E-visionaries: Nick Langley talks to Cambridge professor Andy
Hopper, one of the key movers behind the emergence of the city as a
technology cluster and the man behind billion-dollar spin-offs from
the AT&T Lab he manages
Andy Hopper is the managing director - and guiding spirit - of the
AT&T Laboratory in Cambridge. He is also professor of
communications engineering at Cambridge University.
During the past 25 years, Hopper has been an important influence
behind the development of a range of technologies, from local area
networks to popular micro-computers, wireless broadband and
location-based services. He has overseen spin-offs from the
AT&T labs worth billions of pounds.
Founded in 1986, the laboratory currently has a head count of
60, including 87-year old Professor Maurice Wilkes, creator of
EDSAC, the first computer with an internally stored program. Wilkes
was - belatedly, considering his contribution to computing -
knighted this year.
Over the years, Hopper has attracted £120m of investment into
the lab, the university and the spin-off firms. AT&T is putting
up £1.2m to develop the Laboratory of Communications Engineering at
the university'sengineering department.
Hopper believes the lab's academic/industrial model operates to
the advantage of both the university and AT&T, resulting in a
creative and dynamic environment for advanced research. "We have
pioneered and demonstrated the success of research-based
spin-offs," says Hopper. "By transferring intellectual property,
key personnel and support, these small dynamic operations can
accelerate development and get technology to market quickly. But
not all our research ends up in a spin-off. Other technology
developed at the laboratory has been commercialised by our parent
companies, licensed to third parties, or given away free on the
Internet.
"Our research tends to buck current trends, but if we get it
right, the technology becomes mainstream within five to 10 years.
We are always looking for new lines of work that are best exploited
through spin-offs, and with our track record, getting investment is
no longer a problem."
The lab took a lead in getting affordable broadband technology
into the mass market. They also began working on pervasive
computing 10 years ago, long before anyone but a few academics in
the US would have recognised the term.
The fabric of the building is riddled with sensors, microphones
and cameras. Hopper calls it sentient computing. On a small scale,
it presages a future where computers are able to create a detailed
model that mirrors the real world, and an environment where "the
interface is everywhere, everything can be controlled and nothing
ever gets lost".
Sentient computing uses "Active Bat" ultrasound location
technology developed at the lab. Everyone carries an electronic
tag, and sensors in the ceiling and the devices they use track
their position with an accuracy of three centimetres. "Each person
and object has a defined space with associated characteristics,
rather like a cursor moving across a screen," Hopper explains.
"When changes occur, spaces converge or users press virtual buttons
with their Bats, events are triggered automatically. This means
that people can control applications in a natural way, with no
conscious interaction with the environment."
Examples of sentient computing include personal desktops that
follow you from one screen to another as you move around the
building, or telephones that automatically program themselves with
personal short codes as you approach. The technology is being used
experimentally at Addenbrookes Hospital, where consultants can
carry on videoconferences while doing their ward rounds: the
conference shifts to the nearest monitor.
"The location system is very good at telling us where objects
are, but when it comes to application design we are more interested
in how objects stand relative to one another," Hopper says. "For
example, a mobile desktop application needs to know when a person
is standing next to a computer, but it doesn't care about the
absolute location of the person or the computer. Furthermore, the
concept of 'standing next to a computer' is a vague one and needs
to be translated into something more formal.
"The approach we have taken to this problem is to treat each
floor of the building as though it were a big graphical user
interface, and to allow application programmers to register spaces
on the floor, and to receive call-backs when the spaces interact in
some way. A mobile desktop application registers a space extending
for about four feet around the screen of each display, and a small
elliptical space around each person. As people and displays move
around, so the corresponding spaces move too. When a person space
becomes contained by a display space, the application can bring up
the person's desktop to the appropriate screen. When a person space
stops overlapping a display space, it can move the person's desktop
away from the screen."
Hopper sometimes calls the approach "programming with spaces",
but the analyst community has dubbed it "location-based computing"
since Wap came in.
"We've been doing location-based services for a decade, and we
know more about it than anybody in the world," Hopper snaps,
justifiably impatient that the topic is being treated like
something brand new. "Location-based services are pretty hard. The
systems need to give you information about the real world, but that
must be put into the context of human interpretations of proximity.
You are not next to the person in the room next door, though all
systems now being postulated would suggest that you are. Our system
doesn't because it's ultrasonic, and doesn't go through walls."
One set of walls Hopper has always been able to penetrate are
those between academia and industry, and he was doing it long
before the Thatcher Government began making the universities work
more closely with business.
Born in Warsaw in 1953, Hopper came to England at the age of 10.
