Internet Engineering Task Force 109, 16 Nov 2020

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: IETF109-QIRG-20201116-0500

Description

QIRG meeting session at IETF109
2020/11/16 0500

https://datatracker.ietf.org/meeting/109/proceedings/

A

All right, whilst uh we wait well, let's begin the meeting. uh Whilst we wait for some people to join and while we'll go through the introductory slides, we are looking for a minute taker and a jabber scribe. If you would like to volunteer either join the chat, well say so in the chat or let us know via.

A

A

Your muted rod.

B

uh The unmuting took took several seconds, otherwise you may have to volunteer somebody.

A

Yeah, so while people are thinking about volunteering, I'll go through the usual note, well,.

A

Well, as seen on the screen I'll give it a few.

A

A

Few more seconds for the next.

A

A

And the goals of the irtf.

A

All right so, while people are joining before I start with the rest of the chair, slides, we are still looking for a jumper scribe and a minute taker being a jumper scribe is pretty simple. uh Oh bruno just says he doesn't see the slices, nobody see the slides, or do people actually see the slides.

A

There's a presentation view button at the top.

A

A

Well for bruno there's a presentation button on the top, so maybe that's it uh anyway. We're still looking for jabber's crab and minute taker.

A

C

A

Right jab prescribe: this should be actually fairly straightforward in the online meeting and minute, taking should also, hopefully not be too difficult, given that we're actually a fairly spaced out.

A

D

B

Bruno yeah, my mute and unmute button has extremely high latency like bruno you're, asking to uh to uh share the screen, which I don't think is what you wanted to do is.

B

B

What you wanted to do, at the top in the right there's a series of icons. The left most of those is presentation view not your own not to share your own screen, but to to it should be right about the middle of your screen at the top, or at least it is.

A

A

uh Nobody, volunteering for minutes taking.

B

Shelter can do it.

A

B

You want to be the uh note taker.

B

I see him online, I'm not sure. If he's hearing us or not, let's see.

A

Can't see slides or rodney or hear audio.

B

That's interesting.

B

Is there audio, I can't hear anything.

A

Seems to be working okay for me, yeah! Some people do hear us, because people do react in the chat to some of my questions.

A

Let me just re-share the screen see if that works.

A

Are the slides, visible now and we're still looking for a minute taker? Oh on this shutter shutter still hasn't confirmed.

A

B

uh Bruno says he can do the meeting minutes great uh then we need to so. We need a job.

A

uh Yeah, although jabberscribe last time wasn't uh well was somebody just needs to monitor the chat screen.

A

Okay, uh thanks bruno, so if somebody does actually want to monitor the chat screen and then let us know if we're missing something that would be great uh otherwise I'll just move on.

A

uh So the agenda for today is to spend the first 10 minutes on things like slides, do not work people not hearing audio and other online issues. uh Hopefully, we've got most of that out of the way. Now uh we'll then go for a talk about measurement device measurement device, independent quantum, key distribution by josh slater.

A

uh It says 20 minutes, but it will actually be 30 minutes as a typo.

A

Then chongang will talk about the use case draft and I will talk about the principles draft and then we'll have an open flow.

A

What's the status of the rg uh shortly after the july uh meeting, uh we basically during around the time of our july meeting the department of energy in the usa, announced uh the usa blueprint for the quantum internet uh and with colin's uh encouragement uh rod, and I wrote a blog post to promote the qrg activities. The blog post is online. It's on the ietf blog uh links are in the slides uh and yeah. uh It's still on my wall. It's still online so feel free to share as well.

A

And also the fact that the usa has announced that blueprint for the controversy is quite exciting. uh It's clearly showing there's momentum in this field, so it's great to see that uh other than that. What's the current document status, uh the principles is basically almost ready for uh was basically, I think, ready for last call, but we'll see today at the meeting.

A

The use case draft is almost there as well we'll be discussing some on that today and there's three expired drafts which are expired, but people do every now and then keep asking about them, and there is uh they are of interest. It's just that they'll have to at some point be picked up again uh when somebody finds them interesting again. Other than that uh this is my android's proposal for what we will be doing uh subsequently, because two of the active documents are kind of being wrapped up.

A

So we will rather focus now on education, we'll try to focus mostly on talks like the one that will be coming up from researchers on current topics.

A

uh We might shift to fewer ietf meetings in favor of more interims, which are just shorter and more about just getting up to speed with more with, what's currently being worked on, uh although that is not to say that internet draft rfc work will stop, we can still keep doing that. uh Just that, that's going to still keep happening. It will still be meeting at ietf meetings with that said, uh rodney do you have anything to add to that.

B

Nothing in particular just in my helicopter flying overhead.

A

A

Right so I would like to also introduce our first speaker uh josh. Would you like to share your slides? Whilst I introduce you.

A

So josh later, doctors later has been doing quantum mechanics for over 15 years. Now he started in the area of quantum computing as a quantum algorithm scientist for canada's d-wave systems in vancouver and then the institute for quantum computing in waterloo.

A

Afterwards, he decided to head to graduate school and made the switch to the quantum internet uh in 2014. He finished his phd in calgary canada performing several important experiments in quantum communications, quantum entanglement quantum repeaters and quantum photography.

A

For example, the first field demonstrations of mdiqd, which you'll be talking about josh, led the team of students that built uh calgary's, three-note quantum crypto network and after his phd he did a postdoc at the university of vienna in austria, uh where he worked on the interface of quantum sensing and quantum gravity uh and so having had a whole range of experience.

A

In quantum he joined q-tech in the netherlands, where he was a quantum scientist engineer leading one of the engineering teams focused on raising the technology, readiness level of quantum communications and building quantum internet deployments.

A

Let me now grant his screen and I will let josh do the talking now.

D

Great thank you for check. Thank you for the introduction and uh good morning, everyone yeah, my name is joshua slater, I'm at qtac in the netherlands. So it's about 6 a.m. uh In the morning right now- and this is my first time being at a the qirg or irtf- I etf any of these uh sorts of meetings. This is kind of exciting and fun to see what happens here.

D

I I was asked to give a little bit of a talk about mdiqd and I said well what kinds of things in particular might this group of people be interested in, and I received a list of, I think eight or so different topics and and questions so and a lot of them were more about quantum internet than uh mdiq itself. So I decided to say a few things about uh quantum internet too.

B

So josh you're uh you're not sharing video. I don't know if you intended to or not so so we have your slides, but we don't have the view of you sure I don't know if that is up to you whether you want to share it. But I know your camera works because we saw it during the test session.

D

Because you saw me earlier, I can share video here, we go there, we go good, looks good. Thank you. Thank you for the nice compliment, okay, good. So um what is the quantum internet? I guess everyone in here knows, but just to be clear, a quantum the quantum internet will be a quantum communication network that will exploit the properties of light. Whether or not you know about qubits, it doesn't really matter as long as you know, this is what we're trying to do, exploit the properties of light and that's for a number of different reasons.

D

For instance, there could be users that say I have important documents. I want to authenticate it. There could be users that say I have some particular chunk of quantum data and I want to get this into a quantum computer that will need a quantum internet to do this and there might also be quantum computers situated somewhere in the quantum internet, and these quantum computers might want to connect to other quantum computers share, entanglement share quantum states for some kind of cloud quantum computing.

D

These are the kinds of uh people we imagine will be using the quantum internet, so that was a little why we want to build it. Let's say, let's start with something a bit fun, which is the top seven facts just so we're all on the same board here. The first is that the quantum internet will use quantum technology to provide quantum services to users.

D

I think we all agree on this. The next one, which I hear sometimes some people ask me if the quantum internet will replace the conventional internet and, of course the fact there is no quantum- will not replace the conventional internet, the conventional networks. It's only going to supplement it with new functionality.

D

Third fact is: our communication channels are definitely going to be optical. We have no way in in the physics world to make qubits with radio waves. We don't know how to detect qubits with microwaves all that well, so we are definitely looking in the optical domain and we will definitely be using fiber optics. In particular, we need low loss, especially right now, low loss, optics fiber optics between locations, typically something like 30 db, with no conventional active elements between them.

