Low cost and simplicity make CWDM a good choice for demand-driven networks.
By DR. RICHARD LAUDER and ROSS HALGREN, RBN--Coarse WDM (CWDM) offers significant key benefits for metro optical transport networks. Purpose-built CWDM systems can be low-cost and simple to use and are well suited to the needs of carriers today.
CWDM is a form of multiplexing that has wider spacings between the wavelengths used than dense WDM (DWDM). Also, unlike DWDM, it uses a far broader photonic band spectrum. Up to 18 wavelengths can be transported on a single fiber using CWDM. The costs of deploying CWDM are significantly lower than DWDM.
CWDM made an important step forward in 2002 when the International Telecommunication Union (ITU-T) standardized the CWDM wavelength grid. Named ITU-T G.694.2, it is based on a 20-nm wavelength grid spanning 1270-1610 nm. This achievement has allowed CWDM to move forward rapidly in much the same way as DWDM took off after the development of the 100-GHz frequency grid, now named G.694.1.
Figure 1 shows the CWDM wavelength grid. Most commercial CWDM systems available today use the eight wavelengths from 1470 to 1610 nm, covering the so-called S-, C- and L-bands. Standardization work continues within the ITU-T to develop full application codes for CWDM. The standard, which will be named G.695, is expected to be finalized in October. One of the topical discussions within the G.695 experts group is the wavelength bands to be used by CWDM. Eight-wavelength systems covering 1470-1610 nm will be included within G.695; however, issues being discussed relating to high transmission losses around the water peak in standard G.652 fiber at 1380 nm and high transmission losses below 1300 nm may compromise the ability to standardize 12- or 16-wavelength systems this year.
Demand for CWDM
One of the key drivers pushing demand for CWDM is the move from supply-driven to demand-driven network builds. Carriers can no longer afford to overbuild extensively and hope that they will be able to fill this excess capacity. Today, a visible revenue stream is won and network capacity is deployed to meet the needs of the revenue stream. The network must be profitable from the start and its return on investment must be short.
CWDM is well suited to demand-driven network builds. CWDM has a low first-in cost, much lower than DWDM, and it scales to a capacity in line with demand. There is a need for physically small transport systems that are simple to use and low-cost and provide enough capacity for today's demand and possibly tomorrow's but not next decade's. Better to use low-cost systems that will return a profit soon rather than systems that will supply possible demand for the next 10 years or more but will take most of this time to return on their investment.
Carriers are also keenly aware of the rapid pace of technological improvement. With technology improving so rapidly, it doesn't always make financial sense to overbuild today's technology to supply future demand.
So what are the advantages of CWDM? Two key ones are:
• CWDM's low first-in cost and low channel granularity, enabling deployments to closely match guaranteed revenue streams and allowing capital expenditures (capex) to track revenue generation.
• CWDM's simplicity of network design, implementation, and operation, which enables an easy adoption and implementation of the technology by carriers. That is in contrast to some of the more complicated metro DWDM systems that have been available in the past.
By taking each of these in turn, let's first look at why CWDM is less expensive. A major cost reduction results from the use of lower-cost optoelectronics. Directly modulated CWDM lasers with bit rates up to 2.5 Gbits/sec are optimized for low cost. Their design is based on tried and proven distributed-feedback (DFB) technology. DFB technology has the benefits of a narrow linewidth with highly suppressed side modes, thus providing similar low dispersion performance to directly modulated DWDM lasers. As a result, CWDM lasers are capable of transmitting 2.5 Gbits/sec over distances of 80 km on ITU G.652 fiber.
The low-cost, small-power, and reduced-space benefits of CWDM laser transmitters result from their uncooled design. That means they do not have bulky heat-sinks, control circuits, and thermo-electric coolers (TECs) coupled close to the laser chip, which saves electrical power and space. A typical optical output of at least 1 mW (0 dBm) is achieved with low-cost CWDM lasers.
Low-cost CWDM filters are implemented using thin-film-filter (TFF) technology. They are available as discrete single-channel filter devices and integrated multiplexer/demultiplexer devices with typically four or eight wavelength ports. Various configurations of these devices can be used to implement a multichannel optical-add/drop-multiplexer product. CWDM filters can be specified for unidirectional transmission on two-fiber networks or for bidirectional transmission on single-fiber networks. The latter option has the advantages of lower first-in cost for leased-fiber applications and reduced fiber count for fiber exhaust applications.
Although CWDM systems can be low-cost, it is important to point out that to gain maximum benefit the system must have been designed as a low-cost CWDM system from the start. Taking a metro DWDM system and retrofitting it with CWDM lasers and filters will not result in the cost savings that could be realized from a purpose-built system. A system architecture designed for DWDM will invariably have a design and cost appropriate for high-cost DWDM blades. Taking a banded DWDM metro system and producing a new variant with only one channel per band has even less benefit. There are very little cost savings to be had for such a nonstandard system. It is still a DWDM system requiring DWDM cooled lasers. There are cost advantages to be had from CWDM, but carriers must appreciate that only those systems designed as CWDM from the outset will offer the full advantage.
