Bandwidth is like computer memory; you live with what you’ve got, but more is always better. And while the advent of so-called broadband has satisfied consumers’ demands for a while they will soon be clamouring for more. Fortunately there is plenty to come; the reality is we are still in the Stone Age when it comes to ubiquitous bandwidth.
The benchmark for shifting data quickly is the fibre optic backbone. Scientists and engineers agree that light is the best way to move data over intercontinental distances, and recent developments have seen the adoption of 10 Gbit/s technology. To the public, currently tapping into the Internet on a 512 kbit/s broadband connection, 10 Gbit/s sounds way beyond anything they’ll ever need: so is there really a need for more?
“That’s always been the case though in the history of communications,” says Prof. Ben Eggleton of The University of Sydney. “People argued 25 years ago at the advent of the optical communications that there would be no need for more bandwidth
“But the benefits will be so tremendous that we can’t even anticipate, both in terms of what we can do, and how prices will plummet.”
Prof. Eggleton has a high profile in the world of optical physics in part due to nearly a decade of pioneering work at Bell Labs, the research arm of Lucent Technologies, in the US. He has co-authored over 100 journal publications and is now back in Australia as professor of physics at The University of Sydney, an ARC Federation fellow and the director of the Australian Research Council Centre of Excellence for Ultrahigh-bandwidth Devices for Optical Systems (CUDOS).
The Centre of Excellence is funded under the Australian Government “Backing Australia’s Ability” initiative. These programs are aimed at acheiving scale and focus in areas of strategic significance.
CUDOSbrings together Australia’s leading experts from The University of Sydney, Australia National University, Macquarie University, Swinburne University and University of Technology, Sydney.
So when Prof. Eggleton talks about optical communications people listen.
His vision is for fibre-optic communications to push through 40 to 160 Gbit/s and beyond… far beyond. His job is to develop the science and technology to make this happen, using the resources of CUDOS. Success will make Australia number one in optical communications at a time when the world, and China in particular, will be crying out for more bandwidth.
However, this isn’t simply a case of evolving current technology. The optical systems employed today are restrained by the need to employ hybrid optical/electronic routers, and at 160 Gbit/s these devices run out of steam. Success for CUDOS’ initial five-year project relies on a fundamentally new approach; generating new theoretical physics in the discipline of non-linear optics. And, notably for an academic program, CUDOS eventually plans to go well beyond the “blue sky” research to produce integrated optical chips using processes that can be adopted commercially, although the focus for the first five years is on basic research.
An electronic bottleneck
“I was involved at Lucent with the migration from 2.5 to 10 [Gbit/s], we’ve now seen 40 Gbit/s being deployed, the next step is 160 Gbit/s, but that’s where it gets very difficult for electronics to cope,” explains Prof Eggleton. “We can’t continue to grow without running into fundamental challenges associated with electronic solutions.”
At each node of the contemporary optical infrastructure lies a router. Larger than a PC, and many time more expensive, these units direct packets of data around the network. They have provided a reliable method for doing this at today’s network speeds, and for the next 40 Gbit/s generation (although they will become even more expensive at these bit rates). However, routers have a fundamental speed limitation which makes them untenable at 160 Gbit/s. This is because the router has to convert an optical input to an electronic signal to reshape, synchronise and amplify the data before converting back to an optical signal and redirecting. The switching speed of a router is limited to that of the transistors it employs, and, as any electronics engineer will tell you, transistors are reaching technical limits bounded by the mobility of electrons in silicon and leakage currents from ever smaller devices.
The solution is to dispense with the electronics and perform the tasks of an electronic router using a photonic chip. CUDOS’ idea is to fabricate photonic chips, based on new non-linear materials, for example chalcogenides, to which thegroup has a unique approach. Then, by using conventional microlithography employed in silicon chip manufacture, the group plans to make thumbnail-sized, inexpensive devices to do the signal manipulation currently done by the cumbersome router.
This sounds simple in principle, but presents a huge challenge beyond current technology. CUDOS’ scientists plan to change all that.
The key function of any photonic chip would be optical switching - analogous to the electronic switching of a transistor, but faster... much faster.
“The physics underlying the optical switching is governed by the electron shell of an atom and is consequently very fast,” explains Prof. Eggleton. “Its response is at femtosecond [one millionth of a nanosecond or 10-15 of a second] speeds which is thousands of times faster than a transistor.
“Light-by-light switching can be extended by a factor of 10,000 from where we are now. Even if we can reach the goals we have set at CUDOS, there is still a factor of 100 to explore.”
An optical solution
CUDOS’ five year mission to replace routers with a photonic chips requires detailed research into non-linear optics and the miniaturisation of optical devices based on these effects.
“The challenge is formidable,” notes Prof. Eggleton. “Ultimately we are pursuing a device that would remove the need to convert light to electronic signals and back, be the size of a thumbnail, cost a fraction of a router and consume very little power.”
According to Prof Eggleton, the chip would include much of the functionality seen in existing communications systems, for example de-multiplexing, management of distortion and performance monitoring.