D

We can make just about everything: that's needed for the quantum internet into 19-inch, rack compatible modules. We do this for a number of our modules. Others are doing this as well.

D

We are going to need infrastructure locations and those infrastructure locations will have to be responsible for the support and for the security, for instance the energy, for instance, cooling.

D

If we are talking about security which we'll get into a little bit later, then they need to be responsible for things like access controls and logging, and then there are some questions there. Some some open questions still like what kind of specialized technology or specialized um support can we expect from these locations like dry, cryo, cooling or compressed gases?

D

If you have opinions on that, I think it would be great to know these feel free to drop that into the chat they can drop things into the chat, while I'm speaking right, okay, yeah great, uh also redundancy. We can build redundancy into the quantum network. I could ask this a lot.

D

um We can do this, though the best techniques haven't been explored yet, and the eighth not really a fact, but there's really a lot of uncertainty. Still we don't know who's going to be building the hardware for this. We don't know where it's going to be laid down. We don't know how much control is going to come from agencies like governments, for instance, and we also don't know how we're going to get to a full-blown quantum internet. Nevertheless, we have um something of a road map that has been developed at various stages.

D

Qtec had a particular paper on this about two years ago, which inc the way we lay out. This roadmap is going from certain low-level functionality like just point-to-point, cuba, transmission and increasing quantum functionality over time, so we're down here right now, point-to-point cubic transmission. These are commercial products.

D

You might be aware that you could buy these things and these commercial products allow for things like quantum key distribution at the moment without end to end security and that quantum key distribution is for applications in encryption authentication and whatever else you might use a key for what is slowly coming out now.

D

Ready for industrial development is multi-point to multi-point qubit transmission, so that is for things where you have a larger network and you can do end-to-end security across this network still up to some distance limitations which I'll get to um again for encryption and authentication, but other applications like password verification position, verification there's a few other things.

D

Then we know we have to be developing things like quantum entanglement and quantum memory. There are early field trials coming out in many locations, in particular the netherlands, in 2021, and eventually, as quantum computers start to connect to the quantum internet. So there's really a a large increase in functionality that physicists engineers need to provide before we get up to this stage. But once we get up to this stage, just as quantum computers start to arrive, we start to gain access to things like blind quantum computing in the cloud which will be a big application.

D

That's where you can do processing quantum processing, in particular without the actual quantum computer, knowing what it is you're trying to process all right, but there's a bit of a gap here, as you can see, which means we're kind of ready to roll out these things.

D

But it's going to be some time before we hit up to more general purpose, quantum internet, and so what we imagine is going to be happening is there's going to be two kind of distinct but similar types of quantum internets that are developed. The first one is quantum key distribution networks which has end-to-end distribution of conventional crypto keys by a quantum key distribution.

D

There is some limited distance, limited quantum distance, and thus we need trusted nodes to forward information along, but we can largely do this with today's technology and then later and how much later is definitely an open question right now, there's going to be this second, the quantum information network with end-to-end distribution of entanglement for things like conventional crypto keys, but also running these algorithms on quantum computers.

D

When we get there when we have quantum repeaters, we'll be at unlimited distance and or but right now, we're still at very early field trials.

D

Quantum networks are emerging worldwide. To what extent this is uh everyone is aware of this. I don't know. As I said, there are commercial boxes available from companies in switzerland, south korea, china and the uk. You can buy these things, um they generally work up to 100 kilometers maximum, and thus they need again. These trusted nodes, china is, has launched their qkd trusted satellite. It also distributes entanglement at very low rates and they've been building this 2000 kilometer network using multi-hop trustynos, that's going from shanghai down here up to beijing.

D

It looks like a very nice online, but it's actually composed of many individual trusted node hops along the way. Yeah there are companies in the united states, like quantum exchange, also building 20 20 mile. I always want to say kilometers there 20 mile long links between new york, new jersey and in europe. We have organizations like the quantum internet alliance or kia, there's the open, qkd consortium, building test bed networks in various locations around europe and also in europe.

D

There's there's the quantum communication infrastructure initiative, or sometimes virtuous euro qci, a lot of people involved there. So people are building. um I heard roy check talk about how there will there's a use case document, so I guess uh people are aware of the kind of applications that we're talking about, usually along the lines of secure data connection between two buildings at q-tech. We are.

D

This is our plans for the next about three years, the little quantum internet that we're going to be, laying down we're right now, working on the dolph to the hague link where some of the hardware should be rolling out um onto this link later this month or or in december, and just to say so.

D

Qtec itself is a mission-driven institute and our mission is to develop scalable prototypes of quantum internet with local quantum processors, enable quantum communication, so we're more than just a research institute, and that's why we are specifically tasked with building uh setups like this.

D

We, it's always nice to say something about your home institute. This is actually the campus of tu delft. This was our very first, let's say quantum internet deployment where we had one user or node over here in the qtec buildings and one user node. I think this is about 800 meters away, if I remember correctly down in uh down in another building and a central node provider over here in engineering.

D

This was used for the first time ever demonstrating entanglement in an irrefutably proven way, no loopholes or anything like this. You can read about this. In the 2015 paper yeah- and this is what we're working on right now- so that's the the hague to delft, link we're going to have a central provider over here in another town. It's called ricevike. This is done in conjunction with kpn, like I said, we should be deploying hardware in 2020. just to give you an idea.

D

This is what the hardware that's going into the n nodes here in delft and in the hague. Look like they are 19-inch um rack mounted boxes boxes all made by qtec and inside is a combination of our own off-the-shelf objects and our own custom, electronics and software, and then later in 2020, probably near the end of 2021.

D

Sorry later in 2021, uh these end nodes are going to be upgraded to some quantum computer prototypes yeah good. So I told you a little bit about qtec. I was then asked about mdi qkd in particular, so I want to spend a little bit of time in saying just some general things about quantum key distribution boxes, what they look like and what they do before.

D

I tell you about mdi qkd and hopefully uh convince you that it's a good protocol to consider and then, if there's still time, I say a little bit about quantum heat resolution networks. What might they look like? What might they do so qkd in a nutshell, qkd, it will involve, say something like transmitting devices and receiving devices to transmit and receive optical qubits over standard fiber. This is going to happen.

D

These qubits will be detected immediately at some kind of receiving station to create quantum data, and this quantum data can be processed post-processed into cryptographic, keys, and the key thing is that any eavesdropping on the qubits or on the signals on that optical fiber network, the hubits travel over- will be detectable by the qkd qkd devices, and then this qkd can be used by things like classical symmetric, encryptors and decryptors for secure data.

D

Communication can also be used by, in other things like authenticating documents or authenticating messages in general yeah, but what's in the box, so there's more than just sending qubits. Of course we will. All.

D

Our boxes will have a cubit channel so boxes that send out qubits into the network to be received, but these qubit channels perform best on dark fiber, so they only operate up to some distance limit or or loss limit, let's say 20 to 30 dbs, typically, and while we can make them coexisting with say, wdm channels or conventional signals, any of these signals will decrease the performance and the operational distance of our of our qkd boxes.

D

We in general will also need a quantum control channel for some, let's say, quantum physics, control of certain hardware elements. There are things like lasers inside of these boxes that need to be frequency, locked or stabilized.

D

There are well that's the most important one, let's say, and so we will need something like our own quantum control plane for exchanging data and again we can mix this wdm with the qubits that will decrease performance, so it is generally better if it runs on its own fiber and of course, we also generally use a conventional control, plane or conventional control channel for things like enabling the network delivering service across messages, etc, etc.

D

And again, wdm is fine wdming that with the quantum control channel is fine, so we can mix these together without problems, but it's generally good to keep the cuba channel separate.

D

Trusted nodes, I'll just say so, trusted nodes. We need these because of this 20 to 30 db limit the way we get around, that for people that want to service over 20 over more than 20 to 30 dbs is by essentially having an intermediate.

D

What we call trusted node, which receives qubits, translates that to just classical data and then translates that back to qubits, which sends over an optical fiber again to another receiving device, and what that means is that if you are using this trusted nodes for things like qkb or the exchange of cryptographic key, what that means is that your cryptographic key is visible for some time at this interface at that trusted node.