Now let's turn our attention to simplicity. What makes CWDM so inherently simple to use? The crucial aspect is that CWDM has been designed as a basic transport system with few parameters that need user optimization. If you know DWDM, don't think you know CWDM, as in many ways it isn't so much how simple CWDM is, but how comparatively complex DWDM systems can be to configure.
DWDM systems can require complex power budget calculations per channel, something that is made even more complicated when new wavelength channels are added or DWDM is used in ring networks. Include optical amplifiers in a DWDM ring network and the resultant system is complex and operationally hard to configure or will incorporate costly automatic power monitors and balancers for each channel at various points around the network.
For metro-ring networks, a regenerative CWDM system will provide the simplest possible solution. With regenerative CWDM, each CWDM channel is regenerated at each node. Over the past decade, the focus has been to move away from regeneration toward optical amplification, but the tables are turning. The cost of CWDM optoelectronics is such that it is both simpler and less expensive to regenerate than it is to amplify. Regenerative CWDM is much less expensive and simpler than optically amplified DWDM. The fact that CWDM can be deployed in extreme environments such as that typical of outside plant curbside cabinets also means that an inline CWDM regenerator can be located in such a street cabinet rather than needing an air-conditioned hut, building, or controlled environment vault.
As regenerators fully reamplify, reshape, and retime the outgoing signal, they compensate for any accumulated dispersion¿an advantage over optical amplifiers. Using inline regenerators allows low-cost directly modulated CWDM OC-48 or STM-16 signals to be transported over many hundreds of kilometers.
So what are the primary applications that are demanding the use of CWDM? Three areas are worth investigation:
• The introduction of data services such as Gigabit Ethernet (GbE), Fibre Channel, Escon, and Ficon into business enterprise solutions using a CWDM fiber-to-the-building (FTTB) architecture. The possibility to maximize the use of existing infrastructure and collapse the data access network with the existing voice network at the physical layer are compelling features of CWDM here. Low first-in cost is critical for this market. Carriers are looking for a profitable return on their first business customer in any one location. That user may require several SAN services or a single GbE delivery, but either way low cost for that first couple of CWDM channels is paramount.
• The growth of voice-, video-, and data-service offerings delivered over DSL from remote cabinets on a CWDM-upgraded digital-loop-carrier (DLC) network. Many DLC networks were deployed for POTS service only, with just four or six fiber cables used with OC-3 terminal equipment. With the upgrade to ATM-based remote DLC-fed DSL access multiplexers, an OC-3c backhaul is rapidly filled, especially for DLC rings feeding more than one remote terminal or delivering high-bandwidth video services. In this scenario the reuse of existing DLC infrastructure is paramount and CWDM enables rapid deployment of new higher-bandwidth services at minimal overall cost. The need to trench new fiber is avoided and existing DSL and POTS terminals are undisturbed, thus minimizing customer outages. For this application, the requirement for CWDM to be outside plant-hardened for placement in curbside cabinets is paramount, as are small size of the terminal equipment and low power usage. Commercial equipment suitable for such applications is now commercially available.
• The deployment of low-cost, small-channel-count CWDM systems within the metro core network. In this case, each deployment can closely match the capacity required for the guaranteed revenue, minimizing the speculative capex spending the carrier will have to make. If extended span lengths or multinode rings are required, then regenerative CWDM systems are preferred. The ability to use CWDM rings as an underlay to SONET or SDH rings is optimized if the CWDM equipment can reuse wavelength channels around the ring so that one channel can provide SONET add/drop connectivity to many nodes but by only utilizing one CWDM channel. The growth of video distribution within the metro networks of telecommunications carriers means that drop-and-continue functionality is preferred. In this scenario, the CWDM channel is dropped at a node and the same channel is continued onto the next node in the ring. Regeneration of the continuing channel simplifies network design in such cases. The efficiency of one CWDM wavelength broadcasting a single unidirectional channel to many remote nodes is compelling.
CWDM and DWDM interworking
One aspect of CWDM that gets frequent attention is how CWDM and DWDM can interwork. As a carrier with an existing DWDM network, how can CWDM be best integrated with it? There are three alternatives:
• The two systems can be integrated via standards-based transponders, typically at 1310 or 850 nm. That would be the standard way any SONET ADM would be integrated into a DWDM system.
• The possibility of integrating CWDM functionality onto the core networking products located in the central office. That avoids the need for a transponder-based interconnection and allows greater unification of architectures.
• There is even the possibility to integrate DWDM and CWDM on the same fiber strand. That could be enabled through a combination of the two ITU channel maps, G.694.1 and G.694.2. As shown in Figure 2, it is possible to fit eight 200-GHz DWDM channels inside one CWDM channel. This approach enables G.694.1 wavelengths to be banded inside G.694.2 wavelengths.
CWDM is here, it is being standardized, it fits with today's business models, and it offers many advantages in enabling low-cost and simple transport of a multitude of services.
Dr. Richard Lauder is chief technology officer and co-founder of RBN Inc. (San Francisco). He can be reached at 415-874-3516 or email@example.com. Ross Halgren is senior product manager and co-founder of RBN.