Coarse wavelength division multiplexing (CWDM) is a technology that adds capacity to networks by using the optical spectrum between 1271 nm to 1611 nm (nanometers). It is able to transmit up to 18 wavelengths instead of the 2 standard 1310 and 1550 nm windows. In this article we provide the background of CWDM and how it evolved. We also address challenges for those planning CWDM networks.
To adequately discuss CWDM we need to digress to talk about dense wavelength division multiplexing (DWDM). DWDM evolved first in the 1990s to resolve bandwidth requirements for long distance installations with the first priority to address the need to maximize the spectral efficiency of low fiber count oceanic fiber optic cable spans. This coincided with the invention and integration of the Erbium Doped Fiber Amplifier (EDFA) which could amplify the entire C-band (conventional band) operating from 1530 nm to 1565 nm. By using distributed feedback (DFB) lasers, 40 or more wavelengths could be transmitted over a single fiber. The combination of low loss single-mode fiber (SMF) using the 1550 nm window and the new ITU-T G.654 cut off shifted SMF and ITU-T G.655 Non-zero dispersion shifted fibers (NZDS); this provided a perfect marriage and evolved for use with long haul and metropolitan area networks. Standardized by the ITU’s G.692 standard, it quickly became dominant for use in long distance high-speed applications.
CWDM is used where a lower-cost solution than DWDM is required. This can include metropolitan area networks where transmission distances are shorter and amplifiers are not needed. It is also a great solution where fiber exhaust has limited the amount of fiber capacity. CWDM is approximately one third the cost of DWDM, and can easily be applied to installed standard single-mode fiber installations which are optimized for operation in the original band (O-band) operating between 1260 nm and 1360 nm.
Standardized in 2003 by the ITU-T G.694 and G.695 standards, CWDM addressed wavelengths operating as short as 1271 nm while operating with standard ITU-T G.652 single-mode fibers. With each of the 18 channels being 20 nm wide, lower system costs could be obtained with the use of uncooled Fabry-Perot (FP) lasers and filters.
In comparison DWDM systems require tight channel spacing of 50 GHz spacing (.4 nm), 100 GHz spacing (.8 nm), 200 GHz spacing (1.6 nm) which require higher cost temperature cooled and stabilized distributed feedback (DFB) lasers and wavelength lockers to control spectral drift. In addition, precision optical filters are required for demultiplexing the tightly spaced wavelengths.
With CWDM and DWDM transponders now available in modules, system manufacturers have created migration options where a single 20 nanometer CWDM module can be replaced in the future with a higher density 40 DWDM channel module creating more capacity when needed. The difference in fiber capacity between these two spacings is dramatic. Forty 100-gigahertz DWDM channels easily fit into the 1530 to 1565 nanometer C-band, while two CWDM channels will not quite fit into the same space. This wide channel spacing reduces the cost of multiplexing and demultiplexing optics, and it also allows the use of lower cost active and passive devices. It also allows a future migration to tighter spaced DWDM channels. (See Figure 1.)
CWDM wavelength spacing is specified in wavelength (nanometers) units while DWDM spacing is specified in frequency (gigahertz) units. This means that DWDM channel spacing is uniform in frequency across its operating range, but not uniform in wavelength while CWDM channel spacing is uniform in wavelength but not in frequency units.
Because DWDM devices started with very high reliability and tight specifications, it was easy to simplify these products and develop systems at lower costs. There have been 2 examples where this has been extremely successful: these are the fiber-to-the-home passive optical networks, and the development of CWDM products from 1271 nanometers to 1611 nanometers covering the O-band through the Long-(L) band.
CWDM can be used anywhere where increased capacity is required. In cases where there is a limited amount of available fibers, a CWDM system can be installed to ease fiber exhaust limitations. Simple 1 RU modules for multiplexing, demultiplexing and monitoring make this option easy to integrate into existing networks. This applies to many legacy G.652 fiber installations where today not enough fibers are available in the distribution network such as in FTTx networks.
Another application used in intelligent transportation systems (ITS) is the transmission of multi-channel video and bidirectional data from roadside cabinets. It isn’t uncommon to have as many as 8 cameras monitor the traffic moving in all directions through an intersection. The 8 video signals plus bidirectional signals for camera and access control where CWDM can be transported over 1 single-mode fiber. At the traffic management center (TMC) the optical signals are demultiplexed.
Optical fibers have been developed that are optimized for the various bands. A major difference between CWDM and DWDM is that DWDM is optimized for use at 1550 nm. The G.655 non-zero dispersion shifted fiber is optimized for use in the C-band at 1550 nm but can extend into in the short and long bands, while the G.652 is optimized for the original or O‑band operating at 1310 nm.
A variation of the G.652 fiber is the ITU G.652D, which is designed to operate in the extended or E-band between 1360 and 1460 nanometers where the historical water peak (OH) existed. The G.652D specified fibers in this spectrum with zero water-peak or low water-peak attenuation values. This is critical to recognize as in the E-band attenuation of the G.652 SMF can have attenuation as high as 2 dB/km as opposed to the standard .4 dB/km in the O-band. As the G.652D fiber was first recognized in March 2003, the time frame for installation of the new fiber occurred afterwards.
Here is the dilemma for service providers: did you document when and where the new G.652D fiber was installed in your outside plant? Wavelengths from 1371-1451 may not be usable due to high fiber loss with legacy G.652 fibers with 1391 nm being the highest.
If you do not know where the G.652D fiber is installed then you can design around the E-band by using other channels outside of the E-band that are not affected by the OH attenuation peak.
As optical transmission systems become faster and encompass more wavelengths as in the case of WDM, outside plant testing becomes more complicated as well. We now need to isolate measurements to a specific wavelength, requiring the use of WDM power meters, which are designed to lock onto the wavelength in question. Other equipment can include a WDM channel checker and a WDM OTDR which are available for either CWDM or DWDM applications.
Unfortunately testing individual channels is beyond the scope of most OTDRs which are designed for operation at standard 1310 nm and 1550 nm windows. To test CWDM channels at specific wavelengths we could use the output port of the transmitter as a light source. However, we would still need to order a WDM channel checker which is a specialized optical power meter that makes sequential power measurements for each wavelength. The technician should also select individual wavelengths to see and record the actual values for documentation purposes.
Another challenge is using CWDM in a feeder ring topology and then dropping that wavelength to a distribution fiber. Let’s say you are transmitting the 1311, 1331, 1351, 1371, 1391, and 1411 channels but are dropping the 1351 nm channel from the ring. To test the dropped wavelength without affecting adjacent wavelengths.
CWDM offers a great alternative when additional fibers or bandwidth is required while also providing a future migration path to DWDM. At a minimum CWDM testing requires an optical power meter calibrated for CWDM channels. CWDM is an excellent option to address fiber exhaust and should be investigated for future design options.