Gabriel Odeh
New technologies such as ultrahigh-density and tunable transceivers, Raman amplification, and FEC are reducing long-haul transport costs but are making system design more complex. In this overview, the author presents a four fold approach to understanding the different forces that influence network design.The technologies behind DWDM systems are painting a different picture for engineers who design optical networks now than in the pioneer days of DWDM networks. Just three years ago, carriers deployed DWDM systems with slower-data-rate transceivers and erbium-doped fiber amplifiers (EDFAs).Today, DWDM system manufacturers are offering a suite of technologies such as: ultrahigh-density systems with 10-Gbit/s and potential 40-Gbit/s channels, tunable lasers, forward error correction (FEC), Raman amplifiers, and dynamic optical add/drop multiplexers (OADMs). In addition, DWDM system manufacturers are expanding the C-band (1525 to 1565 nm) with high channel counts, though significant advancements are being made in the L-band (1570 to 1620 nm) and S-band (1480 to 1520 nm) spectrums. In short, these new technologies have changed the landscape that characterizes optical networks.
We will look at the basic system-design questions that DWDM network engineers must consider when building their optical networks. In addition, we will introduce the emerging technologies and how they are being used to build a more robust and cost-effective network.
FOUR-FORCE NETWORK DESIGN MODEL
There are many different issues that a network design engineer must assess when building a DWDM network. These issues are similar to different pieces of a puzzle, where each piece contains limited information. When the puzzle is put together, however, a clearer picture is available. The four-force network model helps network design engineers see more clearly the different forces that interact when building a DWDM network (see Fig. 1).
When designing an optical network, the engineer must consider these four forces: capital budget, network infrastructure, transmission capacity and traffic patterns, and upgrade path scenario.
Capital budget. Capital budget consists of two parts: initial budget and future budget. Clearly, the carrier should provide initial budget guidelines or ballpark figures. These guidelines set the foundation and limits as to what a carrier can build in the proposed network. In many instances, carriers do not really know what their upgrade path will look like or how their traffic patterns will grow over time. For this reason, the future budget is more elusive in nature.Network infrastructure. Network infrastructure looks specifically at the carrier' s fiber-type and distance between regeneration, add/drop, and terminal sites. For example, will the carrier use non-dispersion-shifted fiber (NDSF), non-zero dispersion-shifted fiber (NZDSF), or dispersion-shifted fiber (DSF)? In addition, what margins does the carrier require in the network? The answers to these types of questions shape the transmission capabilities of DWDM networks.
Transmission capacity and traffic patterns. Transmission capacity refers to the amount of aggregate capacity that a carrier transports from one location to another location. To a large extent, this force determines the location of terminals and add/drop sites. The aggregate capacity can be either many channels running at one data rate (for example, 160 channels at 10 Gbit/s) or a mix of 2.5-Gbit/s, 10-Gbit/s, or even 40-Gbit/s channels.
The network design engineer should also know how much and what kind of traffic the carrier plans to transport initially and over the next few years. For example, will the carrier offer Gigabit Ethernet or 10 Gigabit Ethernet services, protected or unprotected services, virtual lambdas? Also, how will these traffic patterns evolve over time? Finally, does the carrier want a truly dynamic network? In other words, will OADMs support channel or band-switching applications? Knowing the answer to these types of questions will help the network design engineer recommend and build the appropriate DWDM network.Upgrade path. There are two traditional upgrade paths that carriers can implement: populate the unfilled C-band (or L-band) channels in a system with high-density transceivers, or open a new band spectrum on top of an existing system (C-band + L-band systems).
Carriers prefer populating their unfilled bands to opening a new band spectrum. The main reason revolves around cost. For example, if carriers want to open a new band spectrum in their DWDM networks, then they will need to buy optical line amplifiers (OLAs) and OADMs that support the new band. In contrast, filling unfilled channels leverages the carrier' s equipment that is already installed. Also, this upgrade path is very attractive with high-density transceivers.
CAPACITY-ENHANCING TECHNOLOGIES
Dense wavelength-division-multiplexing system manufacturers offer a suite of enhancing technologies to support higher channel counts and/or longer distances (between terminals and OLA sites). These technologies have changed the DWDM landscape for carriers.
Before high-density systems were introduced, the C-band spectrum could only support a maximum of 96 10-Gbit/s channels spaced 50 GHz apart. On the other hand, with high-density 10-Gbit/s transceivers, DWDM system manufacturers can support 320 channels in the C-band or L-band spectrum alone. In other words, high-density technologies will allow carriers to transport over 3.2-Tbit/s capacity in a single-band spectrum with 10-Gbit/s channels.
As a result, this technology saves carriers money because the high-density system leverages the carrier' s OLAs and OADMs that are already installed. A high-density system not only postpones a band-spectrum upgrade (that requires new OLAs and OADMs for the new band), but also delays implementing 40-Gbit/s channels until 40 Gbit/s becomes more cost-effective.
Tunable lasers. The DWDM industry foresees tunable lasers as one of the newer technologies that will enable carriers to reduce their fiberoptic network costs. These savings rest primarily on three pillars: spares and inventory reduction, system-component standardization, and operational savings.
