WDM basics

This primer contains information about fiber optics communications and technologies used such as Wavelength-division Multiplexing (WDM) and its subtypes of Coarse WDM or Dense WDM (DWDM). These technologies allow for bidirectional communication as well as capacity multiplication and are widely used at the present to connect routers in different cities and on different continents, constituting the physical part of the backbone of the internet.

The primer is intended for people meaning to learn about communications methods and technologies, so it will be less technically oriented, as more detailed technical information is easily found elsewhere. Instead, I have tried to explain how fiber-optics communications work, explain the basic principles that govern WDM, as well as give information about follow up resources. Little knowledge about these topics is assumed on the reader’s part.

For any information about WDM to be at least partly understood, a few words about the underlying transport medium - fiber optics - are necessary.

Fiber optics

Since the first time someone has signaled another party using torches in the night, mankind has used light to communicate over distance. Light travels very quickly, and in certain conditions it travels far as well, but it has limitations. The environment (air, usually) absorbs it, it must travel in straight lines and was not always easy to generate on command, as anyone who has ever tried to start a fire in the rain must surely know. It took a very long time for technology to advance to the point that light could be used to transmit information efficiently, over great distances and beyond line-of-sight and fiber-optics played a paramount role in it.

Made of high quality glass (silica) or plastic, optical fibers are conduits for light that transmit it from one end of the fiber to the other. Dealing with the old problems of light-based communication, they transmit light over longer distances with little loss, and can be run around corners if the loop is wide enough. Fiber optics cables can be used to connect stations farther away than electrical cables can, and are immune to electro-magnetic interference which, as anyone living in a densely populated area with many wireless access points should know, has become quite a problem. They can be used to illuminate or to carry images and work as sensors, and have lately become indispensable in the Information Age as carriers of information.

However, as a highly technological product, fiber optics cables are relatively expensive and difficult to produce which limits their usefulness. They are most often used inside the datacenter, where many short fiber optics cables are used to connect machines that handle large volumes of traffic between each other, the so-called “backbone” of the internet as we know it. (See the Internet primer for details). These machines need to transmit and receive data fast, and fiber-optics is the connection method of choice because of their high bandwidth.

Example: inside the home, your computers or router might be connected via Ethernet cables, which are made from copper wires, but their length must be less than 300 feet, and the maximum data throughput is about 1 Gbps (one billion bits per second) in higher-end WI-FI routers. Backbone routers on the other hand are connected by fiber-optics links that can transport many times that amount of data tens and hundreds of Gigabits per second, and all this over links that can be between 1 and 1000 miles long. Let’s see how that happens.

The string and cans

A basic communications setup entails two stations that want to talk to each other and the medium used for transmission of information, which in this case is a pair of optical fibers. A transmitter (usually a special laser ) and a receiver that can detect light are used in both ends, and each station transmits on one fiber and “listens” on the other. Turn the light on to send “1” and turn it off to send “0” and you have a basic communications channel.

However, this basic channel, while useful in a didactic role, doesn’t do much for high speed transmission. One can “blink” the light faster and faster, but there are hardware-imposed limits when you reach gigabit speeds - the transistors that drive the lasers just can’t switch the light on and off fast enough. While faster lasers can and have been made, there is a better, more efficient solution - when more bandwidth is required just run more fibers between the stations.

This works if the stations are feet apart, or maybe in the same building, but it becomes expensive quickly if the fibers need to be run between cities and downright impossible if the fiber-optics cable connects continents. (Yes, there are undersea cables that connect land stations on all the continents except Antarctica. You can imagine the engineering and technological feats necessary to lay cable down the bottom of the ocean and laying another pair of fibers whenever more bandwidth is required is not an option.) So a different solution was required.

If multiple fiber pairs were not an option, the next idea was using different “colored” lasers, though I’m using that term metaphorically since human eyes can’t detect the light these lasers emit. Since the most used type of optical fiber transmits some wavelengths a lot better than others, the lasers were constructed to send specifically on those wavelengths (to have a certain “color”), which fall in the infrared portion of the spectrum - around the 1550 nanometer mark. This ensures the same amount of light goes farther and remains brighter, though unfortunately falls short of our visible spectrum and is thus invisible to the naked eye (and very dangerous!)

Lasers also emit coherent light, which means they only emit one color - they have a narrow spectrum. This makes them useful since the colors can be more easily told apart.

