Dan: All Right, welcome back
to the next Quantum Divide.

I'm very lucky to be joined
by Andrew Lord from BT.

Andrew, I um, your surname, I
feel compelled to call you sir,

I know it's not necessary, but

Andrew: I was called you know when
they take the register at school when

you're a child and they read out your
surname first, so I was Lord Andrew

and every single class teacher thought
this was hilarious and the whole class

did, so every year lots of laughter and
I think it's marked me psychologically

for life, Lord Andrew, I am not a lord.

Dan: Okay if I call you Lord
Andrew during the call by

accident, then you know why now.

Listen, we've only met recently,
but from what I can tell, you're

a bit of a stalwart legend of the
UK optical and quantum industry.

You're a senior manager

of optical research at BT, you're a
BT distinguished engineer, you're a

fellow of the IEEE, and you also edit
one of their journals, or a journal

of some sort, which we can come on to.

But welcome to the podcast.

Thanks for joining me.

And why don't you start with a bit of a.

information on your background,
how you got into optical

technology and then quantum.

Andrew: Yeah, sure.

I graduated from Oxford
University in 1985 in physics.

Asked my tutor, what do
you think I should do?

And I remember him saying I, I hear
BT is getting interested in fiber.

And so I thought okay.

May as well.

So I applied to BT and ended up in
a job working with the subsea area.

So at the time in the eighties and
nineties it, the, it was just telephony.

Okay.

There were no.

internet or anything like that.

The question, the challenge was how
do you get telephony across the sea,

across the Atlantic and the Pacific.

And so a lot of activity around
there for the next ten years or so.

And then the internet kind of made
all of that transmission technology

relevant to terrestrial networks.

So every, all the learning that had
gone on in the industry to make very

long span subsea system suddenly started
becoming useful to build in country

networks around the UK and other places.

And so that was really.

fueling the next the next few years or so.

And then I think probably what
happened next was the amount of demand

for data and capacity coming into
optical fibers was requiring WDM.

Multiple colors of
light on the same fiber.

And the WDM revolution really fueled
everything from about 2000 to about 2010.

And then that started to run out of steam.

And I basically have followed
all of this development.

And in 2010 ish, there was a big
coherent revolution where instead

of switching lasers on and off, it
became phase modulation and that.

Gave a huge improvement in
the amount of capacity you

could put onto optical fibres.

Largely because of the compensation
you could get from electronics.

So that's basically where we are now.

We've gone from 2010 to current
day on this coherent wave a

very high capacity system.

And I've essentially followed the entire
process from 85 and to the current date.

Dan: Great.

Yeah, without fiber optics, I think
frankly speaking, the internet couldn't

have scaled as quickly as it has.

So it's probably been an interesting
journey for you, being right from

the beginning of that, pretty much.

so You're now at BT Manager
of optical research.

What does that mean?

Do you have a team of
people working for you?

What kind of stuff are you
working on at the moment?

Andrew: Yeah, I spent my until about 2010
ish working not in the research side,

but more in designing networks for BT.

I then moved to research and built
a team of ten people to essentially

look, slightly longer term, or look
at the new optical technologies

that are could be relevant to BT.

We have.

So ten people focused on anything
from optical access, so PON, systems

to people's homes, to metro and core
networks, so fibre going around the

country not so much subsea anymore
and then this other kind of associated

activities, so for example, can I
do fibre sensing on optical fibre.

How do I monitor the huge
amount of data that's available?

Can I use AI and machine learning to
improve the performance of my networks?

And, of course, quantum, which around
2013 started to emerge as a potential

use for the, essentially the large
fiber infrastructure that BT has.

Dan: sensing one is an interesting one.

I've

Seen optical fibre can be used to
measure vibration and stress and things.

Andrew: Exactly.

Dan: Yeah, are you using that
somewhere or is that something

you looked at some point?

Andrew: Yeah, I can't talk in too much
detail because some of it is sensitive,

literally, as well as commercially.

But but, yes yeah, and I
think we're not the only ones.

I think, generally, the industry is
looking to find other ways to make, get

revenue from their fiber infrastructure.

It's not just carrying bits, it can also
sense under the ground things that are

happening to critical infrastructure
and we should try to use that.

Fiber is incredibly sensitive and there
are a whole range of systems from,

ranging from DAS acoustic sensing, so
putting acoustic signals over a fiber

and looking for reflections, all the way
to actually using what we have already.

So these coherent systems that I talked
about a minute ago have a side benefit

where the, this phase modulation
you can, look at the detection of

the phase and infer something that's
changing in the fiber as a result.

So long as you tap into it the right way,
you can actually get a lot of information

from what we're building already.

So I think, yeah, I think sensing is, has
got a way to go and could be a big thing

for BT and, and companies that have a
similar large national infrastructure.

Dan: Sensing, I think, is a
little bit like quantum and

it's very delicate, right?

It's using the capabilities of
the fiber in a very precise way.

That leads me nicely on to, to quantum.

So you said you started getting involved
in BT in 2013 there, maybe looking into

what was the arts of the possible what's
your involvement in quantum in BT?

And what are the.

What are the links to the
optical domain for you?

Andrew: Now, BT goes back a long way.

In 1994, we actually, apparently,
had a quantum team that was

filing patents on QKD, quantum
key distribution over PON systems.

Those patents have timed out
since then, but very early.

activities.

But then I it was too early.

And back in 2013, we
reignited our interest.

Partly because there was government
funding at the time initiated

through the quantum comms hub.

