, which encodes data as coordinates on a surface. Lathes can be thought of as grids of regularly rotated dots, but, unlike the 2D grids we are accustomed to, the FHE lashes are multifaceted.
Related: The 11 most beautiful mathematical equations
So, instead of describing the position of each data point with simple X, Y coordinates, the number of axes can be quite large, with thousands of coordinates describing each piece of data special. Data points can also be spaced between dots, so each coordinate can have many decimal places to indicate the exact location. This makes it impossible to crack the encryption, even with quantum computers. That’s a promising feature, Rondeau said, since today’s major encryption methods are not quantum protection.
The big problem is that processing this data is very slow on conventional computers – about a million times slower than processing times for unencrypted data. That’s why DARPA has launched a research program called Data Protection in Virtual Environments (DPRIVE), which Rondeau manages, to speed things up. The program recently awarded contracts to startup Duality Technologies an encryption company, Galois software company, SRI International nonprofit and a division of Intel, called Intel Federal to design new processors and software to directly increase speeds 10 times slower than normal, which is 100,000 times faster than conventional processing for full homomorphic encryption.
FHE is so slow because of the way computers perform it. To complicate matters, these data points will not be stable. The researchers found that you can perform mathematical operations such as multiplication or addition by moving data points around the interior of the surfaces. By combining many of these operations, researchers can perform all kinds of computations without decrypting the data. When you close the answer, there is a chance that someone might check it; but that response would still reveal nothing about the data used to make it up.
Related: Top 9 numbers
The overall problem with this process is that moving precisely positioned data points around a high-volume space is far more complex than calculating simple binary data – the standard 1s and 0s of modern computers.
“It’s this data explosion,” Rondeau told Live Science. “Now, not all computing just handles one thing. It handles all this information, all those representations of the dimensions.”
There are two main approaches that DARPA-funded companies can use to simplify things, Rondeau said. One invention is to improve the computer’s ability to handle high-precision numbers, by changing the way numbers are represented in binary code and changing the cycles of chips for more efficient processing. The other is to convert the data to a lower place where the calculation is simpler, which requires new hardware and software approaches as well.
Each team involved in the program takes a slightly different approach, but Rondeau says he’s confident they can move forward with the 100,000-fold targeted improvement in processing speed.
First published on Living Science.