D

And so, if you have many of these trusted nodes, well, you need to trust every single one along the path, and this could be a problem, but at the moment that's the best. We have for solving this distance limitation another or the first. Let's say, big question that many of us in the field have been asking is: what exactly is the service we should try offering with qkd boxes? There are a few things that we often talk about.

D

One is the idea of just offering people pure quantum hardware boxes, so we supply basically only transmitting boxes receiving boxes and we say we can provide quantum secure keys via our boxes, but you're responsible for whatever you do with the keys all the security hardware, the security software around the boxes, you're responsible for the fiber. This is one way we could go.

D

There are talks about um interfacing with particular software and hardware security providers, so that we provide the actual application hardware as well. So we provide the encryption we supply our quantum boxes and also things like encryptors, but still say: okay, user you're responsible for the boxes, you own the boxes, you're responsible for the network, you run the network, but look we supply all the hardware and there's also talks with various organizations about maybe the best way forward.

D

Is we actually offer managed services or something operate, something like an msp and we provide the entire secure communication network. We supply the fiber, the physical network and we simply connect to users who request particular services of us, and then we accept their data and send it across our network. So this is an open question if you have thoughts on that, also love to hear them.

D

We have aside for this. There are some other open questions that we generally have in the field as well like what kind of quantum secret key rates will be expected of us. I know this depends very much on the use case and what services are requested, but at times even ballpark or order of magnitude estimates would be helpful for us going forward. We also generally don't know what quantum service quality of service agreements should look like in the quantum domain.

D

Slas we've had talks with organizations over what sla should look like for quantum in terms of things like priority key lengths, availability security parameters comes up in quantum as well. What kind of certified security parameter we should have, and also I think, what's interesting- is that all of this is likely to be pairwise defined.

D

These are open questions and then also generally, what what security level are? We aiming for business, confidential, government, confidential government secret even- and this has important implications on how we as a community go forward with uh with building the hardware, because, obviously, if we are aiming for government secret levels um that imposes a lot more constraints on the hardware, for instance than say business, confidential, okay, good. So that's a bit about quantum key distribution boxes and a bit of the open questions that we have.

D

Let me tell you now about mdi qkd, so there are lots of different protocols for qkd, I'm sh, you've, probably heard of bb-84 bbm 92, there's many others gg02 cal dps, and there are many ways to compare them: technological readiness level, typical key rates, maximum distance, the security proof, because they're not all fully proved, at least in terms of quantum security, whether or not they can be easily upgraded to quantum internet networks in the future and how susceptible their their hardware is uh to attacks. For instance.

D

These are all things to consider in general, most qkd protocols are point to point, which means, if you have a lot of users, let's say in this case four users that want to communicate to each other. They need either dedicated point-to-point links. Something like this. They all need transceiving hardware, sorry sending hardware and receiving hardware, or they require something like a trusted node or at least some node in the middle. That can reroute signals around something like this. There are a few protocols, an mdiqd that get around this issue.

D

Mdiq pity stands for measurement device, independent quantum, key distribution and you'll see why we call it measurement device independent in a little bit, but the idea behind it is that you have one central node, all your users connect to this. It is a way to go point to multipoint.

D

It is a way as we'll see in a second to more easily upgrade to quantum information networks and along with entanglement based and vbm92. It also gets around a lot of the susceptibilities to attacks on hardware they're, just completely eliminated yeah, so mdiq kitty, we generally say you've been arguing. This is a bit more practical because it's inherently networked into things like star networks.

D

Users only need a fiber link to the central node. Any pair of users can then create secret key. So it's immediately point to multi-point with minimal fibers. It's also more cost effective, so the users can be added anytime. They want just by adding a single connection which is good and also at least at qtek. We design all the hardware so that all the expensive hardware sits at the central node.

D

So there is one major hardware: investment to build a central node to kick off your network, but then there's only inexpensive stuff that lies around at all the users, so adding in more and more users is much much less costly and at, like, I said, upgradable feature I'm going to get to that, but it's also more secure, and the reason for that is this: central node is not a trusted. Node, the central node never has access to the key.

D

It never has access to any sensitive information, and this is not just by hardware design. This is by design of the physical protocol itself. In fact, if the central node attempts in any way to gain access to the key that will appear like any other eavesdropper sitting on the fiber and the users will immediately know this, which means they don't in principle, have to trust the central node and the best that can be done.

D

Even if you give that central note over to an attacker, the best that can be done is denial of service attack the endpoints themselves.

D

They are only send only units and not vulnerable to receiver attacks, and that is a very key thing, because so this is a particular table that comes from the etsy white paper from 2018 on implementation, security of quantum cryptography.

D

So inside this etsy white paper, they listed a large number of security issues against quantum cryptography systems against qkd systems, trojan horse attacks, multi-photon emission et cetera, et cetera, the first, let's say, block of these, and it's about the first four of these are against the senders, and all of these can be avoided by what's called privacy amplification, so pa privacy, amplification pa pa pa.

D

The next block of these are all against receivers, the detectors and there it is much more tricky. So there, the techniques to protect ourselves against these attacks involve things like active monitoring like detector symmetrization.

D

The problem with active monitoring, of course, is that well what happens if someone attacks your active monitoring, what happens if someone hacks into your active monitoring or takes control of your active monitoring, then this is no longer a good security measure. What is explicitly listed throughout the document is look mdiqd, mdiq, kitty. Mdiqt is intrinsically a way to protect yourself against a large number of these attacks.

D

So we always say: mdiqd is more secure.

D

For this reason, and the other point is this: upgrade ability, so I'm not to what extent everyone is aware of this- I don't know, but in general, for quantum internet, almost every architecture at the hardware level for quantum internet has things called quantum repeaters say at two locations and these quantum repeaters about halfway between them, actually very close to halfway between them is a midpoint station that does some kind of bell state measurement. So this entire block is essentially what functions as your quantum repeater.

D

You need two users or repeaters with quantum memories and your midpoint station, and the key thing is that this midpoint station is identical to what we use in mdi qkd.

D

This is intentional we've, because qtec is looking long forwards into the future. We are designing our midpoint station, our central node for mdi q3d, to be identical to the midpoint station needed for quantum repeaters, which means that, if you put let's say you build exactly this set up here in a star with your central node or midpoint station, these users can be upgraded in the future. Just by adding in quantum memory and a little bit of extra control.

D

Hardware and software can be upgraded to quantum repeaters and that's exactly what we're going to be doing in the dutch quantum network. In 2021 after we build after we build this section with mdiqd, we will be throwing, in some quantum memories, uh to upgrade it to quantum repeaters yeah. So what I also like to say and argue a lot is that it is important, especially given how much money is likely to be needed to build the quantum internet.

D

It is important that we consider the longer end goal, that we want full quantum internet with entanglement distribution, and we consider how we build even qkd networks now such that we can easily upgrade them in the future to entanglement distribution, quantum information networks. We know things like we know: midpoint stations are going to be needed. This is almost almost a certainty and we know that um we know from simulations from from analytic work that asymmetric link.

D

So if this midpoint station is closer to one side than the other, that already degrades performance uh quite quickly, as this starts to become asymmetric, so laying down qkd networks, midpoint stations and symmetric links will do a lot uh going forward into the future.

D

Good. Okay, um I'm going to skip over all these nice, this last section with all these nice graphs, but I just wanted to say there are some other open questions which maybe would be interesting to people in the room like whether or not we should.

D

I said at the start that all our quantum internet systems and quantum cryptography systems work better over dark fiber, but whether we should immediately be thinking about wdm multiplexing, we don't know which is the right direction to go. Should we optimize fiber use essentially with multiplexing or optimize the quantum performance. We don't know point to point is point to point sufficient say in the short to medium term, or should we go point to multipoint very quickly with the early networks, these test beds that are being built around europe?

D

It's a question there. To what extent are trusted nodes allowed they're, certainly going to be required in some situations. If anyone says we want a huge network spanning 800 kilometers well, we know we need trusted nodes with, but if your network is only to span 100, kilometers 200 kilometers, or should we avoid trusted nodes anyway? It's likely that users and governments will decide this, but this certainly plays a role on how we build, because protection measures are going to be needed everywhere.

