General description
A good place to start the discussion of fiber optics is referring back to the transmitter-channel-receiver model. The fiber optic transmitter uses a light source to modulate an incoming voltage signal into light pulses. These are then guided by reflection through a fiber optic cable (the channel) to the receiving end. The receiver uses a light detector to transform the incoming light pulses into a voltage signal from which the original data can be restored. The following sections will focus the fiber optic channel.
Physical description
A fiber optic cable is made up by three parts the core, the cladding and the jacket. The core is a thin cylinder shaped fiber of glass or plastic, usually 8-100um in diameter (for reference, the human hair is 50um-150um in diameter). In an ideal cable, incoming light will be confined to the core throughout its path to the receiver. The cladding is a plastic tube that surrounds the core, and is also usually made of plastic or glass. The material chosen for the cladding has to have lower refractive index than the core (see optics review below) to keep the light confined to the core. The jacket is a simply the wrapper for the cladding, it's usually made of plastic and is sometimes reinforced with Kevlar to make it more sturdy. Its purpose is to provide protection for the cladding and core against such hazards as abrasion and moisture.
Optics review
Before going into further detail about the propagation of light in the cable, a brief optics review is in order. A substance's refractive index is the ratio between the speed of light in vacuum and the speed of light in that substance. When a ray of light travels through a uniform substance A, it advances in a straight line. When reaching the borderline to a different substance B, part of the ray will be refracted into B, and the other part will be reflected back into A. The angle measured between the light ray and the line perpendicular to the substances' borderline is called the incidence angle. If this angle is bigger then a named critical angle (determined by the substances' refraction indices ratio) the entire ray is reflected back into substance A (no part of it refracts to B). This is called total reflection. If B's refractive index is very small compared to A's, their critical angle will be very small, meaning total reflection will occur only for a small range of rays (incidence angles). When total reflection occurs, the reflected ray's power is identical to that of the original ray. In cases where part of the ray is refracted however, the reflected part of the ray holds only a negligible part of the original ray's power while the refracted part holds most of it.
Signal propagation
When discussing propagation of light in the fiber optic cable we distinguish between single-mode and multimode propagation. Multimode propagation (multi-path) is the name given in cases where more than a single path of propagation exists in the cable. Rays entering the cable at different angles propagate in different paths. These different paths may differ in lengths. There are two types of multimode cables step-index and graded-index. In step-index cables the core and the cladding each have a different refractive index that's uniform across the entire length and depth of the cable. In this type of cable the light beams move in straight lines. Some of them are refracted out of the core while others are totally reflected throughout the cables in a zigzag shaped path. All of the beams travel in the same velocity (uniform refraction index), but since some paths are longer than other, the beams arrive at the receiver at different times. This causes a distortion in the transmitted light pulses which will be discussed in a later section. The name step-index describes the steep change in refraction index that occurs when crossing from the core to the cladding (a change resembles the shape of a step).
Graded-index describes a cable where the refractive index gradually changes across the rings that make up the core. Inner rings have lower refractive index while outer rings have a higher index. The cladding refraction index remains uniform. This results in the light beams moving in helical paths rather than zigzag shaped ones.
Single-mode propagation describes cables where only a single propagation path exists in the core. All beams entering the core traverse the same path which is the path that crosses the core axis. The number of different paths (modes) that propagate through the core depends on a number of factors - the core's refractive index (n1), the cladding's refractive index (n2), the core diameter (r), and the light source wavelength (
). Specifically, the number of modes is proportional in n1/n2 and also in /r. To achieve single-mode propagation the core's refractive index needs to be reduced to a value close to that of the cladding, and the core diameter needs to be reduced to the order of the light source's wavelength. This change in refractive index will lead to a critical angle that nears 90° most light rays will be refracted and only the ones that are pointed in axial direction will reflect and propagate through.