He read Electrical Engineering and Computer Science at University
College Swansea, graduating in 1974. "It was a fabulous course,
because it combined computer science, the digital side of
electrical engineering, and a little bit of business studies."
The same year, he began his PhD at Cambridge, working on local
area networks, and in the process becoming co-designer of the
Cambridge Ring Lan technology. "Then I did the VLSI [Very Large
Scale Integration - a technique for cramming as many components as
possible onto a very small piece of base material] version as a
post-doc. My thesis was called Local Area Computer Communication
Networks. It should have been called how to design a fast network
and put it on a chip." Next, he worked on computer-aided design
(Cad) techniques for VLSI chips.
In 1977, he met Herman Hauser at a party. Then an academic,
Hauser was to evolve into Cambridge's first "millionaire don", and
is still involved in getting fledgling IT companies off the ground,
as founding partner of the £200m Amadeus Capital Partners
development fund.
"Herman said, 'why don't we start selling some of this stuff?'
So we started a company which sold those network boards."
The company, Orbis, was soon absorbed into what became Acorn,
manufacturers of that pre-PC bestseller, the BBC Micro. Hopper
became research director at Acorn. "We did the Cad for the chips
which went into the BBC Micro, and also provided network chips. The
BBC machine had not only better graphics because of its better
chips, but a proper network approach."
Acorn also used Hopper's Cad to develop ARM CPU chips. The
low-cost 16/32-bit Risc processors are used in set-top boxes,
mobile phones and wireless multimedia devices. ARM, the company
spun off from Acorn in the late 1980s, shipped 12 million of its
chips in the quarter to September 2000, to customers including
Ericsson, Nokia, Nortel and Panasonic.
As the PC pushed alternative micros out of the market, Acorn got
into difficulties, and was rescued by Olivetti, the Italian
typewriter manufacturer which had transformed itself into a
supplier of mainframes and PCs. Making the most of its assets, in
1986, Olivetti founded what was then called the Olivetti Research
Laboratory, with Hopper as managing director. Ten years later,
Oracle joined the lab as co-funder, with equal access to the
research. Then, in 1999, the lab was bought by AT&T.
Have those changes of ownership influenced the lab's research in
any way? Hopper says not. "The companies funded us for what we are;
they haven't tried to mould us. With AT&T it's been great -
there's a fantastic intellectual context. As part of the old Bell
Labs, they are pretty good brand and peer group to be working
with."
He describes what the lab does as "optimising between keeping
those who feed you with money happy, doing what you're good at, and
having fun. We've got this tremendous record of making money for
everybody, and so far we've been able to justify the financial
side."
Are there no pressures to move into shorter-term deliverables?
"We always have a balance of shorter-, medium- and long-term
projects. Once you've got something which is of some relevance to
somebody, then you can start commercialising. That might mean you
sell it, might mean you give it away - how you commercialise itself
keeps on changing."
Examples of free products downloadable from the AT&T Web
site (www.uk.research. att.com) include Virtual Network Computing,
a system providing remote and cross-platform access to Unix and PC
desktops, and OmniORB, a lightweight Corba 2-compliant object
request broker.
Hopper is currently planning an experimental "platform", which
will be used for trying out new applications of third generation
mobile handsets, using student man- and woman-power. "There will be
a number of industrial participants, and I suspect AT&T will be
one of them. We will get the companies to donate handsets, and get
the students, who are an accommodating bunch, writing and running
applications.
"The university is an wonderful middle ground for doing things
with companies, with an underlying academic component."
Hopper believes that within the next five to 10 years, we'll see
a fibre connection to every home and business, with wireless
connections everywhere else. "Internet connectivity to the home
will become one of those items estate agents put on the sales
particulars of a house, alongside gas and electricity and mains
sewerage."
In the meantime, technologies like Digital Subscriber Line (DSL)
will plug the gap, providing high-speed services over the
copper-wire "local loop", which currently carries telephony
services to most homes. Fixed wireless and DSL will also provide
the links where distance makes a physical fibre infrastructure
impractical.
Broadband will enable communications with more "presence" in
them: at the very least, adding video links, although Hopper's work
on sentient computing will enable us to introduce more of our
preferences and personalities into the way we communicate. It will
be up to you whether you want to type, talk, shout, wave your arms
about, whistle or sing - the interface will adapt to you.
"We will harness information from the world around us, and our
personal preferences, to enhance the way we interact with each
other, the environment, and the increasing number of computing and
mobile devices in our homes and offices. We will in effect create
an interface that is everywhere."
