Turning Telecom Fiber into Quantum Secure Links

Summary

Noel Goddard, CEO of Qunnect, a hardware company working to transform standard telecom networks into quantum secure communication networks is interviewed by Yuval Boger. Noel and Yuval spoke about 1st and 2nd-generation quantum secure networks, what one would need to do to create a secure link between Chicago and New York, the intersection of physics and telecom network engineering, and much more.

Read the full transcript on The Quantum Computing Report here

Transcript

Yuval Boger: Hello, Noel, and thank you for joining me today.

Noel Goddard: It's a pleasure to be here.

Yuval: So who are you and what do you do?

Noel: My name is Noel Goddard, and I am the CEO of Qunnect. We are a hardware company located in the Brooklyn Navy Yard that's building and innovating devices to transform standard telecom networks into quantum secure communications networks.

Yuval: When people speak about secure quantum communications, they speak about many different things, starting from QKD to purely quantum networks. I'm guessing you're focused more on the QKD, on the key distribution side. Is that correct?

Noel: Actually, we're on the other side. I think a useful way to think about the technologies right now for quantum communications is that there's a first generation of technologies, which have actually been around for at least 10 years, that are intended for hosting quantum key distribution protocols.

And then there's a completely new set of hardware and protocols coming to light now, which have come out of the research that people are doing towards the quantum internet. So when people start to talk about that, they're typically talking about distributed entanglement protocols.

In order to host those, you need completely new sets of equipment, and the protocols themselves are things we've seen in laboratories, but not necessarily commercially yet. Qunnect is on the latter side. We would consider that to be more of a second-generation technology. First-generation quantum key distribution still communicates with single photons, the same way that we intend to communicate with single photons.

But the difference is that the protocols currently used in QKD are point-to-point protocols. Depending on whatever protocol you're running in the background, like BB84, the user is sending a string of photons between sender and receiver, and, depending on the protocol, randomizes the information, right?

They do it in order to basically exchange and sift, to get a key. In distributed entanglement networks, it's a little different, because it's not intended to be point-to-point in the same way. What you're doing is taking a pair of photons which are entangled and have some information on them, and distributing them between sender and receiver.

In both cases, I think the major challenge in the field has been the fact that we would all like to do this over fiber, but fibers inherently have transmission losses. The longer you try to extend transmission of a single photon through a fiber, the more chance you have of losing it. So something in the 50- to 70-kilometer range is usually the limit people think is useful for having a single fiber link that can carry a photon without significant loss.

When you hit that distance, you need to figure out how you're going to extend it. In QKD, or the first-generation technologies we were talking about before, the way people have been extending the distance is to create a duplicate set of hardware at a node, which basically decodes the process, makes the information classical, then re-encodes the information into quantum, and then sends it again down another segment of fiber.

Those relay nodes have commonly been called trusted nodes, because they need to be highly secure in order to make sure that the process stays secure. But actually, they are vulnerable, which has been the reason that agencies such as the NSA in the US have come out against these types of technologies, because they believe they can't truly be end-to-end secure enough to justify the cost of upgrading the hardware. In quantum networking, the protocol that allows you to extend the distance is something called entanglement swapping.

It allows you to basically project the entanglement of pairs of photons which have never seen each other onto a shared set. And that allows you to gain distance, stepwise, between the original origins of your photon pairs. That process itself, because there's no measurement in the process, is theoretically secure.

Nobody has really shown this in a commercial sense yet, so I always hesitate to say something like "unhackable" or "totally secure," because we don't know how things will work with other types of side attacks and such that people can figure out. But basically, as long as the information remains entangled and there's no measurement, the information is secure.

And therefore, I think there's great promise with this new type of protocol, like entanglement swapping. But the problem is that the hardware needs to be invented. So Qunnect's mission in the past two years has really been bringing the hardware that supports these protocols to the market, and we're not quite there yet, but we've made some good progress.

Yuval: In terms of bandwidth, this is still, at the moment, a low-bandwidth technology, right? So when I spoke about QKD, I meant that your technology would be used to swap the keys or to set the keys, but not for the actual transmission of high-speed information. Is that correct?

Noel: I think the big question mark at the forefront for everybody right now is: once you build it, what are you going to use it for? With QKD, it was very clear that the purpose is to create a key. The reason people talk about keys being the first use case has everything to do with the fact that quantum communications is never going to win on bandwidth, but it will always win on security.

So you need to send something which is sensitive on the security front but can afford to be sent slowly. Keys are a great example of that, because after you exchange the key, you can send whatever information you want. You can wait as long as you want for the secure key. And after you have the key, you and I can exchange information with that secure key in the background.