D

We have questions. We talk about um questions in terms of enrollment. How do we get more people into an existing quantum network? We want to avoid things like vendor lock-in, so I guess we need some standards at some point.

D

Should we allow anyone to connect, there's going to be exchanges at some point ixes, but in the quantum domain should we allow anyone to connect to them? How should we deal with say, networks that connect to these exchanges that start acting suspicious, and what do we do about lawful interception?

D

There's a real question there. If we're going to say that we're offering um a secure network an unconditionally, secure network with unbreakable communication, what happens when we actually have to break it? We don't know these are open questions, though good. So just a couple takeaways, the quantum internet, it's going to start with quantum key distribution for now and full quantum internet will come later. This will take time. I think it is likely to be 15 to 20 years.

D

Probably people have been saying that forever, but 15 to 20 years until we really get to to full quantum internet, our qtex, the dutch quantum network. We are starting up next month and upgrading to repeaters early next year we will have a test bed, there will be. We have a project called quantum q e quantum network experience where people can log in and use the network that will come online at some point as well.

D

There are many open questions that as physicists, we just don't know how to deal with. Like what service to offer wdms trust in those network management, for instance, there are many qkd protocols, I will say again at qtec. We like mdi qkd because of its better security, it's upgradability and we believe cost scaling, and I strongly believe we should be designing the physical topology for these networks now, so that we can easily roll out cue quantum information networks when they come about right.

D

If you want more information, uh this paper here in science from 2018 this laid out, the the qtec roadmap for quantum internet feel free to check out that out. That goes into much more detail. Some of the things I said, and also, if you like listening to me, you can google me on youtube joshua slater qripped. I think I have a 90 minute talk on more details of mdiq pity when I was younger and more younger and less secure. Let's say, but yes, so that's there as well.

D

If you want more details anyway, thank you very much for listening.

A

Great thanks josh uh great talk, so uh questions.

A

I actually see something on jabber from bruno. uh Why do we need qkd or pkc to replace classical public key infrastructure in the first place? Yeah something about that.

D

Yep yep- I normally have some slides about uh about this, but the the the issue is that the issue is that when quantum computers come about, there is this shores algorithm, which will allow breaking a number of algorithms that are used in conventional cryptography, not the symmetric encryption algorithms, but the asymmetric algorithms that are used for key distribution and key encapsulation, and so the problem becomes when these quantum computers are about.

D

If rogue governments or rogue agencies people, we don't trust, have access to them. If they can, they will be able to break the existing the current today public key infrastructure we use for key distribution, which essentially means they'll control the communication network. Honestly I've heard that from other people as well in security, and so we need some better techniques to protect that quantum, that key distribution and one way to do that is quantum key distribution.

A

Right uh we'll have time for one question: somebody has one.

B

I always have questions if nobody else does go ahead rod. Actually, this is actually a fairly low level, simple one. um How long are the links you're? Actually installing? I gather that um one of the links on your four node ring that you drew in the netherlands is actually pretty long for for the current uh technology.

B

D

uh You're talking about going back to this one yeah um yeah, so this link here so okay, so we actually don't have the uh the fibers yet for any of the links other than the first one, so the the distance that this will be or the loss that this will be. We don't know at the moment, but at least this first link uh dolph to the hag. So this covers um 20 db.

D

So it's I mean yeah, I say 20 db because of course uh fiber. That's in the ground is never a straight line and uh of course there are many uh connection points along the way uh that add in extra loss. So we at least know that this link is, is 20 db yeah. Our expectation is this: one will also be 20 db and then these ones will more likely be longer more loss. Yes, but we'll see.

A

Thanks, yup, okay, great uh so thanks josh, uh but whilst we switch over to the next uh say, which is chong gang with his slides, uh if anybody has a question, whilst we switch over uh feel free to ask it I'll stop your screen share. Josh.

A

Okay, uh thanks josh uh hope everybody found that interesting. Oh, I have something from joey salazar. Would you like to unbeat yourself? I think you have to do it.

A

A

I can't hear you.

A

A

B

uh Josh john joyce says: he'll uh put it here on the chat. Okay. There was also another question in the in the chat in the chat from showtime. He said josh. Did you say that the qkd networks have e to e end to end conventional keys.

B

I'm not sure if that's I'm not sure, if he's asking about the authentication of the the nodes or.

D

uh So the the qkd networks can provide end-to-end conventional keys.

B

Okay, so uh it's not, it provides end-to-end keys to the end nodes, but but does not have e to e confidentiality or it does have, and and confidentiality.

D

Okay, I think that's, but that might be a discussion because I'm not exactly sure how short is defining confidentiality. Here I mean. Certainly there will be.

D

It is known that it would be easy to see where the traffic is coming from at any given point, so you would know uh yeah I mean you would know who is distributing keys with who this could be figured out just by monitoring classical traffic, but the key itself would not be known to intermediate nodes, at least in the case of mdi qkd and entanglement, based and and other protocols like this.

D

But yes, okay,.

B

So the uh your midpoint uh bell state analyzer node. There.

B

Does not have to be trusted, but you get secure end-to-end keys across across that right.

D

Yes but and then okay.

B

So hallman.

D

Baker just said: there's traffic analysis attack and I completely agree you would know you would know which users are at which times trying to generate key with each other. That said, the keys would then be storable, for instance, and you don't you wouldn't know from that point when these keys are getting used in principle,.

B

Okay um white check: I think we need to move on we're already running a little over time.

A

uh That's fine if we run a bit over time, we do actually have a bit of leeway this time, uh so we won't have to cut anything short. Okay, thanks thanks, josh, again, uh there's a question you can answer in the chat uh of from joey, but in the meantime, we'll switch over to john gang.

A

E

Hello hi, uh can you hear me real.

A

Yeah, I can hear you.

E

So I assume you're sliding fire right. Yeah, okay, looks good! Okay! Thank you, uh hello! Everyone. um uh My name is cheong gangwon. um So today, I'm I'm going to share with you uh a kind of a short presentation about the status and also the recent updates uh to our uh applications and the use cases for quantum intellect uh document. um So major uh basically want to share with you our primary and major updates after last meeting.

E

B

Background, hello, sorry, um are you, do you know that you're not sharing video.

E

Yeah there you go that showed up. Okay, okay, thank you. uh Yes, let me come back.

B

All right that looks good. Okay, thanks.

E

uh Yeah thanks so kind of just a recap: um the uh little background about the current version, uh o3. So, basically back in may. um This document has been accepted as a research document um and then in july um iutv meeting last meeting, we presented the version zero one uh and then received quite a few good comments and feedback, and after that uh we generate the version learner 2.

E

To address those comments we received from ietf108 and also this version 0.2 has been shared through the email reflector and also received some comments, especially comments from patrick now, ojic or logitech, um so the current version 03 and it's kind of if we compare to the version zero one we have uh presented in us ietv meeting and under the next slide that will show you what's the major updates uh we did after last ietf meeting, so we kind of reorganized the six and six which is uh above the general requirements.

E

We split this section into two subpar subpar, uh two parts, uh 6.1 is a background and 6.2 is the requirements uh in addition to that. um Based on the feedback from last meeting, we ended a new paragraph about the management related requirements um as a part of this 6.2 trying to see uh from the.

E

um If we have the quantum internet in future, we need to manage and correlate uh the quantum resources um either for a quantum node or for the whole quantum intellect and the quantum resource could be, the quantum memory and the quantum code bits and also could be, the quantum collection being established. Among two um quantum notes.

E

uh That's the part we added for from management perspective. There is a requirement, and then we and also based on the feedback. From last meeting, uh we added the two paragraphs in section nine secured consideration basically to discuss privacy related and issues.

E

uh Although one of our use cases applications is about the privacy preservation like the blind quantum computing, uh some terminology we also kind of based on the email discussion um in order to avoid any confusion and make the terminology more um consistent with the principle documents, we replace the quantum sub network with a quantum network, just called quantum network again here are some editorial changes throughout the whole document.

