Dispersion-tolerant modules

By using emerging signal processing techniques, carriers can increase capacity and transmission distance more economically.

By Brian Schreder and Jason Stark, Kodeos Communications

In recent years, service providers have deployed high-channel-count DWDM systems, expecting that they could fill the equipment to capacity as extra bandwidth was needed on specific routes.

But now, the downturn is forcing them to minimise the cost of upgrading each network bottleneck. So, they are focusing on buying extra line cards to finish populating partially filled chassis, and adding OC-192 channels to OC-48 rings to enable mostly filled links to add capacity. Signal processing enabled transponders to employ advances in optical transmission to provide economical solutions.

Equipment vendors can use new transponder technology to deploy systems with fewer electrical regenerators and longer dispersion-limited links, and remove dispersion compensating modules (DCMs) and the associated optical amplifiers.

Metro/regional long-reach markets
From 1998 to 2000, telecom equipment spending grew at over 30% per year, with explosive bandwidth demand driven by the Internet. The market slowed dramatically in 2001, as most initial infrastructure network build-outs neared completion. While revenues from new system deployments fell in 2001, revenues from line-card installations grew at a double-digit rate. In a recent RHK study of 336 city-to-city links, almost a quarter of the OC-48 and OC-192 channels deployed had capacity utilisations of greater than 80%. With demand continuing to increase and capacity utilisation approaching 100%, the need to purchase extra line cards will continue.

Carriers typically light only a fraction of the wavelengths on newly deployed systems, so most fibre routes will not be filled to capacity this year. So, 2002 will be a slow year for greenfield deployment, as carriers fill unused line-card slots and upgrade OC-48 rings to OC-192. As new cards are added, OC-192 line cards can be used to raise capacity on partially lit fibre. At the same time, many carriers encounter transmission problems as the more dispersion-sensitive links become populated. Transponders with increased dispersion tolerance, for transmission over 120-140km, provide significant advantages in both applications, as well as in reducing the number of DCMs and regenerators needed in ring upgrades and greenfield builds.

Consider an isolated space (Fig. 1). In standard on-off keying, the field from the marks will spread into the time slot of the space, leading to intersymbol interference (ISI). This smearing from the surrounding marks raises the bottom of the transmission eye, bringing the received level of the space close to the sampling threshold and diminishing the OSNR margin. Spectral shaping transponders address this problem by opening the transmission eye, regaining lost OSNR margin, even at high dispersion.

But the DCMs are expensive and imperfect, with the amount of dispersion compensation varying over the wavelength band due to the dispersion slope mismatch of the dispersion-compensating fibre (DCF) used in constructing the DCM (Table 1). As a result, each wavelength channel encounters a different amount of accumulated dispersion, in accordance with the dispersion map (Figure 2). Channels which encounter high levels of residual dispersion suffer high levels of impairment. Even in dispersion compensated systems, signal processing transponders increase the tolerance to this residual dispersion, reducing the penalties and costs associated with dispersive transmission.

For a given amplifier spacing, penalties associated with amplifier noise accumulate in proportion to system length. Penalties are low for moderate amounts of dispersion and then rise rapidly when the power penalty reaches a level of 2dB. As a result, systems are typically designed to achieve a fixed level of dispersion penalty, after which the signal must be received and regenerated. Advanced modulation transponders provide additional system engineering margin for dispersion-limited links over traditional transponders.

Advances in optical transmission
Since 1998, new optical transmission technologies have reduced system cost and raised transmission efficiency. In the '90s, optical systems leveraged fibre technology advances and higher channel rates. By using advances in high-speed IC technologies, high-capacity systems have become available at competitive prices. As the advantages of silicon process scaling become available to the high-speed optical systems engineer, technologies common in other communication systems are now being implemented for optical transmission.

Further signal processing techniques are now being implemented which compensate for the most significant transmission line impairments. Transmitters that incorporate spectral shaping are more tolerant to dispersion and keep the transmission eye open over greater distances. These shape the spectrum of optical transmission, tailoring the signal to the characteristics of the fibre link. This process, which applies to Return-to-Zero (RZ) as well as Non-Return-to-Zero (NRZ) modulation, extends beyond simple chirp control. Improvements can be static, targeted for a broad set of applications, or dynamic, adapting to the conditions of specific routes. These spectral shaping techniques yield higher spectral efficiency, for greater bandwidth utilisation, and work synergistically with FEC processing.

