Thinking outside the box: Side Channel Attacks

When we're doing crypto analysis, as in any security endeavor, we have to understand every possible thing that could go wrong.

Let’s talk first about the method that is usually reverted to in order to prove that certain cryptographic techniques are secure.

Logic: Generally, we start with some assumptions. If you don't assume anything is true, you basically can't prove anything. If you don't know any facts, then there's pretty much nothing that can come from that. Then, with those assumptions you feed into the idea of a proof.

If you're lucky or hopefully you have your proof – you can then conclude that whatever protocol you're looking at is secure. But sometimes what happens is that the attacker works outside of the assumptions. They somehow go around this proof and they go to a situation where they break this security. It's called a side channel attack, which is basically doing an attack that wasn't considered in the proof.

Here are some examples of side channel attacks that we've seen in the media.

The first example of a side channel attack is the one that happened in Bitcoin. Hardware wallets are specifically designed to generate private keys and produce signatures but never release the private keys anywhere outside the device. So, people have demonstrated that on those hardware devices that they can, if they measure the signal between device and computer or whatever the devices are interfacing with, actually take that data and recover private keys. To actually perform this attack, you would need physical access to the device while a signature is being performed and taking electrodes and splicing them, etc... So, even though this is considered an attack on hardware wallets, but generally we would consider them to be fairly secure because this is such a major breach; you'd have to lose the device or have the device compromised. Also, somebody would have to figure out the password or the pin to generate a signature on the device and then if they're lucky, doing some signal analysis can recover private keys.

There was another example of a side channel attack that dealt with sound, and the result is if you listen to the sound a device makes when it is signing something, it's possible or at least it's been shown that you could actually recover the private key just from the sound. When security of a digital signature is set up, it's considering the operations the computer's going to do and not the sound that the computer makes when it's doing it. Now, still even with a compromised microphone, it's not a guarantee that a key can be recovered, it's just more of a proof of concept that this can be done. But if you have devices with open microphones, you might want to consider closing them for this reason among others.

Another example of a side channel attack is scripting on PGP. So, there's an email encryption protocol, it's called PGP. You can set up your own PGP setup with software. It's called GPG Going Privacy Guard. There are other software available to utilize this PGP protocol. It's all free. You can encrypt and compute digital signatures with PGP encryption. There was an article that came out a few months that said that if you are using a client that decrypts an email, have your scripting turned on and basically the link gets clicked, then your message could be divulged to a third party. It is considered as side channel attack. It's not really an attack on the protocol itself, it's more of an attack on the implementation of the protocol.

So, anytime you're setting up any type of secure situation, even though you might have a proof and with certain assumptions, it's always good practice to think about what other channels can people obtain information from, and how do you stop people from utilizing those channels.