E

Just to increase readability.

E

I didn't put it over here in the slides. I saw that we already have discussed in an email, but overall these three types of application or use cases are being described in this documentation. uh One is uh based on the code kd. We describe the uh see the secure communications based on quantum k distribution, uh the second application we call the blinds quantum computing. The third one is the distributed quantum computing and that's kind of three um applications being um included in stockmans.

E

uh So uh next step, um this slides kind of uh I pull put a goodreads slides um like before the last friday, and after that I received some very, very good feedback comments from logitech.

E

So I I think the next step is going to um address the comments from audiotech and also uh if there are some uh furthermore, comments. Feedback from the team who already here uh we would like to uh you know, continue to uh address those uh comments, feedback to improve the document. Furthermore, uh that's our goal, um so I think that's kind of uh I don't know if I need to show the um content of this uh the recent document or not, but I would like to um hear any kind of question or feedback from you.

E

um This is the last slide of my.

A

Presentation all right, great thanks, chuang gang, is: are there any questions from the floor.

A

A

uh Joseph did you mean to enable your mic instead of your.

A

Camera joseph here.

A

Okay, uh there are no questions at the moment. I have actually a question, so I did actually so I wrote that email on friday and uh I'm I don't know, I'm still a bit confused about your classification of control, plane versus data plane applications in a quantum context. uh I find the distinction fairly clear in when you draw your analogies in the classical context, but I'm not entirely clear what what what your definition of control plane is in the quantum context, could you actually explain that a bit more.

E

Maybe okay sure uh thank you, what'd, you think uh maybe I'll show that table. I'm.

F

F

I think it comes here. Yes.

E

uh Can you see this table? um Yes, okay, thank you. So here we're trying to do is uh use this. Of course we have the text description. uh I saw this table probably uh maybe easier for us to uh discuss uh so here we we have the three columns here like a classic internet examples in terms of control, plane, data plane uh and then in middle. We have. We call these standalone quantum network examples.

E

uh Those could be when we see standalone quantile network, basically like isolated, just one quantum network and then uh the third column here uh quantum intellect uh in future. If we have the uh you know uh the uh the fulcrum internet, what will be examples in terms of control plan and data plan applications?

E

So let's look at the classic, intellect uh that's very clear. I I I assume uh we have the icmp or dnase those kind of application or pro code they are used uh to. You know, enable the data plan application like a web browsing or email whatever, uh so that so basically control plan here is not for the end user. It's for to facilitate the data transfer data transmission among end users uh and then a standalone quantum network example here.

E

So we were trying to use the same, um not principle same methodology, trying to see if we look at the a single quantum network. What will be the operations? Look like the control plan for classic intellect what would be the operations for the next equivalent to data plan for classic intellect.

E

So here we were saying for the one of the major operation for the um one single quantum network is kind of.

E

We call the entanglement distribution in order to establish end-to-end quantum collection amount to users, that's kind of the the the major purpose for the quantum network and then for doing that um we needed some support, some signaling to control the entanglement distribution, and then we we put over here, trying to say oh, it's kind of similar to the smp or dnas like because the signaling here, it's not for themselves, is for the quantum and for the entire distribution.

E

So from this perspective, we thought it might be a advanced example when we talk about the control plan data plan, but I know there might be something a little bit different from like like we do here for the classic internet, um and then we, when we look at the quantum interlades and this this third economy, I think, is a cinema. More clear here is no matter what we let's say: we have a cocaine.

E

We have some, uh even if uh the the repeater economy, repeater based uh qrn economy, information network, but the data plan is still like very similar to the classic internet could be uh like the video conference, but it has a you know: better security using kokidi based secure communication setup to allocate the quantum case, for uh let's say just one, uh both parties for this uh late conference, it's more secure, that's a data plan application uh and then, for this uh case in order to support this application on the data plan, the control plan will be the cocaine uh or or the quantum theme um you know, in order to uh serve or facilitate this kokidi based video conference.

E

That's kind of the um uh what we were uh thinking about is control plan data plan for for different types of uh on internet classical or quantum.

E

I don't know if that's clear or answer your question, would you take or.

A

uh Yeah, it does so. Basically it's now that you've explained it. It's kind of really clear to me now it's just. I was just really confused for some reason from the text, so maybe I'll just go reread the text and highlight which parts uh confused me, but actually you do raise an interesting point here, which I I don't think has occurred to you before. Maybe just me is that uh when you say quantum ping, I think you imply that some of the quantum traffic also counts as control plane.

A

E

Correct, uh yes, and also this quantum pin actually is contributed by the comments um I remember from patrick or someone uh basically trying to. I think it's a good comment. First of all, it's kind of uh used to test if a quantum collection has been established between uh a pair of quantum nodes, right, yeah, right yeah, because the pin for the classical intellect is trying to say. Oh, if I can reach a ipad address or not, I I will just escape yeah yeah.

A

uh Okay, no thanks, that's clear to me uh thanks for that, are there any other questions from the.

A

Floor, something in the chat, this question might not be completely appropriate, but I was wondering document for guiding the layman, rather than a quantum expert also should the distributed quantum computing section include an example of enhancement gained from distributed quantum computing.

E

uh I don't know if I sorry if I fully capture that or not, could you uh it's in the chat.

A

As well, it's also in the chat, if you just move to the chat window, the question.

F

A

Might be easier that way.

F

A

You and yeah scroll down.

F

This one right, yeah excuse me.

F

F

Example of the enhancement getting from.

E

uh The I think this has one two questions for the first one. um I I think uh there might be uh two um relevant purposes here.

E

One is to um you know it could be as a good document for the namers who are interested in this field, but they are kind of, are you know, kind of challenging for them to dive into the details immediately and then, instead of that they can read over this use case document trying to get a high level uh information like the app what what what quantum inter application will look like. I think that could this document this document could serve that purpose uh relevant.

E

In addition to that is, I think, even for this coirgy, it's more like when we do uh like in the printation from the logitech. We have the principal documents and also there were three uh documents being expired, but people also are interested in that. I think this use case document could be a good input and a complimentary document for uh later on when we standardize or we do a detailed document about different quantum intellect uh making them like the quantum collection establishments and so on.

E

um That's my answer for the first question, but I think probably chairs rod and the logitech could have more insights uh for this question. um The goal of this use case document, but that's my my response. uh The second part should really should a distributed. Quantum computing section include example of enhancements gained from distributed con computing. I think a second question, if I understood correctly, ask us to show some benefits or what we can get out of this if we use distributed quantum computing. uh I think it's a good question.

E

I I think somewhere we put some benefits in the document described uh by using distributed quantum computing, which kind of new features could be enabled by it, but I will double check uh and then anyway, we are going to have a next version 04, so I would object and keep in mind uh this um second part yeah. I don't know if I answer your question michelle or not.

A

I think I just have something to add to the last part. uh I think so there are examples in your document. uh I think uh it kind of relates to the feedback. I also wrote on the email on friday uh that actually, I think, with a bit more detail. It might be a bit clearer as well.

A

Okay, yeah we'll do that, and also for the document, it's probably on its own, not for the limit, but in conjunction as chong said in conjunction with the principles document then yeah, I think it'll be good, for that should be good for the networking expert, not complete layman, to networking.

A

There's a question on the chat from joey salazar: if you scroll down a bit or also from rodney, but I don't know rodney, do you did you have your question answered or.

E

Not yet um yes for the tcp, I think the uh the think packet I I personally would consider it as a control plan, um but you know tcp. The safe is a pro co. It's very hard to say uh it's a control plan pro core uh data plan protocol. I think I underlist that part compared to the segment or packet from the user. I think that from that perspective I would like to say it's for the control purpose.

E

B

Okay, so in conjunction with that, I'm not sure why the third and fourth columns are actually separate. Why is there a standalone quantum network example and a quantum internet example? Why is the distinction of control plane versus data plane different for the two types of networks.

E

uh So like these two columns right um yeah, we could we could merge them together here. um The original idea here is trying to see if we have a standalone quantum network, it's not really starting to support large scale, quantum application yet and then what will be the major operations for this standalone quantum network? We thought that could help us to understand the full operations uh from a small scale, quantum network, and then we move to the large scale.

