Tag: DWDM

It has been almost 20 years since DWDM came on the scene with Ciena’s introduction of a 16 channel system in March of 1996, and in the last two decades it has revolutionized the transmission of information over long distances.  DWDM is so ubiquitous that we often forget that there was a time when it did not exist and when accessing information from the other side of the globe was expensive and slow.  Now we think nothing of downloading a movie or placing an IP call across oceans and continents.  Current systems typically have 96 channels per optical fiber, each of which can run at 100Gbps, compared to the 2.5Gbps per channel in the initial systems.  All of this got me thinking about how it often takes two innovations coupled together to make a revolution.  Personal computers did not revolutionize office life until they were coupled with laser printers.  Similarly, the benefits of DWDM were enormous because of erbium doped fiber amplifiers (EDFAs).

DWDM stands for Dense Wavelength Division Multiplexing, which is a complex way of saying that, since photons do not interact with one another (at least not much) different signals on different wavelengths of light can be combined onto a single fiber, transmitted to the other end, separated and detected independently, thus increasing the carrying capacity of the fiber by the number of channels present.  In fact non-Dense, plain old WDM, had been in use for some time with 2, 3 or 4 channels in specialized circumstances.  There was nothing particularly difficult about building a basic DWDM system.  The technology initially used to combine and separate the wavelengths was thin film interference filters which had been developed to a high degree in the 19th Century.  (Now a ’days photonic integrated circuits called Arrayed Waveguide Gratings, or AWGs are used to perform this function.)  But until the advent of EDFAs there was not much benefit to be had from DWDM.

Fiber optic data transmission began in the 1970s with the discovery that certain glasses had very low optical loss in the near infrared spectral region, and that these glasses could be formed into fibers which would guide the light from one end to the other, keeping it confined and delivering it intact, although reduced by loss and dispersion.  With much development of fibers, lasers and detectors, systems were built which could transmit optical information for 80km before it was necessary to “regenerate” the signal.  Regeneration involved detecting the light, using an electronic digital circuit to reconstruct the information and then retransmitting it on another laser.  80km was much farther than the current “line of sight” microwave transmission systems could go, and fiber optic transmission was adopted on a wide scale.  Although 80 km was a significant improvement, it still meant a lot of regeneration circuits would be needed between LA and New York.  With one regeneration circuit needed per channel every 80 km, regeneration became the limiting factor in optical transmission and DWDM was not very practicable.  The then expensive filters would have to be used every 80 km to separate the light for each channel before regeneration and to recombine the channels after regeneration.

Since full regeneration was expensive, researchers began to look for other ways to extend the reach of an optical fiber transmission system.  In the late 1980s Erbuim Doped Fiber Amplifers (EDFAs) came on the scene.  EDFAs consisted of optical fiber doped with Erbium atoms which, when pumped with a laser of a different wavelength, created a gain medium which would amplify light in a band near the 1550nm wavelength.  EDFAs allowed amplification of the optical signals in fibers which could counter the effects of optical loss, but could not correct for the effects of dispersion and other impairments.  As a matter of fact, EDFAs generate amplified spontaneous emission (ASE) noise and could cause fiber nonlinearity distortions over a long transmission distance.  So EDFAs did not eliminate the need for regeneration completely, but allowed the signals to go many 80 km hops before regeneration was needed.  Since EDFAs were cheaper than full regeneration, systems were quickly designed which used 1550nm lasers instead of the then prevailing 1300nm.

Then came the “ah ha” moment.  Since EDFAs just replicated the photons coming in and sent out more photons of the same wavelength, two or more channels could be amplified in the same EDFA without crosstalk.  With DWDM one EDFA could amplify all of the channels in a fiber at once, provided they fit within the region of EDFA gain.  DWDM then allowed the multiple use of not only the fiber but also the amplifiers.  Instead of one regeneration circuit for every channel, there was now one EDFA for each fiber.  A single fiber and a chain of one amplifier every 40~100 km could support 96 different data streams. Regenerators are still needed today, every 1,200~3,500km, when the accumulated EDFA ASE noise exceeds a threshold that a digital signal processor and error correction codec can handle.

