Interview with Ben Levine, Senior Director of Product Marketing at Rambus at headquarters..

Sanjay Gangal interviewed Ben Levine, Senior Director of Product Marketing at Rambus at headquarters.

Sanjay: Tell us about your presentation at the IP-SoC Conference.

Ben: My focus area is on security, in particular hardware security cores, the idea being that you want security embedded in really any chip to provide security to the rest of the chip in the system. So my talk today was talking about that, particularly for connected devices. The fact that everything is connected to the internet these days means that every device is now exposed to a wide range of threats and attackers. So you need really strong security. So I just talked about some of the challenges, particularly around not only devices being connected, but devices being complex, what the impact is on security, and how you can solve some of those problems with our hardware security core.

Sanjay: What exactly is the conflict between the complexity and security and how do you solve it?

Ben: It's a great question. It's something that really came to the forefront for a lot of people last year when the news came out about Meltdown and Spectre, which were those attacks on modern CPUs. Now if you look at a modern CPU from Intel or AMD or ARM, those are incredibly complex pieces of engineering, and they're designed over many years to be very power efficient, to be very high performance, but not necessarily to be secure. And making something that complex secure is just inherently difficult. One way to think about it is if you have a system, you have different components in that system and they interact in different ways, and a lot of times security vulnerabilities happen when there's an interaction between components that wasn't expected or wasn't anticipated, and that's what we saw with Meltdown and Spectre. Different components interacted in ways that the designers never even thought about.

Ben: The problem is, as things get more complex, you get more components, which means that the number of interactions goes up exponentially and that means the number of potential security vulnerabilities goes up exponentially. And the hard thing, as a system designer for security, you have to get everything right because an attacker only needs one vulnerability. You may have protected a million, but if you leave one door open, the attacker will find it and your system's vulnerable.

Sanjay: How do you solve this problem?

Ben: The solution that we think makes the most sense is basically partitioning security away from other parts of an application that don't need to be secure. So we often talk about siloed execution where you have one silo of execution for things that are security critical, so areas where you have to protect keys, protect passwords, protect identifiers, anything that needs to be protected from attackers or from being read out when it shouldn't. Any sort of security protocols or communications protocols, those all need to be done in a secure domain that can be optimized for security, kept relatively simple and straightforward, leave all of the fancy performance enhancing techniques in the insecure world, and then design your secure enclave, your hardware security core, is what we often call them, make that specifically designed from the ground up for security and then partition it off from the rest of the system. And that's how you get away from that complexity problem.

Sanjay: And does Rambus have any solutions or any things that can help the designers handle this problem?

Ben: Absolutely. So we have what we call our CryptoManager Root of Trust. Their current version is called the RT630, and it's a hardware root of trust that we've developed specifically to provide security to any chip, FPGA design any system, and it's a complete hardware security core. It's based on a custom RISC-V CPU that we developed from the ground up for security. So there are no compromises in that CPU. It's designed specifically for security applications. It has hardware crypto acceleration for standard algorithms, private secure memory, lots of protection against various types of side-channel attacks and other attacks, lots of anti-tamper logic. So, it's a very secure execution domain that can be dropped into any chip and provide this sort of partitioned or siloed execution that we were talking about.

Sanjay: Are there new areas for security needs that are coming up now?

Ben: For a long time, the new one that we talked about a lot was IoT, and as I said, connected devices are exposed to a lot of attacks. More and more things are connected and they're furthermore exposed and also have more severe consequences for compromise. But even more recently, an area that we're focused on is AI. So there's lots of interest in AI accelerators both in the data center and at the edge, and there's tremendous value tied up with all these AI applications, which means that there is also potential for attack, and there's some unique security challenges around AI applications that we're focused on. So our CryptoManager Root of Trust, we think is a good solution for AI applications both data center and edge AI accelerators and AI chips.

Sanjay: I did want to ask you some probably non-technical, non, sort of non-technical question, but are there any scientific achievements you're most excited or scared about?

Ben: Well, there's some exciting ones, but there's one since you asked that I'm really scared about as a security person, and that's quantum computing. So, quantum computing has a lot of promise and a lot of great things it will do, but as a security person with my own narrow focus, it really scares me, and here's why: So we have these fundamental algorithms that we use for cryptography, for what's called asymmetric and symmetric cryptography for encrypting things, signing things securely, doing anything involving security or using one of these standard algorithms, and they've designed to be secure so that if you try to say guess a key randomly for an AES algorithm, which is a common encryption algorithm, it would take you longer than the life of the universe to guess the key with a normal computer. However, with a quantum computer, they don't work the same way as a normal computer where you have to go through each individual key one by one. It can have at the same time many different combinations of bits which means that both asymmetric and symmetric but particularly asymmetric algorithms are very vulnerable to quantum computers.

Ben: So when quantum, and I think it's a when, not if, when quantum computers become powerful enough, they will be able to break the standard encryption algorithms that are used throughout the world, throughout almost every type of device. So it's scary. IBM research came out with something recently saying that if you have something that you want to be secure 10 years from now, you need to start thinking about different types of algorithms because the standard algorithms won't be secure against quantum computing attacks.

Sanjay: Does Rambus have anything in this area to help the designers get to the 10 years from now security challenges?

Ben: We are doing some work there. NIST, so NIST, the National Institute of Standards and Technology, specifies the standard encryption algorithms that are used, and they have a contest now to come up with new algorithms that use different mathematical constructs that won't be vulnerable to quantum computing, and they are soliciting input from all around the world, different companies, different academic organizations, and our researchers actually submitted an algorithm that's made it to the second round, and we're doing other research as well. So we're looking very actively at not only what should those algorithms be, but how do you implement them and how do you implement them securely. But it's research at this point. We're not at the product phase yet, but we hope to be in time to stave off the crisis that quantum computing may face.

This entry was posted on Monday, May 13th, 2019 at 12:40 am. You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.