E

Quantum internet, which has um full-blown uh features to support like cocaine based or some other quantum feature based applications. But we, if we, if you think um it's better to have just like classic interlade and quantum intellect um we could um merge the third column and the second column.

B

I think for for control plane for actual control, plane work. There probably is a substantial amount of work at the border routers, but for for the use cases document, I'm not sure that that even the existence of a border router should actually matter. We should be able to run these applications in theory across the border, although maybe not all networks will support all functionality in the early days.

B

So I think it's more a matter of when you cross a boundary that that changes functionality as opposed to just an administrative border.

E

uh Yes, I think that's valid, that's about it from the use case perspective. I think that's about it. We are kind of not to touch uh like the uh structure or you, the term use a borrowed related or dependent, uh so maybe the the better way is just to maybe for this table just has two column class interlate and quantum intellect.

B

That's my current opinion, but I should probably reread this and think about it a little bit more, but that's that's my current opinion that they should be merged.

E

B

um Yeah, it's now nine after we have 31 minutes remaining. I think if I, if I'm looking at the.

A

Clock four hours we have two hours for the home. Oh.

B

We have full two hours, we go to the top of the hour. Okay, we have plenty of time. I was thinking.

A

uh So at the end, chongang asked a question: is whether documents ready for group last call I'd say it needs at least one iteration, but obviously open to the rest of the group and rodney.

B

I would agree it needs a little bit more too and I should probably provide the same kind of feedback that wojtek did, and I haven't done that yet.

A

But I think it's shaping up quite well and it'll be a useful addition. I do.

B

Too, I think this is going to be a very useful document.

E

Yeah, we will be very happy to uh receive further comments from rania and from the team and uh we'll we'll be updating the document. uh Based on the recent uh comments and the questions, uh people just asked during the uh this meeting, yeah.

A

There's a hand raised from phillip helen baker if they still want to ask it.

G

uh Yeah actually make comment on the security considerations. I think it's important to freight to frame things correctly and what you have here is not cryptography. It is transmission security, which is a separate field from uh cryptography and presenting this as the solution to quantum computing's attack on pki. I don't think is particularly useful for two reasons.

G

First, kerberos is quantum resistant. You know, we've got a 45 year old system that is actually a key distribution system that is resistant to quantum cryptanalysis.

G

The second reason is that communication security is what the field is able to do today. You know, if you look at the breaches that are happening today. They're, not communications breaches, you know, tls is taking care of that ipsec. To a lesser extent.

G

The problem that we have is securing data at rest, and none of what I'm seeing here is presenting me with tools. That's going to secure data at rest, so it's interesting work. I think, there's a valid use case for a quantum internet, but framing it in terms of you're going to solve the problems of the security world. I I don't think is helpful.

B

Can you point to uh specific places in the document where you think that that language needs to be updated.

G

uh Well, it comes in in the security considerations, part uh which you know, I'm a security.

C

Guy, so that's.

G

The part that I read first uh but but it's also you know um they it's trying to present this. As I mean the thing is that the word transmission security doesn't appear in the document, but security is used repeatedly and also from the previous uh uh presentation.

G

I think it's an unhelpful way of framing things. I think it's important that we get the framing right, because what you've got here is basically the killer version of transmission security and quantum is really really good for that and will surpass anything like frequency hopping or whatever and within the you know,.

C

G

Limitations uh that people who go for transmission security today, um you know it's probably even- is easier to deploy than some of the other techniques, but presenting it in terms of this is a replacement for pki. I think really is a bad strategy.

B

This conversation probably becomes long, but but do you think pki or uh pqc is actually does that actually protect data? That's at rest.

B

G

There are techniques that are being developed that are looking at uh data at rest, specifically my work, which is threshold key infrastructure, um so yeah I mean, like pki, isn't ideal for data at rest. You need to extend it a bit uh by splitting up the keys, but it's still pki speaker on steroids.

B

What is sorry, what did you call yours, the the not threshold infrastructure.

G

And the point is that you're not going to be able to take this framework and extend it to data at rest, which is actually what we can? That's, what people are really being concerned about with quantum and doing things like uh all that stored data, the data that we encrypt today that we want to keep secret?

G

That is what is really scaring us, because.

H

G

If I ain't encrypted top secret or compartmentalized security document, it has to be secure for 30.

G

B

I understand what you're saying I'm I'm not sure. Yet if I, if I agree that the document actually needs updating, it does talk about secure communication, setup.

G

Well, it's a question being clear about the types of security that you're providing she provided right: you're, providing transmission security, uh you're, providing confidentiality. uh We have to look into the traffic analysis issue. It's not clear to me that you don't provide any traffic analysis uh control, but we'd have to look at that.

A

uh Thanks for those comments, phil, those would be very useful also on the mailing list, uh so that others could process uh them and they're right.

A

I think that would be greatly valued as well. So if you can also write that on an email summarizing what the comments you've made so that they could be acted upon, that would be very good.

G

Okay, I'll do that.

A

Thanks a lot uh we'll have to move on now so I'll hand over to rodney to chair, because I'm presenting next.

B

All right, I think, the the the last uh or next to last agenda item other than open mic is uh wi-tech's um architecture, principles for a quantum internet. Talk why tech you want to go ahead and share your slides, I'm shang! You can turn off your video if you want you're gonna.

B

um Let's see, I think we put 20 minutes on the agenda for this, didn't we yeah, but we have about 40 minutes left in the in the session. So if it runs uh long, that's okay, although we do probably do want to reserve some time for open mic at the end.

B

This draft also should be pretty close to ready to go for, for at least the rg last call before it goes to irsg. So we should be starting to look at things in you know the final level of detail, including word smithing, and things like that. I think, okay, why tech um I'm going to turn my camera off? When you see me pop my camera back up, that means you've been talking for about 18 minutes.

A

Thanks uh so right, so we've been working on this document now for almost two well a year and a half. Now I guess uh so. The recap. uh The document I'm talking about is architectural principles of a quantum internet uh linked below it's on the data tracker.

A

uh As usual, uh it's been worked on since ietf 104 in prague back in 2019, and the key motivation of this document uh is to basically, we have a charter point uh to create an architectural framework, delineating network, node roles and definitions to build a common vocabulary as the first step towards network architecture, and that was the motivation for which I first actually submitted the first draft and also uh it's also educational, to a large extent, and I think in the end it will become that more uh as well and because it's a great start, I want it to be a great starting point for people with more quantum background, and I think it'd be a very useful document for the rest of the ietf and iot community.

A

If it can fulfill that goal, uh this fyi the document is on github, uh it's. Basically it's just as an update. Just that's where all the changes are tracked and I only update the data tracker before the meeting so that there's an official data tracker update, but I will just update github in between, except that, if it's approaching near completion that will be not happening as often uh I wanted to actually do a bit of also, uh I guess showing off like what I've seen.

A

uh uh I, the document has actually been cited, uh what's been cited primarily by people affiliated to the authors or the authors themselves, but I found two papers that are actually citing uh the internet draft that were not affiliated to any of the authors. So I found that actually quite interesting, given that this document was not written or presented.

D

At academic conferences.

A

To be cited, so it's actually also a good argument to actually expedite publishing it as an rfc, given that internet drafts are not a permanent record and also just like from my own conversations with other people, I often actually share these documents to networking people as a quick introduction uh and they often like it so uh well, as I point out, it's obviously a biased sample, but I hope actually it does serve the same purpose for other people who have a much stronger networking background than they do have a quantum background anyway uh to the actual core material.

A

What are the changes since last meeting, uh I added a control plane and two data plane subsection to kind of explain our distinction between classical and quantum data plane as.

C

Well, as what did we define as the control plane.

A

But, as was pointed out and based on also on our previous conversation in this meeting, is it needs kind of alignment with the use case document uh so that we're all consistent on what's defined as the control plane and what are the data flames we're talking about?

A

uh I added a note based on some feedback on the mailing list, that passive optical elements uh are allowed and that's actually quite important for going forward uh with the network architecture, because uh it kind of emphasizes the fact that we can. We can switch entanglement over short distances without quantum repeaters.