The CUDOS team have identified three pillars to the project in order to develop a photonic chip; basic physics, materials and microlithography.
“We have a very strong theoretical physics program,” says Prof. Eggleton. “Around a third of the team are theoretical physicists, and solving this problem is going to take some fundamental new theory.
“It’s described by Maxwell’s equations; these are well known, but solving them on this scale for complicated linear and non-linear optics is state-of-the-art.”
The second pillar is the materials investigation. According to Prof. Eggleton, the best example of this work is the program looking at chalcoginides. These are compounds which contain sulphur, selenium or tellurium. They are solid state materials which are well-suited to non-linear optics applications. Much of this work revolves around finding the best materials and crystal structures in terms of “optical bandgap” and periodicity to enable the chip to shape and sychronise light packets arriving at different times.
And finally there is a microlithography program which is looking at how to integrate the functionality onto a single chip where the cost savings can be realised. If the functions can be mass produced by photolithography the device will be cheap and commercially viable.
“We use photolithography to create the waveguides, in those waveguides we create [Bragg] gratings, and other elements that allow us to integrate the functionality needed for a fully working photonic chip,” explains Prof. Eggleton.
Proof of concept
Although CUDOS’ ultimate goal is to produce an integrated photonic chip, it faces demands from its paymasters to producs some proofs of concept. Later this year, in what promises to be a genuine “world first”, the team plans to demonstrate a photonic chip optical “regenerator”. This is an important function that would be a fundamental function of an integrated device.
A regenerator “cleans up” an optical signal that has travelled over long-haul fibre optics. It does this by employing a non-linear optical transfer function that suppresses noise and fluctuations in the signal amplitude so that the original data packet can be regenerated. The demonstration will illustrate the speed potential of an optical solution because this early example will run at terahertz switching speeds, which far exceeds current electronic solutions, and represents a world first for an optical device. The device will employ chalcogenide waveguides fabricated using microlithography techniques, neatly combining the three pillars of the CUDOS program.
CUDOS plans to test the devices using a ParBERT tester from Agilent Technologies . This unit will allow the integrity of the regenerators to be checked at 40 Gbit/s to demonstrate that they can suppress noise, regenerate, restore and reshape signals; all of which would have to be done in a commercial device. (See sidebar “Testing optical chips”.)
“Soon we hope to use [the regenerator] in a 160 Gbit/s system to regenerate data traffic that is propagated over long distances to enable truly optical transmission,” enthuses Prof. Eggleton.
“We want to make sure 160 Gbit/s can work and then move forward - we’ve got another factor of a thousand to go.”
The other major demonstration project is a “slow light” device. In the context of the integrated photonic chip, this device will need to synchronise packets of data arriving at different times before redirection. To do this the device will have to incorporate delay lines that act as optical buffers.
“This is currently being done electronically,” notes Prof. Eggleton, “but it’s going to crash a 40 Gbit/s data rates.
“Notably, DARPA [The Pentagon’s research arm] is funding several US groups to look at slow light applications. They are using atomic resonance systems which we believe are flawed; we think photonic resonance is better and this is what we hope to demonstrate. Our device can [slow light sufficiently to] almost store bits of information.”
Australia at the forefront
CUDOS’ efforts to develop a photonic chip parallel those of the development of the electronics IC back in the late 1950s. Only this time Australia is at the forefront rather than the US. Today, even an average silicon processor is incredibly powerful, yet inexpensive. The key to advances in microelectronics has been the use of scaleable manufacturing processes for integrated electronic circuits that could be improved cost-effectively. These are the experiences Prof. Eggleton’s team is applying to photonic chip fabrication.
“We are applying these lessons to optical communications to finally make it the ‘enabling technology’ for telecommunications,” he says. “The success will depend upon miniaturisation and integration of photonic components into a single chip that could replace expensive, cumbersome routers.
“With the next generation of optical communications systems built around these photonic chips, we can open the gate to a path of development that will lead to a level of personal, business and regional interconnectivity unimaginable even by today’s standards.”
Although CUDOS’ brief is not to commercialise its technology, the hope is that other companies will license Australian-owned technology to make the photonic chips. The potential is huge.
As Prof Eggleton puts it: “What we are doing here is to make sure that Australia is at the leading edge of photonic science and optical communications in a decade when the demand for bandwidth is continuing to grow.
“This ‘market for bandwidth’ will be in Asia, and China in particular.”
Beyond the export opportunities, inexpensive and widely-available bandwidth promises much for Australia’s own prosperity. It will mean Australian regional communities enjoying equal access to a range of services taken for granted in urban areas, such as online specialised medical consultations for outback communities.
CUDOS is an ambitious project, but with the talent at its disposal the chances of success are better than even.
“We are a unique centre in that I can’t think of another in the world that brings together the range of expertise and the scale we have achieved,” emphasises Prof. Eggleton. “There is not another group in the world that has the mix of skills and theoretical physics expertise of this one.”