I think we all forget that digital communications had to go through a growth process; it wasn't always a high-bandwidth thing. One of the challenges we've always had in trying to explain why you need to invent new communications protocols and innovate hardware to host them is that we've all enjoyed the luxury of being able to stream tremendous amounts of data on something like our cell phones now.

But in the 1960s, nobody was thinking that was possible when they were sending the first messages between two computers. So I think we just have to accept that there will be first-, second-, and tenth-generation technologies that eventually start to improve bandwidth. It's just hard to imagine today what all the use cases could possibly be.

Key distribution, I think, is a nice first use case.

Yuval: So let's walk through the technology. Let's assume that I'm in the Chicago area, and I think you're in New York, so the distance is more than 50 kilometers. We want to have a secure link between us. We don't want that security to be regenerated, so we don't want to convert classically and then back to entangled photons.

How does the network look right now, and what would it need to have, with your product, to allow us to establish this secure channel?

Noel: First off, the fiber that exists in the ground today is fine for transmitting the types of information we're talking about, as long as you don't have things which are essentially measurement devices on the fiber.

They often call this "optically transparent," but all they mean is that you don't have things like classical amplifiers or classical repeaters on those strands. So the first thing you need to do is find some fiber between you and me which is optically transparent, and then every 25 to 50 kilometers — let's say every 25, so we can sell you more instruments.

But every 25 kilometers, you're going to need to have a rack of instrumentation. Let me start off by saying that to do any type of entanglement swapping protocol, the key devices you need are: an entanglement source to generate a pair; a quantum memory to hold one half of that pair while the other one goes to do something in the network;

and a swapping device, or Bell state measurement device, in a middle node between two sets of entanglement sources and memories. So basically, imagine that every other node has entanglement sources and memories in it, and the nodes in between will have swapping station–type hardware.

At its most basic level, that's a Bell state measurement device. However, in order to make all of this work in a network, you need more than just the quantum devices, because everything has to be really tightly regulated. One thing you need is a distributed clock. Every single node is going to have to have a clock that is able to correlate with the others.

And they have to do so with sub-nanosecond time resolution, which is more than what digital needs right now. Another requirement is that all of the devices in each of these nodes, in order to communicate with each other, need to use very precise wavelength references.

So you're going to need a wavelength reference in every node that has an entanglement source, so that you know it's going to work okay. And finally, because we are sending photons to each other over fiber, we have to recognize that the fibers themselves are subject to environmental stress, and they can change the photon that we're sending.

So on top of that, you need some type of real-time calibration tool, which allows you to test the fiber segments at different times of day — during rush hour, or when it's freezing, or during a heat wave — and to compensate for small changes in the fiber, which are actually going to change the information, because we're trying to communicate with single photons, right?

Every photon counts. So again, there'll be a rack of equipment every 25 kilometers or so. Every other one will have entanglement sources, and between them, there will always be a swapping station. Of course, there will also be some other swaps to make this a hierarchical network that actually functions, but that's what you can imagine.

The way we think about the protocol is this: imagine you have two entanglement sources in nodes flanking a center node that has a swapping station. The entanglement sources are putting out pairs of photons. Half of each pair is captured at the node in the memories, and the other two go to the swapping station.

They're doing this continuously. Whenever you have a successful swapping event, the swapping station communicates back to the memories in those nodes and says, "Successful event. The photons in your memories are now entangled." So each time you have a successful swap, you gain the distance between the two outer nodes that projected to the middle node.

And you can do that across the entire network, or you can do it in hierarchical layers, where you have a bunch of primary entanglement swapping events and then a secondary layer which swaps those. But you quickly come across the problem that you want to make sure your memories can hold your photons long enough to do all of this, and that's something that's obviously still being developed in the field.

Yuval: You speak about quantum memory, but I don't think it's the same type of memory that I would have in my computer, say, 16 gigabytes of RAM. It sounds more like a programmable delay line for photons. Would that be a fair statement?

Noel: It's very fair to think of quantum memories as delay lines, because essentially what you're trying to do is take a photon, not damage it in any way, and then release it in a way that preserves all of the information and quantum states that were associated with the photon before.

There are people who actually try to emulate a quantum memory by putting in delay lines. Our quantum memory is different, because we use a process called electromagnetically induced transparency, which uses a control laser to modulate the transparency of a collection of atoms — in our case rubidium — inside a vapor cell.