And other hubs, but particularly for us,
there was a quantum communications hub.

Government funded, based around
York University as well as other

unis like Cambridge and others.

And we just got involved as a
partner, on the advisory board, etc.

And that was A logical thing to do to
start to build an early understanding

of what these things were capable of.

But at the same time, we also got
involved with Toshiba looking at

how to build proof of concept.

Demonstrations of their early QKD
technology over our installed fiber.

And with that, we introduced them to ADVA,
a company now known as ADTRAN, who were

Making for BT Ethernet systems, so ADVA
would deliver Ethernet for BT's customers.

And my logical question at the time
was, if QKD is an incredibly secure

communications technique, can I combine it
with Ethernet and actually make use of it?

So that was the, the 2013 ask, and
it was our kind of baptism into

the whole area of quantum, I guess.

Dan: is there a a strategy for
quantum that BT has at a higher level?

Because of course QKD is only a small
part of quantum technology as a whole.

just wondering, taking a quick step back,
looking at the high level for a minute.

Andrew: So yeah BT has installed
a quantum program it's managed by

a program manager who oversees the
whole of quantum activities ranging

from commercialization to research.

And I'm responsible for
a fair amount of that.

But there are commercial people involved,
there's planners and installers.

So, So really, it's touching on Dozens
of people across the business, ranging

from, putting some fiber in, installing
some equipment to, to very long term

research on quantum computers quantum
sensing and next generation quantum comms.

But yeah, it does have this overarching
program that oversees it all.

Dan: Very nice, yeah, thank you.

I I did my homework and researched
uh, Lord Andrew a little bit

before before this episode.

And yeah, I've seen obviously you're
very active in the UK quantum scene.

To name a few you're involved
in the IET, UK Quantum, the

IEEE, UK Research Institute.

And there was a project I came
across, AIR QKD, for example.

I guess yeah, it'd be good to hear
the thread of the story through all

of those and how you got involved
and what drives you to, uh, work

with these institutes so closely.

Andrew: Yeah, it's a really good question.

So it's a balance.

I, I have a team of ten, but they're
not all quantum experts, it's probably

two or three focused on quantum, and
we do all these other things as well.

So there's a resource limit to how much
we could do ourselves, and it doesn't

make sense for us to do everything.

And there's I believe in the UK a
very strong ecosystem, especially

down to the university level.

We're very lucky in the UK to have
that resource, so let's make use of it.

And I have, the potential to
have multiple PhD students.

I have five or six at the moment, not
all in quantum, but scattered around

universities like Cambridge, UCL, Bristol,
Glasgow, Southampton, might have missed

one, but essentially these students give
us to spread our wings and to tap into

the fantastic university capabilities.

And then Innovate UK have been a very
important part of the government's

development of quantum as a technology.

Coming out of universities and funding,
startups and getting an ecosystem going,

if you like, from scratch, really.

And I think that's what happened in
2013, that essentially you started to

see Innovate UK funded projects Convince
or enable ideas from universities

to spin out and start to build very
fledgling companies at the time.

And we've taken the attitude
that we would like to work with

them best in class at least.

Because it gives us like
another string to our bow.

It gives us a means to to help support,
to give requirements, and to encourage

what is still a weak UK quantum industry.

But getting stronger and it means that
we can at the grassroots really play

in that space and make sure that the
developments are going in the right

direction, for the UK in terms of
standards, but particularly in terms

of what BT might require in the future.

So it's a balance and we do a lot
of work where we can ourselves.

We file patents and protect the IP
that we're developing, but we're

very um, ecumenically minded.

I work with anybody, essentially
especially in the UK Brexit has

had an issue with our ability to do
quantum in Europe, which is a shame.

It'd be, it would be nice to think
that might change in the coming years.

There's been a bit of a thaw
there, I see, and so maybe we can,

embrace more horizon projects.

In Europe, because they're also very good.

But, it has helped us to very much
focus on the UK as a, the place

to operate in the quantum space.

And I think the last ten years have
gone really well in that regard.

Dan: Yeah, you mentioned Horizon obviously
there was an announcement recently that

Andrew: Yeah.

Dan: Still, we're back into Horizon,

Andrew: Yeah.

Dan: What's that thawing like?

Is it, do you think more of a cultural
thaw or is it just because there

was a gap where we weren't, we're in
limbo for a while and people weren't

really sure whether UK institutes
could get access to funding.

Andrew: iT, there had to
be some kind of agreement.

There are Plenty of associated
countries that are involved in Horizon

that aren't fully fledged EU members.

Yeah Switzerland and Israel, et cetera.

So there was always going to be,
I think, some kind of agreement.

The issue around quantum is that
whether it's included or not.

And to some extent I don't think it is.

So if we wanted to do quantum
collaborative projects in Horizon,

that's, might be harder for us.

But in the more general, for example, more
general optical research space, I think

I'm now freed up to go and do that again,
which is great news, because we have a

lot of really good friends in Europe.

We're still working in European projects.

Yeah, it would be nice to
play a role in that again.

I suspect the quantum might be a bit
more difficult and just to add, I

noticed in the autumn statement from the
government a couple of weeks ago, the

just emphasizing the ambition to work
with countries outside of Europe as well.

So they have missions to, to have
closer collaborations with other

key strategic quantum nations out
there, which might be the US, Japan,

Canada, et cetera, Australia, I think.

I'm not clear whether those countries have
been actually named or identified, but

I think that's also a really good thing.

And maybe, Brexit has helped.

I was thinking a bit more broadly about,
about our strategic relationships.