A

uh As long as we do not use active optical elements, uh but that's obviously, then you're bound still by the distance uh over which you can transmit uh photons without loss, uh as josh mentioned in his talk, uh I also added some additional clarifications and some a point that was kind of long.

A

I've been kind of putting off for a really long time, which is adding references to academic literature, uh they're now all added uh so hopefully the document is, apart from just a slight uh aim: attempt to synchronize the use case draft with this in terms of control plane. It should be complete essentially, as a question is the draft ready for error. Rg last call and that's kind of it for me not much in terms of presenting addresses for questions.

C

A

C

Any last clarifications.

A

Before we move on with official steps for last call stuff.

B

All right, I didn't have to wait until 18 minutes to pop my camera back up.

B

Questions comments, complaints.

B

Criticisms advice agreements, arguments.

B

B

B

B

All right, yeah, white tech- I actually have a comment or two, mostly some small ones. um I went back through my list of uh comments that I sent back in february.

B

Compared to this.

F

B

And um I know we went through some of these on the mailing list, but I don't remember the status of everything every each individual.

B

Concept or each individual point that I had brought up, um did we end up with discussion of the first second and third generation networks in this.

A

Yes, uh contribution to the document.

B

Oh yes, I remember that which, which section is that.

A

That's under error management.

A

There should be something like that.

B

Error management, 4.4.3.

B

Quantum error, correction.

A

I'm trying to load it so I can also share the screen with others, but it's going very slow.

B

uh The couple of paragraphs that are here are fine, but it doesn't really uh well. Yes, it does yeah first generation second generation third generation is here: error management schemes; okay, okay, uh the.

A

Other specific.

B

One we had was, it was um actually kind of a minor wordsmithing stuff. I didn't particularly like having tomorrow in the title of one of the sub-sections, you talked about support for tomorrow's applications in 10 years or 20 years or 30 years. What counts, as tomorrow,.

A

Kind of white I feel like tomorrow's big but yeah that can be modified. There's a suggestion on the mailing list happy to.

B

Change, I did not go go through everything on the mailing list to see sort of how we were relative to each individual item on the mailing list. I was assuming you were taking care of that.

A

Yeah yeah, I I kept track about that. I think I've updated addressed all the.

A

Issues, I guess in the absence of other comments, does that mean the group's positive about having lost colin moving forward with the rfc, barring some awards may think and.

B

One thing I do think we should do before we declare either one of these ready for, for our last call is. We should actually sit down and read them in parallel and make sure the two of them. Don't don't uh disagree in any substantial, um either technical uh or uh or terminology basis. I've read them both.

B

You know with days or weeks between readings, and so so I haven't tried to actually do that.

A

Yeah uh well, I guess perhaps because I actually did actually say in this presentation that uh there still needs to be alignment on the control plane, which addresses your thing that these documents should be consistent with each other, uh perhaps I'll. Let's keep this question of last call for the mailing list, uh rather than answer it right now, because I still need to update that control, plane section at least or this make sure it's consistent.

B

I do want to give it one final, complete end and read before I think before I I'm ready to sign off on it sure that's on me, as as both an author and and as as the chair.

A

uh Cool then that's it for me, there are no questions from the group.

B

B

B

All right, let's ask how many people in the group have read the uh this architecture document. Let's see, what draft is this? This is. This is a version, five right, um how many people have read either a draft four or draft five.

A

uh How about we use the show of hands tool uh so that.

B

Should be a way to do that, I don't know exactly how to do.

A

It yes, so there's: okay I'll.

F

Ask the questions.

A

Are you specifically interested in that version for version five.

B

Well, I think it's probably good to ask how many people have looked at it relatively recently. I think the update between four and five was not huge, so I think either four or five is probably good enough, but you know some of the earlier ones. There were pretty some some pretty drastic changes between you know, 000 and 01 and o2.

A

Okay, so let's see how this works. uh Oh I've just asked the question and I guess now: participants have to raise their hand or not raise their hand.

A

I think for those who are looking for it, I don't know how it shows up to other people.

B

Yeah, it shows up to me: okay,.

A

So I guess, if we wait at least for least half, or at least until it stay converges stabilizes for those looking for it. I assume it's on the left show off hands. I don't know how it actually shows up for other people.

B

Yeah, I think it showed up differently for me because I'm chair too, but it showed up on his participants, chat and then raise your hand, was the left most in the panel with the participants in chat window directly below the raise your hand, join the cue button.

B

Looks like a dozen people have actually read it and that's actually pretty good. I think.

B

A

Okay, well so about that many, let's take it for a while. This can stay on the walls we discuss it. 12 people have read it. So I don't know.

B

Diego uh and uh let's see diego and joey both said, they said that they read it and found it useful that they like it.

A

Thanks uh yeah.

B

I'd say if we've had a dozen people actually read it recently, then we've had sort of a good set of eyes on it, but I do want to give it one more complete, read next to the other, before I'm ready to sign off.

A

Sure sounds good, uh I'm not in a particular rush but yeah, hopefully maybe before the next actual ietf meeting that we do attend uh well.

B

I would actually like to see it for, for our last call and sent to the irs g within the next couple of weeks. If we can achieve that because uh by the first of december, where we're we're moving into end of the year mode, where, where people's distraction level gets high,.

A

Yeah, shall we move on to the open floor then, or is there anything else that you want to say on there.

B

That's all I had on on the the uh on the draft.

A

uh Cool, so the last item on the agenda is open floor, which I guess is an opportunity for everybody to ask any questions they have like. I, I think all questions about the quantum internet should be askable, also opinions on what the curator should do next in light of wait, let me actually share my initial slide for the next uh steps that I proposed.

B

Let's see, josh is still here also so so, uh if anybody has questions going back to the first presentation on mdi, we can also go back and talk about talk about those.

A

uh So if anybody has any suggestions of uh other than what's kind of concurrently on the screen of the next steps for the qrg opinions and also just any questions, uh they can be technical questions as well as to how does quantum stuff work, etc. uh Groups meant to be aimed for collaboration and kind of education is part of their collaboration, so yeah so open.

A

A

H

Hey, can you hear me.

A

H

Good good, um so unfortunately I missed the beginning, but uh it was uh wondering if uh prototyping and testing was uh one of the next steps. You know to try and validate some of the the ideas and the drafts in in experimental implementations.

B

uh That's actually a really good question for either joshua or wojtek. Is you guys, are actually building a network and does your implementation actually conform in some useful fashion to the architecture document.

A

uh So yes, thanks, colin, that's a great question. Yes, so josh! Basically, as he said in his talk, uh there will be a first link up uh between delft and the hague by the end of this year. uh It's going online so that what that is one example uh more practically in terms of how something that directly feeds off from prog work.

A

Then, yes, next year, so josh also mentioned that next year, we're upgrading to quantum repeaters. uh And yes, the protocols we are implementing over. There are inspired by all the discussions have been happening here uh so and like for, I recently uh uploaded to the archive an article of the network layer protocol uh that we will be implementing uh on the actual uh setup, for example, uh and there's also the uh a ln clear protocol that has already previously been published.

A

Essentially, all of like especially the network layer work feeds off our discussions in the qrg and we are going to be running actual tests and prototyping of that early next year, and we will be publishing that soon and though, as josh also mentioned, there will be q, a quantum network experience which will be a web front-end for people to actually run their own quantum distributed applications on top of an experimental network at some point uh next year. Hopefully so, yes, that's a good point to raise we'll keep people updated and posted about. All of that.

B

H

Posting a link.

B

H

Yeah, it would be good to get some some implementation results for experiment. Experiment reports in the group as well.

A

Yeah we'll have some very soon.

B

And josh clearly knows enough to speak the uh the networking lingo so so josh is definitely a keeper we want. We want him around as much as we can get him.

D

Thank you very much rodney, that's nice, to hear. Thank you.

B

I did have a question going back to that, which I think is actually related to show. Does uh wojtek may also uh have some comments on this. You know shot and I were both sort of initially a little bit confused about the the whether the mdi central node actually has to be trusted, or not. um Part of that goes to the definition of what counts as a node, which, in some of the work that we've done, we've sort of waffled back and forth about that.

