Dense wavelength division multiplexing

The fundamental elements of a fibre optic system traditionally have been the source (laser or light-emitting diode), the optical fibre over which the signal is transmitted, and the receiver, which decodes the pulse back into an electronic domain.

In the past, if a carrier needed to increase the number of messages sent along an optical fibre, the response would be to increase the number of pulses flowing through the fibre by developing a faster source and receiver, a process known as time division multiplexing (TDM). As synchronous communications replaced asynchronous transport along a fibre optic cable, faster TDM has translated into higher rates of the synchronous digital hierarchy (SDH) or of synchronous optical network (SONET). SDH rates grew from STM-4 to STM-16 optical networks (SONET expanded from OC-12 to OC-48). Some vendors have even begun delivering SDH-64 (OC-192) systems, which can deliver up to 10 Gbit/s.

Because of the rapid speeds of these lasers, concerns have been raised about using them on older fibres, particularly those subjected to polarisation mode dispersion (PMD). Largely ignored a decade ago, PMD has become a major issue in 10 Gbit/s system designs. In some older fibre plants, PMD can be the dominant issue and can prohibit long-distance OC-192 transmission altogether.

Proposals to use ever higher data rates per channel now dictate that the system designer must examine each fibre span in detail, engineer each span differently and customise the operation of the equipment on a span-by-span basis.With the advent of 10 Gbit/s TDM systems, it is no longer possible to view the installed fibre plant as being data-rate insensitive. Dispersion is 16 times more important an effect at SDH-64 rates than it was at SDH-16 rates. Without some form of dispersion compensation, SDH-64/SONET OC-192 systems are generally limited in distance to 50-75 kilometres. The only alternative to TDM until recently was the installing of new optical fibre cable, which generally was, and still is, a more expensive option.

DWDM: a new way to expand bandwidth

The Internet explosion has caught nearly everyone by surprise, including many of the carriers responsible for delivering its traffic. The result has been that bandwidth capacity requirements have grown faster than the ability of TDM to satisfy them.

This has led vendors to take advantage of a characteristic of optical fibre that had been ignored before. Because light, in the form of photons, travels through an optical fibre in a manner that does not demand space, engineers are able to send multiple streams of information on distinct wavelengths, each of which can travel through the fibre simultaneously. This is the process known as wavelength division multiplexing, or WDM. This has now been enhanced so that it is possible to transmit multiple laser colours with less than a nanometre (nm) of separation between each wavelength, a technique known as 'dense wavelength division multiplexing' (DWDM).

The most common forms of WDM systems use fibre pairs, one to transmit and one to receive, but bi-directional systems also exist where a single fibre is used to handle signals transmitted in both directions. While research scientists have long known that fibre has the capability to transmit multiple colour streams simultaneously, and in fact has an ultimate capacity that for all practical intents and purposes is limitless, DWDM has only become necessary as the requirement for bandwidth has soared, fuelled by the Internet. As the technology has become commercialised there have been technological improvements in such areas as fibre Bragg grating that have allowed systems to go from two and four channels up to 16 and 40 channels.

CIENA Corporation, based in the United States, has become a leading vendor of DWDM equipment because of its ability to be the first to market a 16-channel system (16 lasers into one fibre). With its MultiWave 1600 product, CIENA was able to offer 40 Gbit/s systems (2.5 Gbit/s streams multiplied by 16). An opticalfibre carrying a standard single laser beam can transmit 32,000 voice or data transmissions on the popular STM-16 architecture, but it can carry 512,000 transmissions using the 16-channel DWDM product with an STM-16 interface.

The engineering challenges

The initial engineering challenge was to combine and amplify the signals of 16 transmitters with distinct wavelengths into one optical fibre. The optimum band occurred in the range of 1530-1580 nm. The vast majority of fibres are designed for 1310 nm signals, but it is possible to transform the multiple signals into the 1530-1580 nm range and then back to the 1310 nm range. CIENA engineers had to determine how to separate these signals at the end of the transmission to be routed to individual detectors as earlier filtering technologies had run into problems in this regard. CIENA's answer was to develop a new filter component called a fibre Bragg grating, which consists of a length of optical fibre in which the refractive index has been permanently modified, generally by exposure to an ultraviolet interference pattern. The fibre grating can create a highly selective, narrow-bandwidth filter that functions somewhat like a mirror and provides significantly greater wavelength selectivity than any other optical technology. The result was a clear way to route the signals out of the fibre path and to the receivers.

