CWDM, GFP technologies for service networks

June 1, 2004

Carriers today want to offer a broad range of high-quality services at competitive prices. Equipment providers that enable carriers to meet the service needs of their customers will lead the recovery in the communications market. But what does this renewed focus on services mean for metro- and access-network technology at the physical layer (PHY)?

Carriers need to shift away from technology-driven toward service-driven PHY architectures. The service network is a new concept that embodies this shift.

Service networks enable carriers to operate in a tactical mode of operation, whereby the capacity of the physical layer is increased to accommodate the specific needs of secured revenue-generating services. The PHY is able to efficiently transport the new services, reusing existing infrastructure to the maximum extent, and capacity is deployed only where and when it is required. It's the natural step in a maturing industry.

CWDM and generic framing procedure (GFP) are two technologies well suited to the service network concept where simplicity, flexibility and reliability enable an architecture that allows the carrier to focus on the delivery of services.

Simplicity of use and reliability of operation are two essential design aspects of the PHY for service networks. The term "plug and play" may be overused, but it's a key concept to enable the widespread deployment of optical networks close to the edge of the network.

Equipment suppliers need to focus on the lifecycle of the equipment and make each stage as simple as possible for the carrier from network planning and design to installation and deployment, commissioning and provisioning, and lifecycle maintenance. Advances such as the removal of fans and filters from units in remote locations—thereby reducing the number of scheduled maintenance truck rolls and the integration of complex wavelength-specific modules for WDM systems, simplifying installation and maintenance—are important design aspects for a network focused on services. Put quite simply, the complex stuff going on in the PHY should not be seen or heard; it should just work—in the background—allowing the carrier to concentrate on customers' needs.

Carrier class reliability is the other important design aspect for the PHY. There is more to carrier class than reliability and availability, however. The ability to offer performance monitoring of each channel is important in the delivery of service-level agreements and the creation of specific demarcation points with known performance. Monitoring channels on a link-by-link basis rather than across the whole path of the network also offers advantages in this area.

CWDM fits naturally into a service network—it has all the important attributes. A service network needs a few channels, each efficiently packed with services. That is the best way to produce an overall lowest cost per service. First-in cost is important, and CWDM delivers on that with low up-front and incremental capital costs.

The most important aspect is CWDM's capability to offer a low total cost of ownership. However, it's not good enough to just assume that any CWDM system will be simple to use and offer carrier class reliability. It's important to realize that CWDM can offer these advantages, yet system selection matters CWDM doesn't always work by default.

While CWDM is well suited as the transport infrastructure for a service network, GFP is the multiplexing technology that will allow carriers to efficiently use each wavelength channel. GFP enables data, storage, and video protocols to be framed in a standard way so they can be transported along with SONET/SDH. GFP is the ultimate in simple and flexible service delivery for the edge of the network. It is also the answer to the ongoing debate about Ethernet versus SONET/SDH as a transport technology.

GFP and CWDM are now standardized. International Telecommunication Union ITU-T G.7041 is the standard for GFP and ITU-T G.694.2 and G.695 are the standards for CWDM.

ITU-T G.694.2 defines the wavelength grid for CWDM. It's important to note that CWDM has a wavelength grid, not a frequency grid, as is the case for DWDM. The G.694.2 grid was recently altered by 1 nm to accommodate current industry practice; the wavelength grid is now a 20-nm grid starting at 1271 nm and finishing at 1611 nm (see Figure 1).
Figure 1. The CWDM wavelength grid as defined in ITU-T G.694.2, showing the attenuation and dispersion of ITU-T G.652 fibers (both conventional and low water peak).

Most commercial CWDM systems focus on eight transmission wavelengths from 1471 to 1611 nm. This region avoids the water peak experienced by the majority of the installed base of fiber and uses the lowest loss window. Future systems will use the full spectrum of CWDM, allowing increased capacity and scalability.

ITU-T Recommendation G.695 was approved as a standard in Geneva in February and is thus the newest standard in optical transmission. G.695 sets the application codes and optical-parameter values for CWDM. It describes the optical PHY.

There are two architectures described in G.695, and for very good reason, since there are two very distinct architectures to which CWDM can be put to use. The different architectures are referred to as the "black box" and "black link" models.

In a black box system, the CWDM platform takes tributary inputs that are not compliant with G.695, converts them to G.695-compliant signals, and multiplexes several such inputs together onto a common output. Black box systems work well with existing legacy equipment. In a service network, the carrier needs to integrate services such as voice over SONET, Frame Relay over ATM, data over Ethernet, storage over Fibre Channel, and video over digital video broadcast. Each service platform should be maintained to maximize the reuse of existing infrastructure and minimize costs and disruption to services. That is done by taking the outputs from these respective platforms—typically at 850 or 1310 nm—and inputting them into a CWDM black box system.