If a transceiver is band-tunable, then only 12 10-Gbit/s transceivers are needed to cover the C-band spectrum (instead of over 160 10-Gbit/s transceivers with 25 GHz spacing). By buying and managing fewer spares, carriers should reduce their capital and operational expenditures. Also, from an operational point of view, tunable lasers will decrease the number of different channel configurations that a carrier must support in its networks.In addition to these cost savings, tunable lasers will enable carriers to increase their network' s flexibility and scalability, and support dynamic bandwidth-provisioning applications (see Fig. 2). For instance, if a carrier' s customer requires more bandwidth, then the carrier can easily light (turn on) an extra channel with a tunable laser. This example illustrates how tunable lasers provide not only quick bandwidth-on-demand services, but also a more flexible network. With a dynamic OADM or next-generation optical crossconnect, tunable lasers can reconfigure a DWDM network in real time by rerouting an optical signal based on its wavelength.
Forward error correction. Forward-error-correction (FEC) technology uses statistical algorithms to reduce bit errors in a DWDM system. In other words, this technology enables DWDM channels to tolerate lower signal-to-noise ratios. Because a DWDM channel can tolerate a lower optical signal-to-noise ratio (OSNR), that channel can now travel farther. For this reason, DWDM system manufacturers leverage FEC technologies to increase their systems' channel count and/or distance between electrical regeneration (see Fig. 3).
Dynamic OADMs. Optical add/drop multiplexers enable carriers to add/drop optical channels at particular nodes that are between terminals. Hence, these devices provide carriers with an added degree of flexibility in their DWDM networks (see Fig. 4).
In general, there are two types of OADMs—static and dynamic. Static OADMs require physical changes in the hardware to alter add/drop configurations. On the other hand, dynamic OADMs remotely switch optical channels to either pass through the OADM as an express band or add/drop channels at particular locations.Because dynamic OADMs are flexible, the network design engineer must consider this feature during the early network-design stages. For instance, a dispersion-compensation layout for a static network may be different than a dynamic network' s dispersion-compensation layout. In addition, the channel count for an add/drop band may be different than the channel count for an express band.
With dynamic OADMs, DWDM system manufacturers can trade off OSNR to get more channels in an add/drop band because that band will travel less distance than the express bands going from terminal to terminal. For this reason, the network design engineer must ensure that there is enough channel OSNR (for each channel) to also support the express mode.
In short, as a channel propagates over longer distances and through more OLAs/OADMs, its OSNR decreases. Consequently, it is possible to support a channel in an add/drop mode, but not in an express mode. For this reason, a network design engineer needs to plan for a dynamic network from the start.High-performance amplifiers. To a large extent, the optical line amplifiers and the distance between them determine the total capacity that a DWDM network can support. The two types of amplifiers that DWDM system manufactures offer today are EDFAs and Raman amplifiers. Many DWDM system manufacturers combine their Raman amplifiers and EDFAs to make an EDFA-Raman hybrid for high-loss spans or ultralong-haul applications.
The first and most common optical amplifier is the EDFA. In an EDFA, a pump laser excites the erbium-doped electrons to a higher energy level (or orbital). As the weak incoming signal enters the amplifier, the electrons drop to a lower energy level (or orbital). At the same time, photons are released. In this manner, an EDFA boosts the power of DWDM channels. The EDFA is also a broadband amplifier that operates in the 1550-nm window.
For most long-haul amplifiers, the optical gain is not uniform for all wavelengths. This nonuniform gain phenomenon is known as tilt. Tilt is an issue for ultralong-haul applications and reveals its problems when multiple amplifiers are cascaded over a long-haul route. In short, tilt worsens as optical channels pass through more amplifiers (see Fig. 5).
To mitigate the effects of tilt, many EDFAs use gain-flattening filters. Dense wavelength-division-multiplexing system manufacturers are also deploying dynamic gain equalizers to reduce tilt (and tilt accumulation) in their DWDM systems' gain response. These tilt-reducing tricks enable DWDM system manufacturers to support long-haul and ultralong-haul applications with high channel counts.
Raman amplifiers. There is a lot of talk about Raman amplifiers, which use the existing fiber as the gain medium and harness the effects of Raman scattering to amplify weak optical signals. A Raman amplifier uses a pump laser to excite the vibrational modes of the fiber' s atoms. The incoming weak signal then stimulates the excited fiber atoms to release photons in a specific wavelength window. In this manner, a Raman amplifier boosts the optical power of a channel' s wavelength.
Raman amplifiers are attractive because these amplifiers can operate over a broader range than EDFA amplifiers. Also, many DWDM system manufacturers use Raman amplifiers with their EDFAs for ultralong-haul applications and/or high-loss segments (30 dBm).
Dispersion compensators. As optical channels travel through the fiber, dispersion is introduced. This dispersion spreads optical pulses, which causes the signals to interfere with each other (see Fig. 6). Dispersion also becomes more of an issue with higher data rates and/or longer distances (between terminals and OLA sites).
Because dispersion hinders the ability of a DWDM system to transport information, DWDM system manufacturers use dispersion-compensation units to overcome these crippling effects. The two most generic types are chromatic-mode dispersion compensators and polarization-mode dispersion compensators.
Core networks are very much alive and vibrant, especially as fiberoptic system manufacturers deploy these new capacity-enhancing technologies to reduce optical transport costs ($/bit/km). One thing is certain: new technologies will enable higher channel counts and longer distances between regeneration sites. As these new technologies roll out, carriers will reap the benefits of reducing their optical transport costs.
Gabriel Odeh is a core transport product manager at Ciena, 1201 Winterson Road, Linthicum, MD 21090. He can be reached at [email protected].