Dan: Yeah, the announcement of the
missions from the government was

certainly good on the Quantum program.

And yeah, I think it's it
feels like that it's a stronger

sense of identity around Yeah.

This domain, and makes sense to
what you're saying about horizon.

I didn't realize that quantum
was more of a challenge

Andrew: I think for two reasons.

Firstly, it's security.

And doing any kind of security led
or motivated projects is going to

be harder in a collaborative sense.

And so I think the EU feel that's
an area that they need to protect.

And secondly, I think it's It's a
potentially huge growth area and

so I think there's more care there.

I think it would be the same
with AI machine learning.

So I think there's some topics where,
people are a bit more protective.

Dan: Okay yeah, so let's move on
Now to, as we're talking about

the UK, let's talk about London.

So I'm familiar with the London Quantum
Secured Metro Network which is an activity

that's been going on a long time, I
think within BT and with some of the

partners that you already mentioned.

Why don't you walk us through that?

So I think that's, it's quite
an achievement and I'm keen

to deep dive a little bit on
what's been built and and so on.

Andrew: It is an achievement.

I'm really proud of it, actually.

And for reasons that might
not be immediately obvious.

Yeah, I appreciate the chance
to talk about the motivation.

And it really comes from A few years
ago, we built a point to point QKD trial

with a company called NCC in Bristol.

It's point to point, it connects
two buildings in Bristol using

a BT or an OpenReach product
called OSA FilterConnect.

But it's just point to point.

And that's great, but it begs the
question, how would you scale this?

How would you build something
that connects multiple customers?

Would you just connect them
all on a point by point basis?

Not really.

If you had ten customers, that implies
probably fifty separate connections.

And that's not a network.

Very quickly you Come to the
conclusion that what you need to do

is architect a solution where you
have access links from a customer into

an exchange building, a BT exchange.

And then you have to build like
a quantum network that connects

exchanges together, where the the
data coming from multiple customers is

aggregated onto the same fiber link.

And then you.

Have a much better, efficient
way of using your QKD equipment.

What's happening is these QKD boxes
are distributing keys between nodes.

But you only really need one
QKD box at each end of a system.

And the QKD keys that are distributed
can be shared and used for

multiple customers on that link.

So that's a much more cost effective,
scalable way of building a network.

It's just.

Down in Bristol we hadn't done that.

And so we really were challenged,
what would the architecture be

when we move from point to network?

And that's important.

And that was largely untested at the time
in terms of How do you manage the keys?

How do you make sure the
customers get the right keys?

How do you make sure when you start
to aggregate that everything's

still fine and still secure?

And then the second motivation
was around The challenge I

guess from the leaders in BT.

How close is this to being commercial?

You know Can you integrate your quantum
equipment into regular BT operations?

Can you build a communications link that
manages, alarms and that sort of stuff?

Can you just put it into regular
BT exchanges so that it just

looks like any other telecoms
equipment and is fully integrated?

Because that's what you need
to do for this to take the

next stage to commerciality.

So the London trial was
trying to do those two things.

It was build a network and
show that QKD can scale.

And secondly, integrate it
into business as usual BT so

that it's commercially ready.

So that, we can then go to the next stage.

And that's really The trial in a nutshell.

There's obviously a lot more details,
but that they're the two motivations I

guess the third one then is once we've
done that can we get some customers?

On board and actually look at some
of the use cases that they might

want to trial with this And the
thing went live on April 2022.

Since then, we've had both
EY and HSBC making use of it.

And that's been really exciting.

We're looking forward to having
more customers on the trial.

Not hundreds, but we have capacity
to bring more on in a, controlled way

over the next couple of years or so.

We expect it to run for another two years
to give us, all of the answers and the

experience that we need on those various
questions of, how do you run a network?

How do you manage it commercially?

And how do you make use or bring on stream
customers with their specific use cases?

Dan: Yeah, very interesting.

And many questions on
the top tip of my tongue.

I guess first of all You mentioned
sharing the fiber for multiple customers

trying to get a feel for what, which
part of the architecture that is.

So my understanding is it's a ring almost,
or it is a ring, perhaps a triangle from

the middle of London over to Slough.

And then you're offering services
into that core, if you like.

Are you then switching multiple, are you
giving out wavelengths to every customer

and then switching them over the network

Multiplexing that way?

Andrew: Extremely good question.

Essentially, there's an access and
a core, or an access and a metro.

So the access is a dedicated link.

Let's take HSBC.

So they have a dedicated fiber
link from their head office in

Canary Wharf into a BTX exchange.

That is using an Openreach
product called OSA FilterConnect.

Optical Spectrum Access FilterConnect.

You can buy and it's main
intention is to use that product

to deliver Ethernet to customers.

It's used to do 5G, mid haul,
front haul, back haul as well.

So it's a very flexible access
product where the fibre is

divided into multiple wavelengths.

Either 8 or 16 wavelengths
that you can essentially access

via this OpenReach product.

And what you would do typically is
plug some Ethernet into one of those

wavelengths, or multiple of them.

What we've done in this trial is use one
of those wavelengths to carry quantum,

to carry single photon key distribution.

And engineered the OSA filter connect
to make sure that wavelength is

protected and is well separated
from the other much higher power

wavelengths that are on the same fibre.

We've used other wavelengths
to do the reconciliation.

So the discussion between Alice and
Bob at the two ends to establish the

keys that, that, that is separate
from the quantum channel and that's

sitting on other wavelengths.

And then of course, we're
carrying the data as well.