Receivers with equalisers can dynamically adjust to changing line characteristics, and complement transmit-side spectral shaping. Equalisers directly mitigate ISI in the electronic domain, subsequent to signal detection. This enables offset of penalties from each impairment individually - chromatic dispersion, polarisation mode dispersion, and nonlinear signal distortion. Equalisers take full advantage of scaling trends in silicon device manufacture, replacing expensive optical compensation components with inexpensive, highly integrated electronics.

Applications

• Upgrading OC-48 to OC-192 spans

Service providers want to keep capital equipment costs low, so a problem arises when heavily loaded links are nearing capacity. Traditionally, they had two options: either make "spaghetti" out of the network by routing traffic around heavily loaded links, or add a green-field build at significant cost. A third option is now available by adding an OC-192 transponders, enabling the remaining links to carry four times the capacity while maintaining OC-48 system engineering rules.

Many metro and regional OC-48 rings have a circumference of 100-140km. These were designed on OC-48 system engineering rules, so dispersion compensation elements are not located in intermediate huts or in the optical multiplexing shelves. When a span's capacity utilisation approaches an upgrade threshold, another wavelength is needed. By upgrading to a signal-processing-enhanced OC-192 wavelength capable of 120-140km transmission, the original OC-48 engineering rules are maintained - engineering, installation, facilities, and test costs are reduced, while keeping capital expenditures to a minimum.

• Metro/regional networks

As fibre moves closer to the end-user, metro and regional networks expand, requiring more add/drop nodes. In many applications, traffic is aggregated for transmission over the long-haul part of the network. High-bandwidth metro rings have a circumference of 100-150km.

Metro rings are limited by chromatic dispersion and other linear impairments. Advanced modulation transponders spectrally condition the light before transmission to overcome these impairments, enabling transmission over 120-140km without DCMs and optical amplification. By removing DCMs at intermediate huts, the cost of first wavelength deployment is cut by over 50%. As metro networks evolve, intermediate nodes are inserted into existing spans, requiring DCMs to be re-allocated. With no DCMs, networks built using dispersion-tolerant transponders require no such re-engineering.

Also, as metro rings expand into regional rings (200-300km), DCMs and amplifiers are only needed at one intermediate node about half-way around, if advanced modulation transponders are deployed. This is a significant cost saving over traditional transponder-based equipment that needs dispersion-impairment-reducing elements every 40-80km. As signal processing transponders improve over the next 12 months, rings with circumference of 200-300km will not need to use DCMs or other linear-impairment-reducing equipment.

• Electrical regeneration removal

Using a commercial long-reach module, a 1.5dB dispersion penalty corresponds to 850ps/nm of dispersion, but new signal-processing-based transponders can transmit up to 1300ps/nm without dispersion penalty. So, the 1.5dB of margin formerly reserved for dispersion penalty can be allocated to OSNR margin, to achieve about 1.5 times the number of amplified spans allowed before signal regeneration (Fig. 3). The result is one third fewer regenerators for the same transmission distance.

As this example shows, advanced modulation transponders allow up to 50% more optical amplifier sites to be traversed before electrical regeneration. So, many long-haul links will not need electrical regeneration at intermediate nodes and can be upgraded at lower cost. Electrical regeneration cards can be replaced with a lower-cost optical amplifier module.

Summary
The long-reach market in 2002 will be dominated by adding line cards to existing systems rather than deploying new systems. The new signal-processing-based transponders deliver performance enhancements, while lowering system cost. These produce chromatic-dispersion-tolerant signals, remove ISI impairments from received signals, and allow larger networks with reduced dispersion compensation and signal regeneration requirements. They also transmit OC-192 signals on networks designed for OC-48, extending the life of existing systems. Finally, the signal processing is complementary to FEC techniques - used together, they deliver significantly more OSNR margin to the system integrator, leading to more flexible engineering rules, better system performance and lower costs to the end user.

Jason Stark
Founder, CTO & President
Kodeos Communications
[email protected]