So you essentially open a window in the medium by making it transparent to send a photon in, and then when you turn the control laser off, the photon is basically transformed into a spin wave, which uses trillions of atoms to collectively store the information in the photon. The amount of time you can store it is defined by how long you can wait before you again use the control laser to take the spin wave and the information and reconstruct the photon.

We work at room temperature, which means our atoms are moving at incredible speeds within the vapor cell. In fact, we actually work with a heated vapor cell, so it's an even worse-case scenario. But right now, the best we've done in our laboratory is 800 microseconds of storage.

It's a tremendous amount of time when you think about taking one photon, spreading all of its information in a spin wave across a whole collection of atoms, and then being able to reconstitute it with high fidelity. So, again, you hit on the most important part, which is the idea that memories are necessary for regulating timing in networks.

There are repeater schemes out there right now that don't use memories. But when you start talking about entanglement sources, you're talking about things which are non-deterministic. One of the challenges is that you think you probably made a photon, but it's still going to come out probabilistically in terms of time.

So it's really nice to have something in the network which turns something probabilistic into something deterministic.

Yuval: It sounds like a super complex technology with a lot of different parts. You need photon sources, you need to generate entanglement, you need swapping, you need the delay, you need synchronization, calibration.

How close is it to reality in terms of installing a test network or something that demonstrates that this actually works?

Noel: There have been some demonstrations of entanglement swapping in academic laboratories. Probably the most well-known is a demonstration that recently happened in Delft.

TU Delft in the Netherlands is probably the most advanced quantum networking group in the world right now. They've done a number of interesting demonstrations and developed technology to support various protocols, but the ones they're doing right now to support entanglement swapping are using nitrogen vacancy centers in diamonds as their memories.

This all has to be cooled to a few millikelvin in order to function, so you need an entire laboratory of infrastructure around the devices you want to use. But they have actually been able to demonstrate entanglement swapping over fiber between memories that are supported by all of this infrastructure.

What Qunnect is hoping to do, of course, is to show the commercial version of that using our devices. All of our devices operate at room temperature at this point, so there will be different challenges in trying to put these online. As a company, we just raised our first large funding round, which was to support us building out a networking test bed that we can connect our instruments to, so that we can learn how to do this best over real fiber.

Because we happen to be located in Brooklyn, there are quite a few challenges, because the fibers around here are obviously pretty noisy, since we're in a very urban area. But that's good, because then we'll learn how to deal with that in our networking protocols. I think we're probably two years out from really being able to do what I would call a real repeater protocol.

But entanglement swapping, which is one half of the repeater, is a real goal of ours over the coming year.

Yuval: In terms of the composition of the company, I'm sure there are lots of physicists and people who deal with control electronics. To what extent do you need communication experts, or people who are familiar with how telco infrastructures actually work?

Noel: It's a great question. We think about it a lot, because now that we actually have the opportunity to expand, we're trying to figure out what the right answer is. We started off being largely physicists and engineers, because we needed to design instrumentation, and the instrumentation was the first in its class in many ways, so the challenges were very hardware-specific.

Now that we want to connect things, we're starting to branch over into the network world, and I think we all understand that we would like these things to run on the same fiber network that exists today across the country. There's no way to know how that works without including network engineers in the design process as we keep going forward.

I think it's just a natural transition for a company: first we built the devices, now we need to plug them into something, and plugging them into something brings that. The other piece, which would obviously be very useful on the network engineering side, is to think about the control layer.

At some point, if our devices are working well, there should be a software overlay controlling all of the different nodes, and that's definitely something that starts to enter the network engineering world.

Yuval: So in a commercial sense, a Verizon would one day say, "I can offer you special pricing for a super secure link between two endpoints."

Do you see that as a federal customer? Do you see that as banking? Are there particular applications that stand out as being more attractive?

Noel: Also a great question. I think we all like to think of this as a few layers. If you're talking about exchanging exceptionally secure but slow information, such as keys, that's extremely useful for different types of financial transactions, critical infrastructure transactions, and defense transactions.

There's a whole class of problems that fits. But something we think about a lot at Qunnect is: how do you really take something from being just a communications network and turn it into an analogous internet? The digital internet we have today connects all sorts of devices — computers, IoT, et cetera, right?

It does so over a hybrid structure, which allows everything to be a mesh of connectivity. If we'd like to do the same thing for quantum, we need to think about how we're going to interface a quantum computer with a quantum communications network. How are we going to use quantum networks to actually build quantum sensor arrays?

Those types of things are exciting problems that might be on the horizon once these networks are a little more commercial and widespread. At Qunnect, we're also thinking a lot about the fact that the types of qubits you send over these networks are going to be important to standardize.