The data that's encrypted the ethernet
data is on another one of those

wavelengths all on this fiber, this
SSA to connect fiber and that is a.

An integrated solution on a fiber
to get you from your your customer

premise into a BTX exchange building.

That's the first leg, and then
you do the same at the other end.

So once you come out of another
BTX exchange, and in this case

it's going to Equinix Cloud
it's another OSA filter connect.

And then the middle bit, the metro
bit, is this ring or triangle that you

described earlier where we have three
nodes two in London and one in Slough.

These nodes are now managing
the central bit differently.

So here we take all of the 10G
wavelengths coming in from different

customers, and we will separate those
off and carry those on a fiber and then

the keys will be handled differently.

So those, that data is encrypted
already and it stays encrypted, but it's

carried separately to, to the far end.

Meanwhile our metro network our
core QKD boxes, what their job is

essentially to carry the access keys.

So what they are.

Basically doing is aggregating all
the keys coming in, from all of these

customers, putting them onto a QKD
distribution whose job is literally

just to carry those access keys.

To another node in our triangle.

So think of it as a, as
an access and a metro.

The access is a dedicated to an individual
customer, and then the metro is there

to, in an aggregated way, carry all of
the keys for multiple customers and get

it to the other end of our triangle.

It's hard to describe without a diagram,
but I hope that makes some kind of sense.

Dan: Yeah, no, I'm cheating by
looking at the diagram right now.

So you've got a QKD session, if you
like, on every single link in the path.

Are you then handing keys over from
one to the other, or doing some kind

of there's a word in cryptography
for exchanging keys at a point and

then extending the security somehow.

I can't remember it.

Andrew: Yeah.

So

Dan: you end up with a, like a
like a daisy chain of connections

Andrew: Exactly.

Dan: Alice and Bob are, yeah.

Andrew: exactly.

So the end, the key that, that is
initiated in HSBC, that key will end

up at the far end in the data center.

So the date, the data that's encrypted
at HS BBC stays encrypted with that key.

Not decrypted anywhere until
it reaches the far end.

So what you have to do is get that
key To the far end so it can decrypt.

So that this key is encrypting
our Ethernet data using AES

256 Symmetric encryption and
like I say, it stays encrypted.

So you've got to get that key across your
metro core network and then off the final

access leg to get to the fire end so
that it can decrypt that, that Ethernet.

And so your middle bit
is carrying that key.

So that, that key is carried on a QKD
key using an exclusive or, for example.

There's multiple ways of doing that.

But essentially, you're using those
Metro QKD keys to carry your access

QKD key along with other QKD keys to,
to deliver it to the fire end so that

it can finally decrypt the Ethernet.

And the keys that you are basically
circulating around the metro are there to

be used in a one time pad, a one for one
basis to encrypt edge keys or access keys.

Dan: that's it thank you, one time pad.

That's what I was trying to remember.

Andrew: Yeah, with a one time
pad it's important that you are

doing the best you can there.

But by having one QKD key
per access QKD key, you are

completely, encrypting that key.

And anybody trying to tap into
that will have absolutely no

idea what that final key will be.

Dan: Indeed.

Yes.

So yeah, it's a neat solution end to get
the optical connections end to end secure.

Andrew: It's neat and

it comes down then to things
like the key management system.

So sitting over the top of all of
this, you need some software, KMS

that, Schedules and orchestrates all
of this and make sure that the keys are

in the right place at the right time.

There's enough keys
and that kind of thing.

So KMS is becoming quite
important to consider.

Both in our Metro trial, looking
forwards, we're going to need them.

Potentially even globally,
if we have satellite QKD.

Connecting our trial to some, another
country, then this has to be some

kind of KMS that sits across the
whole thing orchestrating that.

And I, I think it's going to be a
very interesting area of development.

Dan: Yeah, a question on the
KMS of course, this is where the

complexity of the solution is.

And where I guess a higher focus
of the security posture of that KMS

needs to be taken into account, right?

You don't want the KMS to become
a new attack vector potentially.

So it's a very important system.

I guess I'm highlighting that it needs
to be considered with the similar kind

of level of security as the application.

Andrew: Yeah basic things like you
would, want it to be separate, um,

from the QKD that it's managing.

So there's a logical
separacy between the two.

Also, there's ideas where you might
um, although the KMS is managing the

keys, it doesn't actually see them.

So there's, and I'm not a security
expert, but there's ways of making sure

that separacy is implemented properly.

So there's a limit to what the
KMS can actually physically see.

It might be managing, but it's not
actually seeing things in the clear.

Dan: Got it, yeah.

What kind of challenges did you
have in the implementation of this

network, if you don't mind me asking?

I would say, first of all, you know,
were there any issues around Raman

scattering, or effects like that inside
the optical fiber that prevented the

quantum signaling to work, or anything
else that I haven't thought of?

Andrew: Yeah,

there are.

Let's go through the, at least what
I think is from the most serious.

And the one that hits you
most is just the fibre itself.

It's installed regular fibre
that could be in the, could have

been in the ground for years.

It can be high loss.

It can have reflections.

Reflections are a real problem.

Reflections affect standard systems.

And you have to be careful.

But, quantum systems
are even more sensitive.

And so you have to be careful with the
installation and doing things like OTDR.

Looking at that link before you
put the quantum on to make sure

there's no huge reflections that
are likely to cause problems.

And just the performance of that
optical link becomes more important

and more critical when you're building
a quantum system over the top.

So that's the first thing
a major issue for us.

