Historically, much of the running in developing high-speed networks has been driven by commercial interests. However, in the wake of the 2001 dot com bust, partnerships between private industry and government bodies have proved a more fruitful means of developing new technologies.
That approach has been particularly evident in Australia, where a number of high-speed networking projects are currently under way that involve partnerships between major networking equipment vendors, universities and other government-funded research bodies.
The government is investing funds in developing national networks on an ongoing basis. In October 2002, it officially launched GrangeNet, the GRid And Next GEneration NETwork, a $14 million project to link together universities in Brisbane, Canberra, Melbourne and Sydney on a 10 gigabit network. Industry involvement was a major theme, with Powertel providing the optical fibre connections between the sites and Cisco working as the technology provider. That funding was provided under the government's Advanced Networking Program.
"The network will support a number of Australia's research communities that require significant levels of bandwidth, which include computational physics and engineering, bio-informatics, environmental and climactic modelling, media services and astronomy," Communications Minister Senator Richard Alston remarked at the time.
Just a few months later, in December 2002, funding of $42.5 million was approved for the Australian Research and Education Network (AREN), which will connect together universities across Australia. "The availability of accessible and affordable bandwidth is a major issue as it is central to the operation of modern universities," Alston said. The project has a special focus on regional universities, which have relatively low bandwidth compared to their metropolitan cousins.
One of the furthest reaching projects is CeNTIE (Centre for Networking Technologies for the Information Economy), which has been developed as a collaborative partnership between the CSIRO, Nortel Networks, IP1 Australia, the University of New South Wales, the University of Technology in Sydney and the Western Australian Interactive Virtual Environments Centre.
"We usually talk about CeNTIE as bringing the future of computing to high-end user groups today," says JB Clarke, who heads up Nortel's Wollongong Technology Centre, one of the key partners in CeNTIE. "It's about putting together networks to take advantage of bandwidth."
While CeNTIE made its public debut in April 2003, work on the project began almost two years earlier. CeNTIE also makes use of existing projects such as GrangeNet, and draws its $14 million funding from the $40 million government BITS project.
The CeNTIE consortium began working on its original proposal to the BITS just before September 2000. This was around the time of the Sydney Olympics, which, as CeNTIE director Dr Terry Percival laughingly recalls, was "not the ideal time to be putting something like this together". Bids closed in December, the approval process took place between March and May 2001, and serious work on the project began in October 2001.
Choosing a commercial partner took some time, according to Percival. "We wanted to work with a vendor with an Australian presence in terms of research and development," he says. Nortel's carrier-grade gigabit ethernet system, the Passport 8600, is at the core of the CeNTIE project.
By April this year, the first fruits of CeNTIE were being demonstrated to sponsors and the media. For a government-funded project, especially in the complex area of technology, this represents a remarkably quick turnaround.
CeNTIE's basic vision is fairly simple. "The future of networking is going to be extending the local area network into the metropolitan area network, and then into the wide area network," says Percival. "The common thread is the need for quality tele-presence. Basically we're looking for videoconferencing and audio that works properly. Everyone sees high bandwidth as the goal, but you also need the advanced networking technology to put these things in place."
"We've developed a technology called Digital Video over IP. This runs at 30 megabits per second (compared to conventional videoconferencing systems, which typically operate at 512 kilobits per second), and has CD quality audio. It's cheaper than a good quality system, because you don't need all that processing power." With the network in place, Percival estimates that a $10,000 investment can be used to provide several high-quality video streams, as well as an audio stream. "The fact that it's based on PC technology is why it is cheap," he says.
"The basis of all this telecollaboration is that one stream is not enough. There's face to face, but there's always two other things." For instance, in a medical environment, doctors may wish to view a live scanning process, as well as reviewing existing medical records. This may be technically complex, but as Percival notes, "You've got to make it as simple as possible."
One of the unusual characteristics of such networks is that they don't have typical traffic patterns. "Some communications will last for weeks, some need to be instantaneous but last for just a few seconds," says Nortel's Clarke.
Two basic networks were set up by CeNTIE, one operating in Sydney and connecting around 20 sites, and one in Perth. The choice of Sydney seems obvious enough - it remains the main hub for most of Australia's technology industry activity - but what did Perth offer? One key incentive was reducing the travel time for traditional collaboration. "It takes almost three days for someone based on the east coast to have a one-day meeting in Perth," says Percival.