Attenuation and bandwidth
In this section we will focus the discussion on silica glass which is a popular material choice for fiber optic cores. We will also focus on the 0.5-1.6um wavelength range as the manufacturing of low loss cables and efficient sources/detectors is more convenient at this range. The silica is mixture made by fusing silicon and dioxide (SiO2). The molecular locations in the silica are random and inconsistent in contrast with structures such as the crystal which is characterized by fixed reparative molecular patterns. To achieve different refractive indices the silica is doped with other materials such as titanium, thallium, germanium, boron and others. Losses/attenuation in the glass fiber occur due to absorption, scattering and geometric effects. The following paragraphs will describe each of these in further detail.
Absorption can be divided into two kinds intrinsic absorption and absorption due to impurities. Every glass, regardless of its exact composition or level of purity, absorbs light at certain wavelength regions this is called intrinsic absorption. Glass absorbs heavily in the ultraviolet region of the spectrum (0.01-0.4um). The intrinsic absorption loss diminishes towards the visible part of the spectrum (0.4-0.75um), and then rises again and peaks in the infrared (0.75um-1mm). intrinsic abruption therefore bounds our stated region of interest on both sides but is insignificant inside it. Of the impurity related absorptions, the most significant two are transition metal ions and transition hydroxilion (OH) ions. Special measures are taken to guarantee low levels of these impurities during the manufacturing process. To achieve acceptable loss properties metal impurities need to be kept under few parts per billion, and OH impieties under few parts per million. Even when kept at this level OH impurities still cause a significant loss in the region around 1.4um wavelength. This region has to be avoided to assure reliable communication.
Scattering losses have to do with the structure of the glass. As mentioned earlier the silica has an erratic structure, this results in localized refraction index changes. These are modeled as small scattering objects (much smaller than the discussed wavelengths) embedded in a homogenous material. A light beam passing through in this model, is scattered upon bumping into these objects and loses some of its energy. This is known as Raleigh scattering and it becomes significant only in wavelengths below 0.8um. This restricts the use of small wavelengths.
Geometric effect attenuation refers to the loss caused by cable bending either macroscopic or microscopic. Macroscopic bending means large radius bending, such as bending a cable around a duct corner. A beam propagating through the cable in total reflection will necessarily hit the bend with diminished incidence angle. This will possibly lead to an angle lower than the critical angle in which case the beam will refract out of the cable causing a energy loss. Microscopic bending refers to small axial changes that occur along the cable after it's been sheathed in its jacket. This may cause destructive coupling between otherwise unrelated modes (paths) in the cable which results in energy loss.
The discussion above focused on losses that occur inside the cable, leaving out losses that occur on cable connectors and splices. These too should be taken into consideration.
Taking the total of the above losses (all except bending which has to be addressed specifically) we get the diagram shown below.
The described attenuations dictate three practical wavelength windows that can be used [0.8,0.9um], [1.3,1.35um], and [1.45,1.6um]. The last two windows have very low attenuation 0.2dB/km (roughly 4% loss). The first window has higher attenuation but the manufacturing of light sources intended for this window are very convenient. This means repeaters can be spaced ten of kilometers apart from each other (compared to few kilometers apart in copper media). This is significant in lowering the cost of optic fiber systems and also in reducing errors introduced in repeaters. Each of the three windows has a bandwidths of 25-30THz several orders of magnitude larger than can be found in radio or copper media. This allows for Gbps data rates.
Distortion and noise
Light pulses are distorted in the fiber optic channel by material dispersion, waveguide dispersion and by multimode spreading. These distortions types do not equally apply to all previously described cable types. Both types of dispersions mentioned here are caused by variations in refractive index across wavelength. This means different wavelength beams travel in different velocities in the fiber. A practical light source emits a light pulse that's made up of different wavelength pulses. In the presence of dispersion these will reach the end of the fiber at different times and when added together, the resulting pulse at the output will appear spread out (in comparison to the source pulse that is). Dispersion caused by natural properties of the core material is labeled material dispersion. Waveguide dispersion doesn't have to do with material (it can occur even in non dispersive materials), but with the structure. Since the light inside the optic cable structure is constrained to a very specific zigzag motion, the principles of unbound light refraction no longer apply. Refractive index is wavelength dependant in this case (again regardless of dispersive properties of the material).
Multimode spreading is similar to the previously described dispersion phenomena. The parts that make up the pulse propagate in different modes (paths) in the fiber, and therefore arrive at the output at different times. This again causes the pulse to spread in time.
