PC Hardware Is Physically Leaking Your Encryption Keys
All it takes is a dedicated hacker, an antenna, and some signal processing skills.
Your computer has an aura. That sentence causes me actual physical pain to type, but there's not really a better way to put it—whether you're surfing really fucked up porn or writing the great American metaphysical novel, there is an ethereal layer of electromagnetic radiation, heat, and noise surrounding the tactile machine underneath. This is sort of just a fact of computer hardware. Would it be better if I called it a "halo"?
Regardless, according to computer science researchers from Israel Institute of Technology (aka the Technion) writing in the Communications of the ACM, this aura bears potentially revealing information about the goings-on within your machine and the associated security measures that machine employs in its protection of your data.
It's been demonstrated in the past that physical encryption keys can be extracted from very small devices, such as RFID tags or smart-cards, but only now has it been extended to a full-on hardware organism such as a laptop or any other "PC class" computer.
"For attackers, ramming the gates of cryptography is not the only option," the paper notes. "They can instead undermine the fortification by violating basic assumptions made by the cryptographic software. One such assumption is software can control its outputs. Our programming courses explain that programs produce their outputs through designated interfaces; so, to keep a secret, the software just needs to never output it or anything that may reveal it."
This assumption turns out to be quite wrong. Hardware leaks.
Initially, hacking via hardware leakage—known generally as a side-channel attack—would seem pretty much impossible given the complexity and speed of a modern PC. Complexity makes for some very, very noisy leakage, which makes for likewise very difficult signal processing. Meanwhile, the actual harvesting of that leakage has to be done at a speed faster than the target computer's clock rate. The requires the use of expensive and specialized lab equipment.
Side-channel attacks typically require long, uninterrupted access to the target machine
The third major barrier is actually accessing the device to be hacked. Side-channel attacks typically require long, uninterrupted access to the target machine. This is a trickier matter when that target is someone's primary computing machine (versus just some smart card).
Nonetheless: "We have identified multiple side channels for mounting physical key-extraction attacks on PCs, applicable in various scenarios and offering various trade-offs among attack range, speed, and equipment cost," the Technion researchers write.
The three methods the group employed targeted acoustic data, electric data, and electromagnetic data: the sound the computer makes (known as "coil whine"), the changes in electrical potential across its physical chassis, and the electromagnetic radiation it gives off.
The hacks were effective in every case, but not equally so. Extracting an RSA key by analyzing coil whine required around an hour of up-close recording, while monitoring electric potential just took a second and could be accomplished with a simple touch. Monitoring radiation, meanwhile, has the benefit of only needing an antenna—in some cases nothing more than a basic AM radio antenna.
With the data actually captured, the rest is basically just signal processing. Methods for extracting real information from even very noisy sources are plentiful and not particularly obscure. It's mostly just software.
So, what now? Well, keep in mind that these hacks still require a hacker to target you specifically and to also be reasonably close to you for possibly an extended period of time. Guessing passwords on the internet, this is not. But still, the paper notes a handful of fairly obvious countermeasures, including hardware muffling and shielding schemes and introducing ciphertext randomization into the software itself.
"To fully understand the ramifications and potential of physical side-channel attacks on PCs and other fast and complex devices, many questions remain open," the paper concludes. "What other physical channels exist, and what signal processing and cryptanalytic techniques can exploit them? Can the attacks' range be extended (such as in acoustic attacks via laser vibrometers)? What level of threat do such channels pose in various real-world scenarios?"