The second thing is if you are trying to
co exist putting classical and quantum on

the same fibre, and we are, then um, it's
a question of filtering and a question of

making sure that the the classical signals
are completely excluded from the quantum

channel by careful design of filtering.

And I think um, the trial itself is
relatively modest in terms of distances.

So things like Raman that you mentioned
are still an issue, but not to the

same extent that you might find for
systems where you're trying to put lots

of Classical channels, so we're not.

If you wanted a full C band's worth of
100 gigabit channels, carrying multiple

terabits, I think then, over, 30, 40,
50 kilometers, then you're going to have

bigger Raman problems than we're seeing.

But we, we still have to be
careful about Raman and, in some

places, make sure that the QKD
keys are at a different wavelength.

Maybe down at 1.

3 microns, for example.

There's still a balancing act because 1.

3 microns has higher loss than 1.

5.

The classic coexistence conundrum, really.

You really want to put your keys
where the loss is lowest, but

that's where the classical data is.

And this then ultimately
begs the question, should

you do coexistence at all?

And I suspect that In metro
environments like ours it's okay,

because, the distances are just
not so great that it's manageable.

But I think, ultimately, if you're
trying to build national networks and

where that national quantum network
is gonna be Supporting multiple

classical secure systems you would
probably dedicate a fiber to it and

make sure you do it properly and
get it to work as well as it can.

So and that's just my feeling, but
I think we will see less coexistence

because it's just so challenging
in, in those bigger geographies.

Dan: Yeah, if you've answered my
next question, ultimately it's going

to be around the coexistence piece.

It's great to see it functioning
well, but of course it is, like I

said at the beginning of the episode,
a very sensitive, precise process

and can be disturbed pretty easily
like many other quantum technologies.

Yeah, and I guess a
supplier like BT, right?

You've got a lot of fibre, so
there's many options on the table.

You may even have fiber that's
unused and can can productize in

such a way that you use two fibers,
but still make it cost effective.

Andrew: yEah.

Dan: yeah.

Andrew: But, if it's an access link
into a customer, then you really want

to maximize the use of that fiber.

It's likely to be going just a few
kilometers and you can, there are

ways of, engineering a co existent
solution with Ethernet and QKD.

And it's only going a few
kilometers, so that works fine.

And you really don't want to have to
put lots of fibers into that customer,

because that's really expensive.

Dan: Okay, that's great.

So I'm familiar with OSA.

As I said before we met today.

I actually implemented one a long
time ago in a different life.

What are the details in terms
of the performance that you get?

And we've spoken about the configuration,
but in terms of the throughput for, is

throughput the best way to measure the
quantum channel, or is it more around

the success rate of keys generated?

Andrew: Yeah, the number one
characteristic is key rate.

So how many keys are you getting
through your fiber link per second?

And keys will be measured in usually
measured in 200, 256 bit blocks.

Because that's what you
need for an AES key.

Yeah, that, that's,
the best measure of it.

There's an open question
about how many you need.

So in an access link you probably
don't need that many because you're

just using those keys to protect
the single customer channel.

And, you might need a 256 bit key which
you might refresh every second or so.

That's not a very high key rate, so
it's fairly modest, so you need a few

hundred bits per second to do that.

In the metro side it's different
because you then potentially have

dozens of customers all needing
their keys carried, so they're all

giving you, hundreds of bits of key.

Per second.

So then you clearly need
something much bigger there.

And so there, I guess we're talking
more kilobits or tens of kilobits even.

And our trial's generating
those sorts of levels.

Okay it's very dependent on the
distance and the classical channels

and we've got some variability but
we're seeing high tens, hundreds of

kilobits, megabits in some cases.

So plenty of keys.

And yeah, that's so it's very healthy.

There's a lot of headroom For even
considering doing things like one

time pads, so you could even imagine
for a Particular use case you might

actually dedicate all of these keys
to Encrypt something that needs to be

incredibly secret like a even a secret
voice call or some secret financial

trading or something where you could,
if you really wanted to apportion

those keys on a one time pad basis.

But, generally I think we're assuming
that most applications will be

implemented at AES256, where we're
essentially, taking 256 keys and

changing them every few seconds or so.

Dan: Oh, it's that fast, the
exchange rate, if you like, of keys

Andrew: Yeah, I don't know, I don't
think we're really clear how It

whether it needs to be that quick.

There's some rule of thumb that says you
should try to replace your a s key after

about a terabit or a terabyte of data.

Now clearly, uh, 10 gigabits per second.

That's lots of seconds.

A hundred seconds or so.

Or maybe more if it's terabytes of it.

Yeah, so I think we're changing
the keys probably way more

often than we would need to.

But, ultimately customers might want
to upgrade to a 100 gig access link.

That's what we're seeing at the moment.

So there you would need
to refresh more often.

But the technology's capable
of doing that, right?

So as long as we can refresh of order
seconds, I think we're covered in terms

of that yardstick or that rule of thumb.

Dan: Yeah, and the encryption
itself can be performed anywhere

in the customer network, right?

Depending on the kind of infrastructure
they've got on the end, and

that's really their prerogative.

, Andrew: I mean we, we are
using the keys to encrypt a

ethernet to carry customer data.

But you know, These keys could be made
available at the application layer.

They could be made available at the IP
layer to do, you know, IPSEC, MACsec.

So there's different ways of using those
keys from a customer's perspective to

encrypt the net parts of the network
that they really want to encrypt.

The simplest thing for us is
just to encrypt the entire.

stream coming out of a customer's
communications room, and just put

it onto a 10 gig Ethernet trail
but you don't have to do that.