Hopper thinks that to date, electronic communications have made
it too easy for people to be anonymous. Broadband will help turn
the world into the kind of village community which regulates itself
because everybody knows what everybody else is up to. "I subscribe
to the basic premise that it's for the good of the world to make it
easier and cheaper for people to communicate. We're not even close
to saturating the potential desire. When you have unlimited
capacity for next to nothing from everywhere, then I'll be
content."
Ultimately, broadband will make it possible for us to have an
immensely high speed communications connection wherever we go,
without having the cost constantly at the back of our minds.
Hopper is a life member of the National Trust for Scotland,
enabling him to visit crumbling highland castles without charge
whenever he wants to, and he says he'd like a similar arrangement
with a comms provider. "I want to write a cheque, and from then on
and forever, all the electronic communications I need will be
available to me from this supplier."
A limited number of Cambridge residents already have a taste of
the wireless broadband future. A network set up by the AT&T
Labs, working with the Laboratory of Communications Engineering and
Adaptive Broadband, is providing interactive high quality video and
audio services over a 5km radius. Student hostels are among those
benefiting.
The Adaptive Broadband technology delivers 25 megabits per
second (mbps) asynchronous transfer mode or 10mbps Ethernet across
wireless connections, still well behind state-of-the-art hard-wired
networking speeds, but miles ahead of the slow trickles of data
which users of PSTN and ISDN connections have to put up with. ADSL,
which is slowly coming onstream in the UK, promises from 1.5mbps to
9mbps to the user - but a maximum of 640 kilobits per second (kbps)
in the other direction.
The contrast with mobile wireless is even more stark. Wireless
application protocol (Wap) phone users are waiting impatiently for
General Packet Radio Service which will provide data at up to
115kbps - a fourfold increase on GSM technology. Universal Mobile
Telecommunications System, expected to begin experimentally next
year, will offer up to 2mbps.
Ironically, Hopper is not among the blessed. "It's a
line-of-sight technology, and because of a big tree, my house is
not in line-of-sight. There are colleagues in this laboratory who
have 25mbps to their homes, and I don't."
Hopper says that when high-bandwidth wireless connectivity
becomes the norm, the volume of digital media generated by an
individual will no longer be governed by the amount of non-volatile
storage he or she can physically carry, but will be essentially
limitless. "It may not be too long before all of the media assets
you have, like photos, home movies, music recordings, TV
programmes, and automatically generated photo-logs of your car
journeys, will be held in a little terabyte box in the corner of
your living room, or in one part of a big petabyte box at the data
centre of the XYZ Digital Media Corp."
The lab developed the AT&T broadband phone as a way of
accessing all that data. Beside the familiar handset, it has a
touch-screen interface which gives access to broadband multimedia
services. It's a thin-client device, with no application software
to crash, no information to lose, no processor or storage to
upgrade, no risk of viruses, and zero user maintenance. "This is
similar in concept to an existing phone," Hopper points out. "If it
is replaced with another one, the user doesn't lose their phone
number or access to any of their services. Compare that with the
applications and data on a home PC."
The broadband phone uses IP telephony, which means that voice is
carried over the same network as data and other digitised material.
Users can draw diagrams and maps for each other, share photographs,
or browse the same Web location while having a conversation, as
well as sending faxes and e-mails, watching live or stored video,
and ordering goods online.
This is the technology that Internet-enabled call centres have
been waiting for. Currently customers who want to talk to an agent
while navigating a Web site need two different phone lines. Because
the technology is entirely network-centric, Hopper says, it's
infinitely scalable, but also viable for a small enterprise.
Of course, the broadband phone isn't a phone at all, but a
computer, which could just as easily take the form of a wireless
tablet, a panel mounted on a wall or even a car dashboard. "Because
the system is designed to interact with the endpoint using very
simple protocols, it can easily be re-targeted at different
devices". Hopper says. "When working at home, your phone could take
on all the facilities normally available at your office. Your
personal phone display could even be made to migrate to your PC
screen, to a call box, or to your car dashboard."
Hopper says the phone is just the first of a line of remotely
managed, simple-to-the-point-of-no-brainer devices that will
provide access to broadband multimedia services over the
Internet
The broadband phone has helped with a cultural problem at the
AT&T lab. Readers may have felt uneasy at the description of
the building, with its Big Brother-like array of sensors and
bugging devices. Some laboratory staff felt that way too.
"When we deployed this multi-media system of computers with with
microphones and cameras attached, we were concerned to make sure
that microphones could be switched off, and users could see that
they had been. It's actually impossible to listen in to another
office. Our users were a bit queasy until that happened.
"We recently put in the broadband phone system, which is also
computers with microphones attached, but because they look like
phones, nobody said a dicky-bird. There's no difference between the
two, except trust."