B

So your midpoint node that serves as the bell state analyzer we have debated off and on for years as to whether that counts only as part of the link and therefore is not visible in hops. Further down there further down the the path or, if it's actually a full first-class network node, and we had decided at one point that we thought it belonged only to the link.

B

But we had not looked at at this multi-point switching star-shaped kind of of architecture that you're talking about for mdi more recently for some of the work that we've been doing on graph state quantum repeaters which we haven't published. Yet we have decided that that the the the bell state analyzers need to actually be pretty sophisticated and need to participate in the end-to-end communications.

B

So our opinion on whether those are nodes or or or whether they're nodes that are visible at the network level or whether they're just part of the link, our own opinion about that has changed in the last year or so I don't know. If either of you wants to comment on that.

A

Right so you're saying the midpoint, should I be active participant in the network now there's some more experiences.

B

Bruno also commented on that on the in the case of mdi qkd, the middle node facilitates star shape, so it also includes mems cross connects and things like that, it's more than just a bsm, it's more than just a bell state measurement device or a bell: steel analyzer. uh Yes for mdi.

B

I think it is, and also for the graph state based repeaters, which we haven't talked about yet at all in qirg at all, but I think for both of those cases, the bell state analyzer probably has to be a full-fledged network that participates in routing and security and appears in the network, topology and all those sorts of things. Whereas you know a year ago, our opinion was that it shouldn't.

B

Speaking as an individual, you know running running our own research group.

B

Josh said in the comments here, he also believes that the midpoints and center notes should be active participants because it's acting to bring point-to-point uh point to multi-point functionality, yeah yeah, so the mdi use case actually brings that up pretty clearly to to uh to how important that is. In that respect, I suppose.

A

Yeah, if I the way my thinking has been going, is that the bell measurement station, the midpoint is a very so especially the star networks that mdi qkd will work on and josh showed.

A

I find to an extent an analogy between local area networks and layer, two networks and what can be achieved with such a midpoint station, uh where we have a kind of local area network around a single bell state measurement station, where we can have multiple end nodes connected to one and the switching is done by the valve state measurement in the middle.

A

And this involves no entanglement swapping well, it does, but no quantum repeater entanglement swapping uh just that midpoint station and then, when we want to go beyond these local area networks, we need gen, proper quantum repeaters, with quantum memories which can perform an entanglement swap on a stored, qubit and start qubits and that's kind of closer to our what we know as layer 3 network layer networks.

A

So I do actually agree that perhaps the midpoint well, not. Perhaps I do agree that the midpoint is an active element and will draw analogies to local area networks and in fact, most realistic deployments. Early deployments might actually be just multiple end nodes around a single midpoint station, at least practical deployment. Talking about experimental deployments will probably be a single central station and multiple end nodes.

B

B

Other questions anybody have another topic: they want to bring up.

B

Can one cascade and one cascade midpoints?

B

um You can certainly cascade optical switches if your end-to-end loss is low enough, but the end points with the measurement devices themselves or sorry. The midpoints, with the with the measurement devices themselves, probably not.

A

Yeah the way, I understand the question, I'm going to say: no, because you need a very good synchronization between the two end nodes using the submit point in order to.

B

uh uh Kiriti the extended the explanation a little bit if abc, connect to md1 and md1 connects to md2 and then md2 connects to def. So now, in this current hardware, the mds only receive right. So your connection between md1 and md2. There it's only going to receive, but if you put another, if you put node g in between md1 and md2, then you would have a large network with multiple hops and then you're doing multiple hops of mdi, where you have multiple memories involved.

B

um Josh could probably answer that.

A

B

Bruno says: you need quantum repeaters to do multi-hop, md iq qkd, which that certainly agrees with maya.

B

My understanding at which point um at that point, you're doing um entanglement based qkd. You know maybe like the eckerd protocol, the e91 protocol rather than pure mdi, and at that point you are doing end to end, but you also have to be able to generate end to end entanglement, which is still the hard part.

B

um Josh, did you want to you want to pop in voice wise and talk about what you were just talking about there.

D

Yeah sure sorry, it was just uh in regards to the discussion of whether midpoints are um active in the network. So the what I was writing was that the the chinese network is essentially a long connection of um almost like super super nodes. They're trusted nodes or something like super nodes, and then many users can connect to these super nodes uh in almost a star-like fashion.

D

So one way to see this network is really a large, a chain of midpoint type stations with um each midpoint station having its own spar network around it.

D

Yeah and I believe the the american network um between new york and new jersey uh will have a similar physical topology physical construction, which which then to me, uh adds to the argument that the midpoints are are active elements inside the network.

B

Who's leading the new york to new jersey work and is that is there something public about that.

D

This is an organization called quantum exchange.

B

Okay, I know then um joey also has a question in the chat here josh. You might be the right person to answer that.

A

uh I'll just read it out loud uh if the measurements and qubit operations during the secure communication setup referring to the use cases, draft and follow-up from ietf regarding data description can be potentially used to identify truck the user establishing that setup. So, basically, is the qkd connection. Setup.

A

Anonymous and I would uh I would say well it's as anonymous as normal control, plane traffic. To be honest, it's actually not unless you protect it, yeah. No, it's not anonymous. Unless somebody corrects me.

B

It would be interesting to try to build something anonymous on top of it, but the most obvious implementation, I think, is not.

A

To get anonymous communication, I think so for quantum key distribution. I think it would be difficult to build anonymity on top, but there are other applications which do quantum based distributed, quantum-based applications which do guarantee anonymity, but we have to go beyond quantum key distribution networks into quantum information networks uh and they go into very.

A

This is beyond my expertise, because it's application level for quantum, but there are things such as like anonymous quantum anonymous transmission in the quantum domain, where you can transmit quantum data without your receiver, actually knowing who it was transmitted from uh so there's scope for much more anonymity, but it has to go, but it's beyond quantum key distribution.

B

um Before we get to the last couple of minutes here, let's see is colin still here if colin's still here, I'm pretty sure we did this uh last time, but since we're since that was several months ago, and we're closing in on final call for for these, uh for these two drafts colin, do you want? Do you want to to uh quickly toss in a word or two about about what the process needs to be to finish these off.

H

uh Yeah I mean that the process for finishing the draft self is that you, the the group and the chairs, come to the uh conclusion that the work is done uh and then you send it to uh my myself.

H

uh There's the you can do this in the data tracker and then I arrange for the irsq to review it and give feedback um and uh most likely there will be some comments and the draft gets revised as a result of that, there's a check that it doesn't conflict with any work in the ietf, which I think in this case is unlikely and then it gets published as an rfc.

B

B

B

Any other uh comments or questions uh we've got about 10 minutes left. If we want to use it or we can give people an extra 10 minutes to to break for before before their next.

B

B

Bruno we'll look forward to your uh to your notes.

A

Yeah thanks for the menace taking, uh I think that basically wraps it up I'll stay online for the next 10 minutes myself, at least. uh If people want to ask questions, but overall I'd say, the meeting is basically over.

B

I think we can declare it closed um the I do think by the way we that we should actually try to meet face to face the next time all of iatf gets to meet face-to-face just because it's been a nice long time. um The next, let's see march, has already been declared to be online. Is that right.

A

I think I don't require to be online.

B

So hopefully, in july next year, hopefully we'll see everybody in july. I think that's optimistic.

A

B

I think march is right out, but I think, but I think july has a reasonable shot of actually happening.

B

All right, thank you all then, we'll see you we'll see you on on the mailing list and hopefully in person sometime middle of next year,.

A

B

Everybody thanks wojtek.

A

Yeah thanks rod.

B

And thanks colin and josh uh and everybody who uh contributed.

B

Bye y'all, I'm stepping out.

A

C

Can can you hear me joshua still now.

A

I can hear you.

C

Hear me: okay, thank you. Just wanted to check no questions or anything just wanted to to make sure that my mic was actually working.

A

C

Thank you for staying, um I'm actually just using the space here for the chat a little bit: okay, okay,.

C