It has been known for a long time that white light is composed of different colors, and it can be split into its parts using a prism (like in the Pink Floyd album cover.) This effect, however, applies to all light, not just the kind we can see! So using this effect on the lasers that transmit the data was a natural step forward. Helped by the fact that lasers are sources of coherent light, it meant a lot of them could be crammed into a relatively small portion of the spectrum without there being much overlap - this is important for reasons outlined below.

That is the setup currently used in WDM. Multiple lasers, used at the same time but on different wavelengths can be combined, sent through the fiber optics “pipe” and dispersed at the destination into the original “colors”, recreating the original transmission streams. This was a huge breakthrough for the telecommunications world, as it enabled more and more data to be sent on the limited number of fiber-optics pairs available.

Thus, one pair of fibers can be used to connect a pair of backbone routers in different cities, and 1 Gbps or 10 Gbps lasers can be used, depending on the requirements. Should the need for bandwidth grow, more lasers can be added to each router, and using special optical equipment the separate streams can be combined into one, sent down the fiber and separated on arrival. However, the routers only see one huge link, not separate streams, which makes the setup flexible, as lasers can be added whenever an increase of capacity is needed.

The WDM concept has been around since the ‘80, but lately has caught traction with the telecommunications companies as time passed, since better understanding of the intricacies of light propagation in fiber optics as well as technological advances in miniaturization has made the necessary equipment less expensive. Also, WDM allows the expansion of capacity of any given network without laying more fiber. Where there was just one 10 Gbps link between two cities before, nowadays the same fiber pair can be used to transmit upwards of 400 Gbps and even more, as the capacity of any given link can be expanded just by adding multiplexers and demultiplexers at each end.

Coarse WDM

CWDM, or Coarse/Conventional WDM differs from Dense WDM in number of signals. Up to 8 separate light sources, on different frequencies are combined and used on a single fiber pair. The relatively low number of channels means the frequencies of the emitting lasers can be further apart, making generation and detection cheaper. At the lowest level, CWDM means just two signals that travel on the same fiber - one around the 1310 nm wavelength, and one around 1550 nm. This way, the signals interfere very little with each other and just one fiber strand can be used for bi-directional transmission (this is important in congested city fiber networks, where using a pair of fibers for every connection has become inefficient.) Since the channels are very wide apart in the spectrum (by current standards, as we’ll see), the transceivers (emitter/receiver laser combo) can be made cheaply. This type of CWDM is widely used in metropolitan areas for cable television networks, to have upstream and downstream channels available, and lately to provide fiber to the home in the (fortunate) areas in which this service is available.

Dense WDM (DWDM)

Dense Wavelength Division Multiplexing takes the thinking behind CWDM further, by diminishing the spacing between the various laser wavelengths that are used. By using special equipment, up to 40 and even 80 different wavelengths can be used on the same fiber, though in this system the fiber strand is uni-directional and all the wavelengths fall around the 1550 nm mark. This happens because this way, erbium doped fiber amplifiers (EDFAs) can be used to further the reach of the transmitted signal, and EDFAs are most effective for wavelengths between 1525–1565 nm, in the so called C band of the spectrum. The spacing between individual channels must be very small, compared to the previous CWDM example and more sensitive (and more expensive) equipment must be used. The main advantage of EDFAs is that they amplify the whole spectrum of the signal that is fed into them as long as it falls into its amplification band. Thus any kind of signal can be sent, and any number of channels can be transmitted (up to the limits imposed by the equipment that must separate them, of course) and the amplification works regardless of the signal structure. This leverages the cost of the amplifier, making it more efficient the more channels are used on the fiber pair.

