O Band WDM

Expand Capacity without Adding New Fiber

By Wim Mostert and Dawn Emms, Aurora Networks

Today it's all about bandwidth per subscriber. Cable operators need ever more so they can continue to offer customers the sophisticated services needed to stay ahead of increasingly aggressive competition. Given the rapid adoption of high definition TV (HDTV), more video on demand (VOD) programs, and higher-speed data services demanded by the mainstream consumer, the trend for more bandwidth will continue, and cable operators will need to increase their network capacity to stay in the game. Bandwidth expansion, however, cannot be at any cost - with everyone looking to save expenses, it is important that any bandwidth upgrade be cost-efficient.

Many factors need consideration when planning an increase in network capacity: availability of fiber, cost of construction, network architecture and type of node deployed. Cable operators have worked hard to standardize deployed networks - typically a 1,310 nm topology with what was considered to be sufficient fiber counts as well as 1,550 nm topologies to cover longer distances and to overcome fiber scarcity. However, for today's needs, fiber availability is typically still a major limitation - and a big challenge to overcome with the high cost of installing new fibers. Overhead cable installation is on the order of $15,000 per mile, most of which is construction cost. For underground installations, the cost can grow upwards of $40,000 per mile.

Forms of WDM

Since the 1990s, technology vendors have been pioneering dense wavelength division multiplexing (DWDM) technology for carrying narrowcast signals. This technique enables more and more wavelengths, and hence narrowcast services, to be carried on the same fiber. The wavelengths are all in the C band of the ITU DWDM wavelength grid, positioned between 1,530 nm and 1,565 nm as shown in Figure 1. This technology has evolved from early eight-wavelength systems on a 200 GHz-spaced grid to the current industry-standard (ITU-T G.694.1) 40-wavelength system with 100 GHz optical frequency spacing.

FIGURE 1: Representation of the ITU wavelength grid

In principle, there is no technology limit to further develop this and for more narrowcast services to be supported on one fiber. Initially, the DWDM wavelengths carried just eight RF quadrature amplitude modulation (QAM) narrowcast channels, containing a mix of video, high-speed data and voice, to each node. Since its introduction, the technology has matured to support 32 RF QAM narrowcast channels per wavelength, with some vendors able to support full carriage of the complete 550 MHz-1 GHz narrowcast band, namely up to 75 RF QAM narrowcast channels, to each node. While DWDM is now widely deployed, it has primarily been a bandwidth expansion tool for cable operators who had standardized on 1,550 nm HFC networks. A typical application, distributed DWDM is shown in Figure 2.

FIGURE 2: Distributed DWDM, dual fiber trunk line architecture optimized to serve lower density areas

For the O band (1,260 to 1,360 nm), solutions have been developed using coarse WDM (CWDM). The original concept behind CWDM technology for digital transport in the telecommunications industry was to enable use of lower cost, uncooled coaxial lasers, which experience wavelength drift over temperature within the window of the CWDM optical filters. However, in our "analog" broadband industry, the high linearity requirements on CWDM transmitters for forward transport necessitate strict temperature control of the lasers. This has resulted in the use of higher cost TEC butterfly lasers to overcome the non-linearity problems, but this defeats the low-cost advantage of uncooled lasers. Today, CWDM forward transmitter applications have so far been largely limited to two wavelengths over a very short distance (typically 7-15 km), limited by inherent fiber characteristics such as dispersion and nonlinear effects, among them stimulated Raman scattering (SRS). These interact with the inherent characteristics of the laser transmitters to cause nonlinear RF signal distortions.

O band DWDM solutions had been investigated before for application in cable TV networks. However, other than the limited CWDM systems just discussed, this region had been avoided because there were no obvious cost-effective solutions to overcome the many non-linear effects in this spectral region where the zero dispersion of the fiber resides. It should be noted that basically every fiber optimized for the 1,310 nm window of operation, including SMF-28 and SMF-28e fibers, and installed since the early 1980s, shares these similar zero dispersion characteristics.

Eight wavelengths

Now it is possible for cable operators to deploy more densely spaced multi-wavelengths in the O band. This is represented in Figure 1. This O band multi-wavelength solution enables up to eight separate wavelengths to be carried in this band in the downstream on one fiber and for a reach up to 30 km. Previously, the limitation was just one wavelength in this band, using traditional 1,310 nm distributed feedback (DFB) lasers, which have been extensively deployed in our industry for the last 15-20 years.

Multi-wavelength solutions for the O band were first introduced in 2006. After careful study of O band non-linearity effects, it quickly became apparent that it was choice of wavelengths and their relative position to one another that would be critical to developing a field-deployable system. The correct wavelength positioning "neutralizes" or avoids the impact of fiber phenomena such as fiber dispersion, four-wave mixing, SRS and cross phase modulation on RF signals modulated onto the laser light. These phenomena, interacting with laser transmitter characteristics in the multi-wavelength system, all cause distortions, mostly composite second order (CSO), in the RF spectrum. Further research has led to the tightening of specifications for optical filters while still incorporating the same robust and reliable filter technology that has been proven over many years in DWDM C band applications.