You can use Ethernet encryptors, but
you could feed it into at the IP layer

or even the applications layer and
encrypt email directly or something

like that if you really want to.

Dan: Right?

So the, the encryptors is you're
using doing it at the layer two,

basically, which is a lot simpler and
invisible to the customer, if you like.

Andrew: Yes, exactly.

Dan: Okay.

Thank you.

So let's move on to to
some barriers for the QKD.

It's a very advanced technology.

It's fantastic to see it working and
with real world use cases and providing

assurance for end to end connectivity
across London, and we're seeing it in

other cities and universities, so on.

But what about barriers for QKD?

So I would say cost is
probably quite a high one.

in that you're adding this
additional layer of devices to

the infrastructure to carry keys.

whAt other kind of barriers are there
in the industry or in the market that

you think are worth highlighting here?

I

Andrew: Yeah.

Good question.

Four or five and in no particular order.

Cost, clearly and that's
related to volume.

And also related to the fact that
the solution is hardware based.

So clearly it's gonna cost,
it's not just an algorithm.

And we have a post quantum algorithms
that are being developed that

will be presumably much cheaper.

But QKD requires hardware and at the
moment, because the volume is very low

in terms of sales from QKD vendors.

It's expensive.

I would expect that to drop not
to zero but to drop to The point

where it increments regular
transmission by a percentage.

In other words, you could buy
a non QKD protected link for X,

and you could buy a QKD protected
link for X plus something percent.

That something could be, I don't know.

I'm not going to say, but hopefully
it's not going to be an order

of magnitude more expensive.

But we're not there yet.

I think volume will help
significantly with that.

Other barriers are around the
immaturity of the technology.

So customers just still
not being aware of it.

And we would like more
customers on our trial.

We'd like to expose that
the whole technology better.

And we're clearly in, in a phase
of telling customers about that.

Thirdly, the National Cyber Security
Centre, NCSC have a white paper on their

website that it let's see, negative, but
it issues concerns or warnings about QKD

and things that need to be put in place
for it to be better welcomed, by them.

And those concerns are around things
like assurance the idea of QKD is a very

strong, it has a security proof but you
still need to implement it properly.

And we don't have currently in the UK
is a, an accreditation process that, It

essentially stamps the QKD box to say this
has been built properly, that this has

been built according to a sound design.

And it fulfills the security
proof requirements from the

theoretical perspective.

And that accreditation or assurance
process could be quite involved and needs.

potentially an independent body to
actually look at the box and say and ask

some basic questions and do some tests.

So we don't have that yet and that's,
I think, a barrier, for sure, and and

needs to be, needs to be resolved.

So yeah, immaturity lack
of customer awareness.

Cost and then the, the main regular,
regulator or security over, overseer

in the UK is not warmed up to QKD
yet because of some of these issues.

Dan: think it's probably the same
reason why you've got that in the U.

S.

as well, that there's no
recommendation for use of QKD in

public sector, customers, and yeah,
interesting about the assurance.

I guess there's a technical
assurance and then there's also the

manufacturing and, proof that the the
technology is working as it should.

I think one of the things
behind that is standards, right?

There isn't really any standardized
way of doing QKD other than

the mathematics behind it.

tHere's a number of protocols
which can be implemented.

which achieve different things.

But ultimately you end up with with
knowledge of the shared key on each end.

Any thoughts on the standards
domain and what's being done to

stimulate that going forward?

Andrew: Yeah lots of work.

ETSI have been focused on QKD, have a
special QKD working group for a long time.

ITU can starting to do
work as well in the area.

They say standards are, bodies and
activities are in place and are developing

the kinds of requirements that we need to
standardise the assurance in this space.

It, so that's good.

I think different countries are going
in slightly different directions

which are I guess is inevitable,
and we're not seeing the same.

Restrictions that we've got in the U.

S.

and the U.

K.

and other parts of the world South Korea
and Singapore are happily building QKD

networks to do 5G backhaul et cetera.

So I think, also when you look on the
NCSC website, they talk about principles.

So I think we're gonna see a more
principles based approach whereby what

you try to do is set some principles that
this box has to conform to, and then you

test against those principles being met.

And that will make it a more transparent
process for QKD to to find a way

forwards through, through this problem.

And standards clearly have a major role
in defining what those principles are.

And that makes it a bit easier and a
bit more agnostic, because you're right,

there are multiple QKD protocols And
the main two, one around digital, if

you like single photon BB84 based, but
the other one CV or continuous variable

QKD, which is extremely popular now
and getting a lot of interest among

mainstream vendors, is a very different
beast in terms of how you would check

that it's actually doing what it should.

It's a slightly harder thing to verify
that it's at the right security level

because it's all based on having enough
difference in terms of noise threshold or

the gap between the signal and the noise.

So that's a slightly harder
thing, I think, to work through.

And it's certainly different than for DV.

Based QKD.

So yeah, standards have a big part
to play, and they've been active

for a long time, although they have
been quite academic in, over the

previous years in their outlook.

I think that's changing now, and
we're seeing much more pragmatic

standards addressing this issue.

And BT ourselves are involved
actively in those standards,

trying to make sure that we get to
something that is actually useful.

Dan: Yeah, you mentioned CV and DV.

Of course, there's
entanglement based as well.

And as such those three different
methodologies are extremely different.

anD Other than creating, it would
be bad enough to have a standard for

each of them and then have multiple
standards bodies creating standards

like they tend to do globally.