Also critical to this process has been development of the erbium doped fibre amplifier (EDFA), which can amplify a signal along an optical fibre by cascading new photons along the line. This allows signals to be boosted without the need for costly regenerators (which convert the signal back to electrons before amplifying it back to photons). DWDM would not have been nearly as feasible had it been necessary to boost every channel by costly regenerators at relatively short distances.

Further improvements came about as CIENA bid to provide its system for the Cable & Wireless Communications' link between Porthcurno and London in the United Kingdom as part of the GEMINI trans-Atlantic fibre system being undertaken by WorldCom. Previously, CIENA had been using 30 dB amplifiers to cover 120 km using four spans but this would not meet the Cable and Wireless specification that the DWDM equipment should be able to reach the entire 550 km between Porthcurno and London without using regenerators. The result was a 25 dB amplifier that could concatenate up to nine amplifiers in a row.

Bringing interoperability to DWDM

The 25 dB amplifier was to become an important feature of CIENA's second-generation DWDM system, the MultiWave Sentry, which offers the potential for spans of up to 1,000 kilometres without regenerators Sentry also brings another important characteristic to DWDM. It allows DWDM systems to be connected directly to ATM switches and Internet Protocol routers/switches, without the need for SDH or SONET multiplexers. This ability to connect directly to a variety of protocols is a distinct advantage of DWDM, and provides an added flexibility that is leading to a greater acceptance of the technology.

Also important to a DWDM system was CIENA's leading-edge development of optical add-drop multiplexers (OADMs). These are important because they allow signals to be added or dropped along a fibre optic route, adding important flexibility to the network architecture. OADMs also can be used to replace EDFAs.

OADMs allow up to four STM-16 (OC-48) channels to be optically added or dropped at any optical amplifier site, without interruption to other channels. They feature three optical channel types: express channels that pass through the OADM without being added or dropped; drop channels that drop selected channels at the OADM site; and add channels that are inserted at the site.

Sentry features a TMN-based element management system. A 17th channel is provided in addition to the 16 traffic channels to provide communications between network elements along a route. It can also locate bit errors and transmission defects via the SONET/SDH overhead (B1 bytes) and report that information to the network management system. Sentry also can detect signal routing errors using the J0 section trace bytes in the SONET/SDH overhead.

DWDM in the local exchange

Until now, DWDM products have been aimed at satisfying growing trunk capacity requirements. Recently, however, local exchange carriers have been complaining that they also are running out of bandwidth; they face some of the same problems that long-distance carriers do, including the fact that the cost of installing new fibre (which can mean civil engineering works) is a very expensive alternative. High-speed TDM also introduces problems of adding and dropping smaller amounts of traffic in local applications. The challenge has been to reduce the cost of DWDM so that it becomes affordable for short-haul applications. CIENA has responded with two product announcements: the MultiWave Firefly and the MultiWave Metro. There are a variety of cost reduction technologies built into these systems with the most important being that the system does not require optical amplifiers.

Firefly, which will be commercially available this year, provides operators with the capability of delivering up to 24 channels over one fibre for distances of up to 65 km. By expanding the number of channels, the capability expands to 60 Gbit/s (24 x 2.5 Gbit/s). Like Sentry, Firefly can interface either with traditional SDH/SONET equipment or directly with devices carrying ATM or fast IP traffic. Metro is currently under development to provide a system for inter-exchange rings and high bandwidth local loop services.

DWDM-centred network architecture

Major DWDM vendors in the USA and elsewhere recognise that DWDM will play a central role in the evolution of network architectures; the CIENA product will be based on the MultiWave core architecture, a DWDM-centred architecture based on packet and cell switching with SONET, ATM and IP short-reach interfaces. It connects Sentry with Firefly and Metro to form a seamless network that will overcome both long-haul and short-haul bottlenecks that plague existing networks.

With this architecture, individual large businesses would be able to have their own wavelengths, which they could use securely to send traffic between different sites. The result will be a network of tomorrow that is far more secure, cost-effective and cost-efficient than today's.

Stephen B Alexander, VP - Transport Products, Ciena Corp, Linthicum,Maryland, USA