Practically, with the new O band multiwavelength transmitters and optical filters, building blocks are available that form the basis of this tool for increasing network capacity. The transmitters used are in essence the same proven and reliable "1,310 nm" DFB transmitter technology used for many, many years in this industry. The only difference is that the laser used is wavelength-specific in the O band. However, these lasers are produced using the same process as traditional 1,310 nm laser technology, with the laser wafer growing process now targeted toward specific wavelengths. This production process has been successfully replicated by multiple laser suppliers.

The other consideration, laser wavelength temperature control, is the same as that used and field-proven with cooled DFB lasers, which are used in standard 1,310 nm and DWDM C band applications. Good temperature control is needed for acceptable linear analog multiple RF carrier distortion performance for any technology, DWDM C band, CWDM or multiwavelength in the O band. However, it should be noted that it is wavelength choice and relative position, rather than clever circuitry, which overcomes the non-linearity of the fiber and allows the use of existing laser transmitter technology. The optical filters used to multiplex the wavelengths onto the fiber and, in the field, de-multiplex the wavelength from the fiber are built with the same thin film filter technology proven many times over with DWDM C band applications.

Two to eight

The first introduction of DWDM O band technology was limited to just two wavelengths; subsequent technology advances and production refinements have resulted in eight wavelengths now being supported. Today, these systems can support up to 30 km optical links, depending upon the number of wavelengths and the desired receive optical power at the forward path receiver. However, further refinement in the choice of wavelengths, improved laser technology specifically focused in areas where laser parameters interact with fiber characteristics and non-linearities, and improvement in electronics to further counter non-linear fiber effects will continue to further evolve this technology toward more wavelengths and longer distances over a single fiber.

The current development enables the use of up to eight wavelengths with different narrowcast QAM loadings to be carried on one fiber (along with the analog and digital broadcast channels common on these wavelengths); the cable operator no longer needs to use new fibers to increase network capacity. This results in a very cost-effective solution for increasing network capacity through node splitting. Using traditional methods, to achieve full four-way segmentation in the downstream would typically require four fibers, one fiber for each 1,310 nm DFB transmitter. This is a major capital expense for a cable operator and would take considerable time and planning, driven primarily by the deployment of new fibers.

One fiber

Using today's O band multiwavelength WDM and scalable node technologies, the same end result can be achieved using just one fiber and, if a full scalable node is already deployed, by simply adding additional downstream receivers to the node. This is now a task that can be undertaken in one standard maintenance window, with a minimum of disruption experienced by end-customers. This is shown in Figure 3. Indeed, this example shows how a cable operator can feed a fully-segmented scalable node from one fiber with capacity available to expand the service area further when needed. This example indicates how an additional three nodes can be added and fed via the same fiber.

FIGURE 3: DWDM in the O band; a fiber-efficient tool to increase the capacity of 1,310 nm architectures

Obviously, the choice of scalable node will determine exactly the degrees of scalability and flexibility. Notwithstanding the node choice, there are solutions available today which can support up to eight O band multiwavelength WDM downstream and 20 upstream segments using CWDM, on two fibers for 1,310 nm downstream transmission, and 40 downstream and 80 upstream segments for 1,550 nm DWDM transmission on two fibers. A subset of those wavelengths, carrying downstream and upstream traffic, can also be transported on a single fiber to further preserve fibers and wavelengths. A cable operator now can selectively increase network capacity on a node-by-node basis with both 1,550 nm and 1,310 nm transports.

A further, hidden benefit to WDM technology when used in node splitting applications is that fibers can be freed up. These "new" fibers can now be re-deployed in new applications. For the cable operator wanting new revenue streams, this fiber can be re-deployed to serve a local business, potentially generating significant additional revenue. Alternatively, these fibers can be used to drive nodes deeper into the network, moving toward a fiber-deep network. This architecture not only provides significant bandwidth, supporting both today's and future generation's applications, but also provides considerable operational advantages over a traditional HFC architecture.

Conclusion

Staying ahead of the competition now coming from telco, satellite and local municipalities is critical for cable operators; bandwidth to the home is today's hot commodity. To maintain a competitive advantage, cable operators will need to provide high bandwidth to support the video (in particular, the transition to HDTV), data and voice services that end customers are demanding. However, providing too much bandwidth too soon can be a financial drain on the balance sheet.

With the WDM solutions now available and field-proven for both the C and O bands, cable operators have very cost-effective and fiber-efficient tools in their engineering arsenal. It is critical that technology companies continue to innovate and develop solutions that are customized for cable operators; with installed HFC networks, their network upgrade challenges are unique. With the "weapons" developed for their use today, we believe that cable operators are armed for success to meet the bandwidth per subscriber requirements, both now and into the foreseeable future.

Wim Mostert is director, product management - Optical Transport Systems, for Aurora Networks. Reach him at wim@aurora.com. Dawn E. Emms is director, marketing - Optical Transport Systems, for Aurora. Reach her at demms@aurora.com.