To create this lattice of
different standards for people

to follow, I think would just not
be very helpful to the industry.

Um, And also with this is just
a reflection on my behalf.

You've got all these multiple layers.

You mentioned the hardware.

There's also the quantum
technique that we mentioned.

How are the, um, qubits being sent across?

Or is it entanglement to to then
measure the entangled photons and

then make conclusions from that.

These are the algorithms that
run across the QKD links.

Some of those have been,
they're very well defined.

But again, I guess as you say, it's the
implementation complexity, which the

standard bodies won't necessarily cover.

Um, yeah, interesting
point about the principles.

Andrew: they won't.

Just to be clear, we're not doing
entanglement in our metro network.

It's all entirely BB84.

We are not at the stage of
implementing entanglement.

I think that still has a way to go.

But yes, you're right.

One of the, it's another barrier
in a sense, is that vendors

are quite protective over what.

They're built, how they're
implementing and for good reason,

otherwise they'll just get copied.

So the actual implementation and design
of a BB84 or a CAO or an entanglement

based protocol or whatever, CV will be
very bespoke, actually, because that's

where the vendor can differentiate.

And actually quite protective.

It is probably not public how a
vendor A has actually implemented

QKD, and they might not want it to be.

And that's an issue, because
there has to be enough.

publicity behind what their implementation
to make sure that it's accredited.

So yeah, very interesting
and difficult problem.

And you don't want to force vendors
to make QKD in the same way because

they need to differentiate because
someone might differentiate on cost.

Someone might differentiate
on performance.

Someone might differentiate on ability
to integrate with classical equipment.

And you want that, And you want to
encourage that, but at the same time

you need to have enough visibility
of what they're doing to make sure

that it's okay what they've built.

Dan: Yeah and I think that comes
in different lenses, right?

You've got the is it, is the technology
creating the keys as designed?

whAt are the attack vectors?

How is the key management supported?

There's all these different things which
can be certified, I'm not sure there's

a single organization doing that yet.

I don't know whether the ETSI
standards are covering the whole

QKD stack, if you like, when it
comes to how it could be accredited.

Perhaps that's one for another episode.

Maybe if you could connect me with
somebody in ETSI who'd be willing

to talk, that would be good.

Andrew: Absolutely.

Yeah.

And I'm sure they would love to talk about
things like common criteria and principles

based approaches, et cetera, what you what
areas, what aspects you can standardize

and assure and accredit that you're right.

It's a bit of a minefield.

I'm not an expert in it, but
yeah, happy to put you in touch.

Dan: Yeah, thanks for
talking to me about it.

Also, you mentioned costs and the
potential of using Cryptography

algorithms, which are running in software.

W where do you see the
future market for QKD?

tHere, there's some, there's the
ability to use quantum safe algorithms

as it stands across links and then
to create keys albeit with some

additional exchanger information between
endpoints, but still being quantum safe.

For example, the McEliece
cryptography system.

Provides keys in that way
which can be used for MACsec

and other types of encryption.

So I guess that this, these techniques
will impact the future market for QKD.

What are your thoughts on that?

Andrew: It will, for sure.

Firstly NIST is as you probably
know currently working through post

quantum algorithm competition, and
is set to complete that next year.

However, some people think,
okay, it's done and dusted.

I heard someone say the other day,
actually, that's just the starting gun.

Once they've produced one or two
algorithms that, that do a quantum

safe version of key distribution.

Then it, that needs to then be
implemented and built and tested.

And that's a multi year project.

And then it needs to be rolled out.

Companies overturning or
renewing, refreshing their key

distribution infrastructure.

That, that's not trivial.

So The Yes you're right.

There is a question around PQC.

But that area is, has
still has a long way to go.

But the kind of question that I
think should be in people's minds

is that's mathematically complex
thing to break, but it's not.

information theoretically secure.

It's just based on a
complicated algorithm.

What is your, as a company now
who's investing in security, what

is your the chances that you're,
or the probability that you're

assigning to that PQC being broken?

And If it's 0%, then that's a very
strong claim, given that these

things get broken quite regularly.

Now I'm not saying that
these will, because they're

clearly the best of the best.

But, there are a lot of people out there
trying to break these codes for a living.

If, if you're assuming there is a chance
that these PQC codes are not completely

safe, then doesn't that mean you should
be looking at other techniques as well?

And I know of some companies
that would like to see multiple

layers of security or protection.

And QKD, for the, for the most
um, secure requirements and data

I believe should be in there.

Now, whether you need it
everywhere, I'm not so sure.

And there might be places
at the edge of the network.

Like my smart meter in my garage
probably doesn't need QKD on it,

but the integrated measurements
of the whole nation's smart meters

including, billing information, etc.

Probably does need protection.

And so it's horses for courses.

And at the very edge of the
network, I guess QKD is overkill.

And a PQC algorithm is fine.

But I think in the center of the
network you probably want to do

the best you possibly can and
make sure that you've taken all

precautions to protect your data.

Dan: Yeah, very well put.

It comes back to risk
management, doesn't it?

Really assessing the risk, assessing
the likelihood, assessing the threats.

And then, defense in depth
you mentioned is a common.

And yeah, I guess QKD is
another tool in the toolbox.

So when it comes to looking
at what that, what that looks

like for your organization.

Andrew: Yeah, sometimes these
things are pitched as either ors,

and I don't think anybody in our
community sees them like that.

So I would fully expect PQC to be sitting
there working alongside QKD anyway.

In any case, PQC will be probably
part of an authentication algorithm

to get any QKD system up and running.