The network also makes use of a Melbourne-Perth connection provided by IP1 Australia. "The IP1 Australia network can expand seamlessly to many thousands of times its current capacity and so will provide an immensely valuable backbone for Australia's future telecommunications requirements; linking the east and west coasts and providing a conduit for international traffic to both hemispheres," says IP1 Australia CEO Steve Demetriou.
Not all this traffic has to be at high speeds, however. "We also did some trials with 10 gigabit ethernet, which is quite exciting, but we don't need that amount of capacity most of the time," says Percival. "When we're researching a big experiment, we can fill it up, but that doesn't happen all the time. It's not a telco network, it's a research network. It's like a laboratory."
That laboratory has to have a commercial basis, nonetheless. "One of the key successes of the project to date has been the industry focus groups," says Percival. "We wanted to look at what the current problems were in a range of industries, and what the possible solutions were." Key areas for research included telemedicine, the health industry and education.
An early demonstration included connecting together conservatoriums of music in Sydney and Perth for a saxophone quartet master class. This is a particularly unforgiving type of application - any delays will invariably result in sound not being transmitted at the same time, ruining an individual performance. As well, musicians rely heavily on non-verbal cues for interpretation.
A more stressful application is the world of medicine. "In the medical world, one of the areas where we've found a lot of interest is in real-time diagnosis for rural hospitals," says Percival. "Medicine is very specialised these days. If you've got a different or unusual case, you want to direct it to other specialists in the same field to discuss it. There's a lot of people who already have monthly meetings, for instance in the cardiac world."
One recent success has been the development of a haptic system, which allows for 'virtual' training, where surgeons can provide physical training without having to be in the same location as their students. This was first demonstrated in late 2002. "The trainee could feel the instructor guiding his hand around the scene - even though they were on opposite sides of the world," says CSIRO virtual reality researcher Chris Gunn. "They could also co-operatively feel, stretch and manipulate a simulated gall-bladder, liver and bile duct, and guide each other's hands as though they were in the same operating theatre."
A full-scale project in the medical area is due to kick off at the end of 2003. "Doctors have to be comfortable that they can diagnose using this kind of system," says Percival.
The movie industry is also another prime candidate for such technologies. "The film industry is another area we're working with closely," says Percival. A lot of film post-production work is done in Australia. Because of the size of files involved, currently a lot of work is done via sneakernet, and it's all done in batch mode."
Not only does such an approach limit the ability of production houses to outsource work during periods of peak demand - such as the period running up to the final edit of a full-length motion picture - it also restricts their ability to take advantage of Australia's 'follow the sun' capability, where rushes can be produced overnight and then transmitted to the US, which remains the economic centre of the global movie production industry. A high-speed network would instead allow for directors, editors and special effects crews to collaborate simultaneously on editing the same piece of footage, even if all three are in different locations.
"By linking major post-production houses in Sydney we will be researching new work practices, such as collaborative online editing," says CeNTIE Chief Technologist Dr Dean Economou. "This should significantly shorten the post-production time and give Australia's world-class film industry an edge."
Another area of development - education - is driven by even more pressing financial imperatives. "Universities have got cost problems, so why not share their courses to avoid overlap?" says Percival. High-speed networking makes this a practical possibility, enabling individual campuses to develop strong areas of expertise while sharing these with a wider student base.
To date, CeNTIE has run fairly smoothly. "There's been nothing that's really hit us from left field," says Percival - with one notable exception.
When the network was first set up, CeNTIE hoped to make use of the large amounts of dark fibre already in place in Australia. Unfortunately, that didn't happen. "It was more economical to build our own network in parts, which was very disappointing," says Percival. "We've learnt that often it was cheaper to build a new network, even though there was fibre there. Eventually, we got sick of fighting. One of the things we've discovered is that modern telcos are on a very short cost recovery curve."
Had CeNTIE not been underwritten by the CSIRO - which still aims at cost recovery over the life of a project, but takes a decade-long view of how costs should be recovered - then Percival doubts that it would have been feasible. "We took a ten-year return on investment view. We do this all around CSIRO," he says.
CeNTIE itself is currently at the halfway point, and is scheduled to finish at the end of 2004. Percival is confident that, based on results to date, further research funding from other government schemes will be forthcoming. "We are starting to work on the metrics to prove the benefits to date," he says. Future plans for expansion include extending further into other regional areas.