Right now, there are a number of different types of qubits that different groups use, and a large number of wavelengths that different groups use. This is because quantum has a lot of ways to work right now, but it's always very specific to the material that hosts the quantum effects. In our case, we use rubidium vapor, which is a very classic material for a lot of physics.

So a lot of the wavelengths we use are defined by that, which means the communications wavelengths would be defined accordingly — as opposed to somebody using a different physical system, like rare earths, who would be using a different wavelength.

If you're going to do that, then you need frequency converters, or you need to decide on a standard that everybody plays at. So I think we have to evolve to a point where people can actually start talking about standards in the industry. That's not close, right? That is some distance away.

But telco providers like Verizon, which you mentioned earlier, are watching and eagerly trying to learn so that they can prepare and understand what they may or may not need to do in order to host quantum information in the future. We use polarization qubits at Qunnect, which are traditionally thought of as more of a computational qubit, not a communications qubit.

We do that for a number of reasons. Our quantum memory natively stores them, which is perhaps the most important part. But polarization is also something you tend to get for free out of a lot of entanglement-type processes. So right now we're building all of this equipment to show that polarization qubits can be used over networks, which I think is not something people really believed could be true.

But they are also a very native match to quantum computers and quantum sensors, so we can start thinking about the next-layer problem, which is how we connect devices over this quantum communications network so you're not just sending keys.

Yuval: As we get closer to the end of our conversation today, I wanted to shift away from the technology to ask two other questions.

You mentioned that a lot of the advancements were actually international. You mentioned TU Delft. I read someplace that there was a secure video call between China and Austria. Do you worry that there are going to be export restrictions on your kind of technology?

Noel: Oh, there already are. The world is changing; I think that's just part of the real story. In America, ITAR restrictions already apply to quantum encryption technologies. Qunnect doesn't build encryption technologies. We build all of the stuff that hosts quantum information, so you could think of us much more as the infrastructure that hosts quantum information, as opposed to the end devices that do the encryption.

Right now, you would not be able to export anything from the US that was a quantum encryption device. Philosophically, as a company, we would very much hope that the quantum internet is something that develops in Europe, in the US, and in other countries independently, and can eventually be connected the same way as the digital internet.

But I think its special use cases, which go back to that defense use case, are just part of the reality of where we are as a nation in the world. We understand that cybercrime is basically the crime of our lifetimes, right? And it's only going to get worse once quantum computers come online.

So we'd be naive not to say that, of course, we understand there are going to be restrictions at some point on some things. But for right now, I think everybody's benefiting from it being a more open technology, so that we can all find the best solution globally.

Yuval: I think you are a physicist by training, and now you're running this exciting technology company.

If you could have dinner with someone in the quantum space, dead or alive, who would that person be?

Noel: That's interesting. My training is actually in biophysics, so I wasn't trained in AMO physics, which is primarily what we do here. I started off working in a photonics lab in graduate school, building instruments and doing those types of measurements, and then I saw an opportunity to do a hybrid degree program.

When I went into biophysics, I spent quite a bit of time on the front end of biophysics, also building instruments to measure hard things. So most of the heroes in my world, if you had asked this question early on, would have been people largely in the biophysics space.

But something quite exciting just happened in quantum communication science, and that's the Nobel Prize. Just a few weeks ago, the winners of the Nobel Prize in Physics were announced, and of course they got it for showing entanglement technologies. I've been rather blessed to have had some light introductions to those teams, but it would be really fun to sit down and have dinner with all three of them and to see how special they believe it is that their very basic research science is now something people are talking about as a commercial reality.

Yuval: Excellent. So how can people get in touch with you to learn more about your work?

Noel: We just launched a new website, which we're rather proud of, and it shows a number of interesting videos and animations. There's an animation on how a repeater network would work using our devices, so it's a great place to get started.

We also have good intentions of making animations about all of our different devices. If you go onto YouTube, you will see our quantum memory animation there, and as we continue to roll out products, we'll have more. We thought this was a great way to educate people on how these devices work. As well-intentioned as white papers are, I think they don't necessarily get as much reading as we always hope, whereas an animation is something everybody can appreciate — visual learning is how we do things now.

So that's a great place to start. If you happen to be in the New York area and you'd like to come by for a lab tour, we've been doing more of those recently, so that's another thing — you can always drop us a note and ask. We have a general email line: contact@qunnect.com.

Yuval: Noel, thank you so much for joining me today.

Noel: It's really a pleasure to be here. Thank you for the invitation.