DWDM Systems

A basic system has to consist of the following: A transponder that gets data from a client, like a backbone router, in the form of an optical signal. The transponder then converts the wavelength of the signal to a predetermined, standardized wavelength in the C band, called “channel”. This is usually done by transforming the received pulses of light into electricity and then converting them to light again but on a different frequency. In a 40 channel system, the channels are numbered from channel 21 to channel 60. A multiplexer then adds the different wavelength optical signals from all the transponders into a single light stream. An amplifier boosts the signal and sends it down the line. At this stage, depending on application, the signal can be powerful enough that it could reflect off shiny surfaces or even white walls, so protective eyeware should be worn when handling plugged in fiber optics or amplification equipment. The fact that the beam of light is invisible does not offer protection, and permanent blindness or scarring of the retina is possible. The signal travels down the optical fiber, and is attenuated by the silica glass. Every 80-100 km there usually is an intermediate amplification site, where EDFAs are used to re-amplify the signal. Otherwise, the light would go dimmer and dimmer until the receiving equipment could no longer distinguish the pulses. If necessary, in such sites an intermediate optical terminal can be used, which drops and adds certain wavelengths from the stream, and can insert diagnostics and control signals into the stream. This ensures the setup is flexible, and point-to-point as well as full network length channels can coexist. At the remote site, the process is done in reverse. An amplifier is used again to boost the received signal, because the de-multiplexing process that follows introduces enough attenuation that combined with the attenuation from the previous fiber span can render the signal useless. A demultiplexer is used to separate the channels from the stream and send them to individual transponders. Transponders are used again to modify the wavelength of the individual streams, in order for the signals to be sent to the corresponding interfaces in the remote router. This basic setup can be extended for various configurations. Usually, intermediate sites have more than just amplification and also add and drop channels, connecting the sites to the bigger network; the latest improvements in technology have allowed for software-tunable transponder modules, as well as special equipment which can be configured via software - telling it which channels to add or drop, making the entry points into the system very flexible. However, the signal can only travel so much distance before impurities in the fiber distort it beyond recognition so regeneration of the signal was required. This usually happened in the intermediate sites where transponders were used to transform the weak optical signal into an electric signal, reform its shape and transform it back into a fully formed optical signal - dubbed OEO regeneration (optical-electrical-optical). Advances in technology as well as availability of software-on-a-chip error correction algorithms have eliminated the need for OEO regeneration except for the longest of links and at the moment, 1000 kilometer fully optical links that carry 40, 80 and up to 160 channels are possible.

This allows for great capacity multiplication between remote sites, as once the bare-bones DWDM system is in place, all 40 (or 80, or 160) channels are available at once and increasing the number of used links is just a matter of installing another pair of transponders.

Thus, DWDM systems are of great use to telecommunications networks, though the small spacing used between channels and the quality of equipment required to achieve this high performance mean the equipment is more expensive than CWDM. This has made the technology useful at the backbone layer of the usual telecommunications network, as the requirement of high bandwidth necessary over long distances has made the more expensive equipment easier to justify financially. However, CWDM still has its uses in metropolitan areas, where bandwidth is relatively low but other constraints - like number of available fiber strands or pairs, and resilience to fiber imperfection - dictate the layout of the network.

Comparison between CWDM and DWDM

Though more expensive and higher-tech than CWDM, DWDM actually came first. However, the need for a scalable, low-cost solution in certain areas made telecommunications companies drive to get some of the advantages of DWDM in easier to use packages, and for a lower price. Thus, CWDM appeared.

The differences between technologies can be summarized below:

CWDM

  • lower cost
  • low capacity (usually 1 or two channels, but up to 8 )
  • short-range
  • wavelengths used are far apart
  • frequencies for each channel can be wide
  • more resilient to fiber attenuation
  • easier to install (some solutions are small enough to carry around the office, AC power compatible)
  • amplifier use is possible, but usually more expensive than it’s worth, capacity wise.

DWDM

  • high cost
  • high capacity (usually 40 channels, but up to 160)
  • long-haul (1000 kilometer links are possible)
  • wavelengths used for each channel are tightly packed
  • frequencies used for each channel have to be narrow
  • less resilient to fiber attenuation
  • installation tends to be more involved, as equipment is usually sensitive to temperature variations, has to be sheltered properly
  • can benefit from amplifier boost

Conclusion

This all means that the two technologies don’t compete, as CWDM systems are usually well-suited for metropolitan use, to connect various offices between them or to the central backbone network, while DWDM is better put to use in the backbone network itself, to connect cities and countries. Both technologies are used at the same time in the same network, and while DWDM links are almost always amplified, CWDM systems are being promoted as low-cost access solutions and amplifiers are usually not cost-effective. DWDM equipment on the other hand needs the environment to be stable in order for the lasers to perform at their peak precision, so cooling and sheltering adds an important overhead to installation. Add that to the high quality needed from the (de) multiplexing equipment in order to add or separate each channel to or from the closely spaced bunch, without disturbing the other channels, and the basic solution becomes prohibitively expensive except for use at the backbone level.


QR Code
QR Code wdm_basics (generated for current page)