In step-index cables all three distortion causes apply. The total spreading in such cables is typically tens of ns/km. The contribution of multimode spreading to the total spreading is several orders of magnitude larger than that of material and waveguide dispersion. Graded-index cables are also affected by all three distortion causes, but pulse spreading is much smaller than it is in step-index, the reason for this is explained next. In graded-index, paths that are contained in the inner core rings are shorter, but also slower (high refractive index). On the other hand paths that extend into the outer rings are longer, but faster (low refractive index). The length-velocity trade off creates a situation where all paths take approximately the same time to traverse. This leads to reduced multimode spreading distortion. Total spreading in graded-index is typically few ns/km. Singe-mode cables, as can be expected, suffer only from material and waveguide dispersion. The dominant distortion in the single-mode case is the material dispersion.
It's important to state that spreading is only linear with the fiber length in case of short length cables (few kilometers). For long cables, spreading is linear with the square root of the cable length.
It's been found that pulses of a special shape called solitons are almost unaffected by dispersion, this is of course very desirable and efforts are made to adopt these into next generation systems.
Regarding noise, besides thermal noise which is present in all systems, a fiber optic system also suffers from shot noise and modal noise. We'll start with a description of shot noise. The detectors used in optic receivers follow the following principle incoming optic signals generate discrete charges that together make up the current carried into the receiver. Since these charges are generated at random times, the total current is not constant, it fluctuates around a constant average. These fluctuations are referred to as shot noise. A second type of noise suffered by optic systems is modal noise. In multimode cables, the different modes interfere with one another, some modes add together, while others subtract. This creates a certain light pattern at the fiber end. The light pattern changes with variations in source light wavelength. Such variations may occur due to temperature variations or vibrations. Ideally losses impose by fiber link components (such as connectors) are independent of this light pattern, but in practice this is not the case. Continuous random temperature shifts or vibrations will therefore cause random variations in loss and consequently variations in received power. These variations are referred to as modal noise.
Optic systems are not affected by external electromagnetic fields, so they do not suffer from impulse noise or crosstalk. They also do not radiate any fields that might cause interference with neighboring systems.
Applications
Television Modern cable TV networks are hybrid fiber-coax networks where the backbone consists of fiber optic cables and only the last stretch leading into consumer homes is still coax.
Telephone Fiber optic cables play a major role on a number of territories in the telephone network. It is used on long distance telephone routes which span 1500 km on average and carry few tens of thousands of voice channels. It's also used for rural exchange trunks. These connect between telephone exchanges in neighboring towns. They can range up to 160 km and need to carry up to 5000 voice channels. Yet another territory where fiber optic cables are employed is intra metropolitan trunks. These are 12 km on average and need to carry as many as 100,000 voice channels. These trunks do not include repeaters. In the long distance routes and rural trunks mentioned, optic fiber competes against microwave transmission which is also widely used. Fiber hasn't yet phased out twisted pair in the routes that connect the home consumer to the nearest exchange but as the telephone network rapidly evolves into a platform for audio data and video this is ultimately expected to happen as well.
LAN A number of different physical layer standards for LANS are based on fiber optics. The 10BASE-FP Ethernet standard specifies use of a 850 nm light source, allowing a 10Mbps rate with repeaters needed every 2km. The fiber distributed data interface (FDDI) standard specifies a 1300 nm light source and achieves a rate of 100Mpbs over 2 km. Going back to Ethernet standards, the 100BASE-FX also achieves a 100 Mbps rate with physical specifications very similar to those of FDDI. Unlike FDDI 100BASE-FX specifies the use of two fibers (one for sending and one for receiving) and requires repeaters every 100 meters. Fiber is also a very popular medium in the gigabit Ethernet domain. The 1000BASE-SX standard delivers a 1Gbps rate using a 850 nm light source with a multimode fiber and requires a repeater every 550 meter. A second gigabit Ethernet standard is the 1000BASE-LX which specifies that a 1300 nm light source be used with either a multimode or a single-mode fiber. The multimode variant requires a repeater every 550 meter, and the single-mode one requires one every 5 km.