Of course, since the gain region of the EDFA was limited to about 40 nm of spectra width, great emphasis was placed on fitting the different optical wavelengths as close together as possible.  Current systems place channels 50GHz, or approximately 0.4 nm, apart, and hero experiments have done much more.

In parallel, new technologies have increased the bandwidth per channel to 100 Gbps using coherent techniques that we have discussed in other blog posts.  So a single fiber that in the early 1990s would have carried 2.5Gbps of information, now can carry almost 10 Terabits/sec of information, and we can watch movies from the other side of the globe.

CWDM, just as DWDM, use multiple light wavelengths to transmit signals over a single optical fiber. However, there are still some differences betwwen the two techologies in many ways.

CWDM uses a 20-nm wavelength spacing that is much wider than the 0.4 nm for DWDM. The wider wavelength spacing in CWDM means lower product development costs. This is one reason why CWDM is less costly than DWDM.

Most CWDM devices operate in the 1470-nm to 1610-nm range. The frequency grid for DWDM and the wavelength grid for CWDM systems are defined by the international telecommunication union (ITU) standard G.694.1 and G.694.2, respectively.

CWDM provides a maximum of 8 lambadas between two CWDM multiplexers over a single fiber pair as compared to DWDM Multiplexers, which support up to 32 lambdas (based on 0.8-nm or 100-GHz wavelength spacing) over a single fiber pair. some long-haul DWDM systems can support up to 160 lambdas per fiber pair.

Each CWDM channels uses a specialized gigabit interface converter (GBIC) or small form-factor pluggable (SFP) transceivers are commonly known as colored GBIC and SFP. Each CWDM channels uses a different “color” GBIC or SFP because each lambda represents a different color in the spectrum. In this case, the native GBIC or SFP in the client devices are substituted with a colored GBIC or SFP.

CWDM multiplexers are usually passive (i.e, not powered) devices containing a very accurate prism to multiplex eight separate wavelengths of light along a single fiber pair. And passive CWDM devices cannot generate or repeat optical signals.

No amplification is possible with CWDM because CWDM uses wavelengths that cannot be amplified with EDFA amplifiers. Therefore, the maximum distance for a CWDM link is approximately 100 km.

The Cisco ONS 15501 EDFA, which has a wavelength range of 1530 nm to 1563 nm, can only amplify two signals (1530 nm and 1550 nm) out of the eight signals that are multiplexed onto the fiber pair.

CWDM provides an alternative solution to DWDM for low-latency and high-bandwidth requirements associated with synchronous replication application. However, DWDM is more scalable than CWDM. DWDM also has a longer distance capacity than CWDM because DWDM can be amplified. The main benefit of CWDM is its low cost. It is a cheaper solution than DWDM. In other words, CWDM is optimized for cost, while DWDM is optimized for bandwidth. For enterprises that have access to dark fiber and have only limited scalability requirement, CWDM is a relatively inexpensive way to achieve low-latency and high-bandwidth interconnections between DCs. The CWDM implementation also results in less complex installation, configuration, and operation as compared to DWDM.

CWDM can be deployed in point-to-point, linear, or fiber protected ring topologies, It is limited to a distance of up to 120 km for Gigabit Ethernet and 100 km for 2-G FC in a point -to-point topology. It is typically used only for extension of the FC fabric in a metro or campus application. As CWDM carries only eight lambdas on a single fiber pair, there are limits to the number of possible drops and the number of sites that can be interconnected. A ring or linear topology reduces the distance depending on the number of OADMs traversed by the CWDM channels because each CWDM OADM introduces additional power loss in the network.

CWDM can also be used to enable multiple ISL connections between the switches over a single fiber since it requires less fiber for interconnecting two metro sites. The same benefit applies to port channel implementation between the switches.

In short, DWDM is a solution that provides a higher number of connections and longer reach, or extension, at a much higher cost while CWDM is a more cost-effective solution for metro or campus solutions where the distance is limited.