Innovative optics targets next-generation telecoms

30 Jan 2004

Despite the telecoms downturn there is still plenty of innovation emerging from the R&D labs. Steve Ferguson of Marconi Communications examines 10 optical technologies.

From Opto & Laser Europe February 2004

There has been plenty of bad news in the telecoms industry over the last two years, but if you take a long-term view then telecoms has been a resounding success. Revenues of telecoms operators have risen enormously over the last 150 years, the amount of traffic per bearer has increased steadily and analysts forecast that the market for "next-generation optical products" is healthy.

Equipment vendors and network operators that want to take advantage of the long-term potential of the industry must now watch the innovation going on in research labs in order to see which optical technologies will be key for future growth.

In the near future, industry emphasis will continue to be on consolidation and cost reduction and many products will be based on existing components and product types. However, looking beyond 2005, there are a variety of emerging technologies that could well prove invaluable to this industry.

Enhance and advance Much of the research currently under way builds on techniques and ideas that already exist and combines them to create dramatically better systems. Wavelength switching and broadband passive optical networks are key examples.

Wavelength switching can be used to reroute traffic, independently of payload protocol. Remotely reconfigurable optical add-drop multiplexers (R-OADMs) provide the first stage of wavelength switching, and Marconi has already deployed more than 300 R-OADMs in public networks, each capable of supporting up to 32 x 10 Gbit/s.

Before full photonic cross-connects, which promise massive savings in cost, space and power consumption, can be deployed, traffic needs to grow. Current 32-, 40- and 80-channel systems still have light traffic loads.

With greater loads, many operators and vendors view R-OADMs as a significant feature of future networks. Wavelength switching will offer the greatest value when combined in nodes with electronic switching, under a single level of automated management control.

Broadband passive optical networks (B-PONs) are an attractive way to deliver fibre-to-the-home (FTTH). Although the major investments in B-PONs in the mid-1990s were too early to impact the market, investors have recently begun to show increased interest again.

Japan leads the way in FTTH deployment and authorities predict more than 7 million users there by 2006. This is in part because of Japan's high density of housing, but also owing to a national policy for broadband. Some European countries, such as Italy and Germany, are also deploying FTTH in areas such as greenfield housing.

One application that could also drive FTTH deployments is the provision of video services, which several operators have earmarked as their best bet for good future margins. While asymmetric digital subscriber line systems combined with MPEG4 coding could deliver video services, there are many constraints. Delivery across fibre removes these.

Electronic integration The use of electronics to improve the performance of photonic systems is another major research area, and the technology is now moving beyond 10 Gbit/s. One example application is enabling multimode fibre (which offers low installation costs but is often thought to be limited in terms of speed or distance) to achieve 10 Gbit/s in B-PON or access network applications that require a range of around 1 km. Techniques that could help this include forward error correction and fibre compensation.

Forward error correction (FEC) improves photonic-performance margins by digitally correcting errors after they occur but before the data are used. The fast response of electronics, compared with today's typical photonic devices, is an advantage when compensating for polarization-mode dispersion. FEC uses powerful processing electronics for 10 Gbit/s systems and is being developed further for use at 40 Gbit/s. In addition, true digital photonics techniques could give 160 Gbit/s.

Fibre compensation is a precursor to the true integration of electronic and photonic technologies on the same chip. This technique compensates for fibre impairments at the photonic level to prevent digital errors before they occur. The technique is similar to that of the phone-line modem in a PC, but 200,000 times faster.

Many fibre routes need conditioning for use beyond 2.5 Gbit/s owing to the effects of dispersion. The effects of current photonic-compensation techniques are marginal for 40 Gbit/s and introduce a 20-30% loss. Electronic compensation is not as effective as true photonic compensation, but its costs are much lower.

Digital boost Although still at the research stage, digital photonics techniques such as multiwavelength regenerators, multiple integrated photonic digital devices and optical packet switching could greatly improve system capacity, cost and functionality.