So you need it there in that case.

I can't see why you wouldn't have
it sitting there alongside QKD.

NCSC say that's important anyway, and
I think they're right to say that.

Dan: Brilliant.

Thank you, Andrew.

I've just got a couple more
questions before we wrap up.

One around boosting QKD.

Maybe this is like a roadmap
type thing for you and BT.

In terms of.

We mentioned longer haul fibre
what about things like proxy re

encryption I was reading about?

There's packetization, and you
mentioned satellite photonic sources.

That's really interesting domain there.

Is this areas that you're looking
and, what can you share with us?

Andrew: Yeah, we've done some recent
trials with Juniper actually looking at

how to integrate keys into the IP domain.

There hasn't been much.

Too much work in that area, but I
don't think it's not rocket science

but it's certainly, we are a supplier
of IP networks and so our customers,

many of them are multinationals running
their own IP networks, so that's a

layer where I think QKD encryption
could play a part, but only if it has

a global reach, so that immediately
makes you think of satellites And there,

we're seeing a lot of progress we're
expecting some early missions next year,

but certainly by 2025, 2025, we would
expect to see some satellite missions

starting to fly, and we have a lot of
connections and interest with those.

That for us.

really enhances QKD's capability
and removes the objection that

it's very geographically localized.

And then the interesting question that
is what is The I guess the balance

between terrestrial and satellite.

On the one hand, you might
imagine satellite is just

there to connect countries.

But in, in country, it's
entirely terrestrial.

On the other hand, you might say no,
I'm gonna have multiple satellite

ground stations in each country.

And the satellite's perfectly capable
of delivering some of those long

haul links within a country as well.

Dan: Yeah, thank you.

And finally, future plans for Quantum
in the UK or personally, anything

you'd like to share um, just to pique
the interest of people listening maybe

conferences or activity that you're
going to be involved in going forward.

Andrew: Yeah, we have been very, in, in
BT, very focused on building this trial.

And that's taken a lot of our energies.

We are now, we have that up and running.

It's very mature.

And so we get, we're,
Looking to the future now.

We're excited by the autumn statement
and we see a new phase of quantum.

There will be a new phase, phase three
of the quantum hubs launched next year.

And we plan to be involved in those,
whichever ones are successful.

Good luck to them all and for us that
means getting involved in next generation

quantum quantum entanglement networks.

We are helping in a project called the
Quantum Data Center of the Future with

companies like Orca leading it where we
think, ultimately quantum Comms networks

will connect quantum computers and
produce something that, that people tend

to talk about as the quantum internet.

We don't, I don't really know even what
that means, let alone like the phrase

but for us embracing the kinds of things
that would be needed, like quantum

memories quantum repeaters, quantum
entanglement is where we're heading next.

And um, I'm..

one of the um, subcommittee members
of the quantum track at OFC, the

Optical Fibre Communications Conference
in San Diego next year in March.

So I've just had the the privilege of
helping the committee review all the

submissions, the quantum submissions into
that conference, of which there were many.

And the successful papers
will not be announced until

next week or the week after.

But, so I can't say, I can't talk about
it, but I was really excited by the the

quality and the range of submissions,
which seems to me to have taken off.

It feels like we're entering into a
new era of quantum comms, where all

kinds of Companies and academic groups
are doing research, which wasn't

the case, four or five years ago.

So that conference should
be really interesting.

And I'm actually helping lead a
workshop at the conference, which

is asking a really basic question.

And the question is this what
do, what does the house think?

Do we believe that quantum is all
about QKD, and that's all you need?

Or do we believe that actually QKD is a
bit of a red herring and what we really

want to do is build a quantum internet?

Or finally, do we think that actually
to get to a quantum internet we need to

dip our toe in the water and start to
build very pragmatic QKD networks now?

Because that's the only
way to build experience.

So that's going to be a workshop
in San Diego at OFC, so please come

along and let's have a lively debate.

Sounds good.

Dan: I'll look out for
my ticket in my email.

Thank you.

Yeah, that is a very interesting topic.

QKD is the only active use case really
of the quantum network at the moment.

But of course there are many
others and that's yeah, I'd

love to see the results of that.

Andrew: I was privileged last week
to be one of the external examiners

for a PhD Viva down in Bristol.

Dr.

Marcus Clark, congratulations
to you, if he's listening.

So that was, and I mention it because his
PhD was all about this quantum internet.

And in Bristol they have I think
it's 19 or more nodes connected

together using entanglement.

A hero experiment, a phenomenal
piece of work that Marcus and

the team have done down there.

But it shows you what's possible,
really amazing how much, how

quickly things are progressing.

So although QKD is, where
it's at the moment, that's

going to change quite quickly.

Dan: Hmm.

Yeah, it's an exciting domain, absolutely.

I agree with you there.

Let me wrap it up then.

Thank you very much for
coming to speak to me.

I've learnt a lot and best of luck with
all the endeavours that you mentioned.

And we'll talk again at some
point, thank you, Andrew.

Andrew: Thank you very
much, nice to talk to you.

cheers.

Dan: I'd like to take this moment to
thank you for listening to the podcast.

Quantum networking is such a broad domain
especially considering the breadth of

quantum physics and quantum computing all
as an undercurrent easily to get sucked

into So much is still in the research
realm which can make it really tough for

a curious IT guy to know where to start.

So hit subscribe or follow me on your
podcast platform and I'll do my best

to bring you more prevalent topics
in the world of quantum networking.

Spread the word.

It would really help us out.