It is essential to have an optical-electrical-optical (OEO) stage in a black box system. The incoming 1310-nm optical signal is converted to an electronic signal, then to a G.695-compliant CWDM optical signal multiplexed with other such signals and output (see Figure 2). An electronic step enables system designers to put the intelligence in the electronics and integrate the OEO stage into a low-cost integrated module.
Figure 2. CWDM black box system design as defined in the ITU-T G.695 standard, which was ratified in February, showing the point where application codes are defined in the standard.

By integrating the OEO stage, it is possible to offer a simple system that does not have wavelength-specific interfaces, transceivers, filter modules, or the like. For integration into existing equipment for which a transponder function is required, an integrated OEO system that removes such modules is simple to deploy and maintain.

An integrated OEO system can also be designed as a reconfigurable optical add/drop multiplexer (ROADM) or CWDM regenerator, allowing full remote configuration or network extension. The use of a CWDM regenerator enables deployment of CWDM ring networks that extend significantly beyond 80 km to several hundred kilometers. CWDM ROADMs and regenerators are commercially available.

With such integrated plug and play design, a technician in the field can deploy a system without even knowing the system uses WDM technology. This type of a system design is in line with the concept of the service network where the smart technology is hidden "under the hood," allowing carriers to focus on delivering quality services to their customers.

In a black link system, the CWDM system takes tributary inputs already compliant with G.695 and simply multiplexes several such inputs onto a common output. Thus, black link systems are more suitable for new deployments. These systems reflect the direction the market is moving, using pluggable optics at the network edge. Black link boxes enable the use of G.695-compliant, small-form-factor-pluggable (SFP) transceivers on the edge devices and CWDM multiplexers that do not contain OEO transponder stages (see Figure 3).
Figure 3. CWDM black link system design as defined in ITU-T G.695, showing the points where application codes are defined in the newly ratified standard.

This architecture reduces the capital cost of the system, but some work still needs to be done, particularly in the area of performance monitoring to ensure that the black link system is easy to manage and maintain. If something goes wrong, how easy is it to determine where the fault is? These challenges will be solved and represent opportunities for enterprising system manufacturers. Going forward, the ITU-T intends to take G.695 toward full transverse interoperability. As shown in Figure 3, when that happens, the optical multiplexer and demultiplexer may be sourced from different vendors with the interface between the mux and demux specified by G.695.

Of course, one of the advantages of the use of pluggable optics on edge devices is the flexibility that they allow. GFP edge devices will almost all use SFP transceivers on their aggregate output, enabling these service-oriented devices to be deployed with a low-cost 850-or 1310-nm output. That may be appropriate for initial deployments where WDM is not yet required. If the GFP system is added to an architecture that has predominantly legacy edge devices, which do not have pluggable optics, the use of an 850- or 1310-nm output connected into an integrated black box system creates a powerful architecture that is efficient, proven, and reliable. In short, the basic architecture building blocks needed to create the ideal network for any given specific scenario or set of requirements already exist.

A lot of debate recently has focused on the role of hybrid CWDM/DWDM systems. As discussed here, CWDM fits well into the design for service networks. DWDM was developed for the creation of large-capacity transport systems and is less oriented toward services.

The industry is still a ways away from large-scale deployment of wavelength services. As such, CWDM and DWDM have important roles to play in the network. Is it not therefore better to have an optimally designed CWDM system and a separate optimally designed DWDM system? CWDM systems are lower-cost not only because they have lower-cost optoelectronics, but also because they do not contain many of the required features of large-scale, high-capacity DWDM systems.

If a carrier does feel the need for basic cut-down DWDM to enable a scalable service network, then the flexibility of pluggable optics will prove to be the savior. An edge GFP device with an SFP aggregate can be deployed with a "gray" 1310-nm interface to start with—no WDM is required. As the network grows, G.695 CWDM pluggable optics may be used on the same aggregate to increase capacity. Recent indications from the market show that DWDM SFP transceivers are emerging, so the natural expansion to DWDM will use the same SFP aggregate port on edge devices such as GFP multiplexers. The resulting architecture has flexibility and scalability and retains simplicity and cost-effectiveness.

Dr. Richard Lauder is the chief technology officer and co-founder of RBN (San Francisco). He can be reached at [email protected].