Multiwavelength regenerators promise to remove the planning burden of analogue characterization for long-distance routes. Two approaches are being studied at present. The first employs multiple regenerators in a chip, which is difficult to do and is really just putting more electronics in a package. A more innovative approach is the regeneration of multiple (about 10) wave-division multiplexing (WDM) channels within one optical-processing device, typically a variant of a semiconductor optical amplifier. Such devices could be deployed in around three to five years.

Multiple integrated photonic digital devices (devices integrated onto one substrate and suitable for mass production) could be used to create devices for FEC, demultiplexing and more intriguing applications such as encryption. Photonics researchers want to be ready for the day when substrates other than silicon will be used.

As is the case with many photonics applications that burst onto the market, distributed switching and control have been the focus of much research in universities. However, improvements in all-optical switching and optical memories are still needed, and optical packet switching will probably first appear as optically assisted packet switching in single-router nodes.

Holey structures Photonic-crystal devices and fibres rely on photonic bandgaps and promise impressive control over the properties of optical devices.

Just as electronic bandgaps are the basis of transistors and integrated circuits, photonic bandgaps could form the basis of many types of active and passive optical devices. A photonic crystal is a lattice in a dielectric material created, for example, by an array of holes in an optical waveguide. These holey structures can be tailored to create photonic bandgaps - a range of frequencies in which electromagnetic waves cannot propagate.

On a substrate such as silicon, a pattern of microscopic holes or similar discontinuities shapes the light field, enabling overall device sizes just a fraction of those today. For example, 3 dB fibre couplers now on the market are tens of millimetres long. In a photonic-crystal device, the equivalent functionality could fit into just 20 µm.

Most work at present is on 2D devices - structures patterned in two directions and constant in the third. Fabricated via photolithography, plasma etching and metallization, such structures can produce nearly lossless filters, waveguides and mirrors, lasers with a low current threshold and holey optical fibres (see below).

Three-dimensional devices are just starting to emerge. These are much harder to create, but will be needed to make active photonic-bandgap-based devices. Scientists in Japan plan to build commercial 3D optical crystals in space, relying on the zero gravity to prevent distortions in the lattice.

Photonic crystals offer the possibility of single-substrate integration of photonics and electronics. When combined with digital photonics, this could enable the multiple parallel processing of ultrahigh-speed signals.

Photonic-crystal fibre is the most commercialized photonic-crystal technology, and comprises a long thread of silica glass with a periodic array of holes (containing vacuum, air or liquid) running along its length. It can give extreme properties, such as selective forbidden regions of wavelengths or high nonlinearity. Other fibre designs can offer greatly reduced signal degradation for transport applications, nonlinearity 100 times lower than current fibres and almost perfect control of chromatic dispersion.

In the longer term, the attenuation could potentially be much lower than on legacy fibre. The industry is not there yet, however - current loss from photonic-crystal fibre is 13 dB/km compared with legacy-fibre loss of 0.2 dB/km - and there are big questions regarding what to do about splices.

Data storage The final category of emerging technologies is a wild card - optical signal storage. Researchers in this field have come up with many ideas but, so far, few useful techniques. For example, there is still no optical random-access memory. Such a device could revolutionize router design because optical memory is potentially much faster, although it would not necessarily be smaller or denser than electronics.

There is considerable ongoing research in this area. Light has already been slowed by 2000 times in a solid-state device and the latest results show that it can be slowed at room temperature. This could prove a crucial breakthrough. After all, laser development only began seriously once the first room-temperature lasers appeared.

Commercial potential All of the above technologies could prove highly useful in tomorrow's optical networks. Photonics is inherently as powerful as electronics, and the potential of the emerging combination of the two is awesome. There are plenty of areas for innovation, as is evident at the key industry conferences and meetings. Positioning these ideas for commercial success is a real challenge but, crucially, the long-term market in telecoms is still healthy.