Computer and Communication Networks (paperback)
14.2. Basic Optical Networking Devices
An optical network comprises optical devices interconnected by optical transmission links. Other basic elements on an optical networks are: tunable lasers, optical buffers or delay elements, optical amplifiers , optical filters, wavelength-division multiplexers, and optical switches. 14.2.1. Tunable Lasers
Tunable lasers can continuously change their emission wavelengths , or colors, in a given spectral range. These changes work in collaboration with optical switches to select a particular wavelength for connection establishment. A tunable dispersion compensator is used to compensate for fiber-dispersion losses over the length of the fiber. 14.2.2. Optical Buffers or Delay Elements
Buffering in optical nodes also faces serious limits forced by optical-technology shortcomings. Optical buffers , or optical delay elements , can be implemented by using a certain length of fiber to delay signals. No practical optical memory is available with the current technology. 14.2.3. Optical Amplifiers
Often in non-all-optical networks, a signal may not be able to remain in optical form and may have to be regenerated, requiring the amplification of the signal or the conversion of the signal from optical to electronic. The use of optical amplifiers allows achieving large-distance communication without the need of a regenerator. A major milestone in the evolution of optical fiber transmission systems was Erbium-doped fiber amplifiers (EDFAS), or simply optical amplifiers. An optical amplifier can amplify signals at many wavelengths simultaneously . The amplification of a signal involves the extraction of the clock from the signal and reclocking the signal, resulting in the creation of "speed bumps" in a network. 14.2.4. Optical Filters
Signal filtering is often needed in optical networks. An optical filter equalizes the gain of transmission systems and filters the noise or any unwanted wavelength. In Figure 14.2 (a), an optical cable carries four wavelengths, » 1 , » 2 , » 3 , and » 4 , and is connected to a filter. This particular filter is designed to allow only » 1 to pass and to filter » 2 , » 3 , and » 4 . Figure 14.2. Basic communication devices for optical networks: (a) optical filter; (b) wavelength-division multiplexer; (c) optical switch
The design of an optical filter has a number of challenging factors. Insertion loss is one and is the loss of power at a filter. A low insertion loss is one of the specifications for a good optical filter. The state of polarization of the input signals should not affect the loss. Keeping the temperature at the desired degree in optical systems, especially on filters, is another factor. Temperature variation should not affect the passband of a filter. As a result, wavelength spacing between adjacent channels on transmission systems should be large enough so that the wavelength shift does not affect the entire operation. The passband of filters in an optical system can be narrower if a number of filters are cascaded. The objective is that at the end of cascading the broad passband , each filter causes a small change in operating wavelengths. 14.2.5. Wavelength-Division Multiplexer (WDM)
Wavelength-division multiplexing (WDM) transmission systems divide the optical-fiber bandwidth into many nonoverlapping optical wavelengths: so-called WDM channels. As shown in Figure 14.2 (b), a WDM mixes all incoming signals with different wavelengths and sends them to a common output port. A demultiplexer does the opposite operation, separating the wavelengths and dispatching them onto output ports. On the common link, each channel carries information at light speed with minimal loss. This multiplexing mechanism provides much higher available transmission capacity in communication networks. 14.2.6. Optical Switches
An optical switch is the heart of an optical network. The objective in using optical switches rather than semiconductor switches is to increase the speed and volume of traffic switching in a core node of a computer communication network. The optical switch acts as an optical cross-connect (OXC) and can accept various wavelengths on network input ports and route them to appropriate output ports. The optical switch performs three main functions:
Figure 14.2 (c) shows a simple 2 x 2 optical switch. A signal with wavelength i arriving at input j of the switch denoted by » i,j can be switched on any of the four output ports. In this figure, » 1,1 and » 2,1 arrive on the first input port of the switch; » 1,1 can be switched on output port 2, and » 2,1 can be forwarded on output port 1. This basic optical switch is a switch element in larger-scale switch architectures. The basic switch can be made using various technologies. Such switch elements are broadly classified as either non-electro-optical or electro-optical . Classification of Switch Elements
Non-electro-optical switches have simple structures. For example, a mechanical optical switch uses mirrors at a switch's input and output ports. A switch can be controlled by moving the mirrors and directing a light beam to a desired direction, and eventually to the desired output port. The advantages of mechanical switches are low insertion, low cross talk, and low cost. However, they have low speed. Another example is the thermo- optic switch , which is built on a waveguide. In this type of switch, a change of temperature in the thermo-optic structure of the waveguide can change the refractive index of the material, thereby allowing a switching function to be formed . Although similar to mechanical switches, thermo-optical switches operate at low speeds, but their cross talk is lower. Figure 14.3 shows a typical 2 x 2 electro-optic switch , which uses a directional coupler . A 2 x 2 directional coupler is an integrated waveguide that can combine or split signals at its output ports. A coupler is the building block of filters, multiplexers, and switches and is known as a star coupler . A star coupler operates by changing the refractive index of the material placed in the coupling region. The switching function can be achieved by applying an appropriate voltage across the two electrodes. Figure 14.3. Smallest optical switch constructed with directional coupler: (a) architecture; (b) symbol
The switching function is achieved by taking a fraction, , of the power from input port 1 and placing it on output port 1, with the remaining fraction 1 - of the power on output port 2. Similarly, a fraction 1 - of the power from input port 2 is placed on output port 1, with the remaining power fraction on output port 2. The advantages of this switch are its speed and a modest level of integration to larger modules. However, the high cost of its production is a disadvantage of this switch over the other switch types. Contention Resolution
Designing large-scale switching devices is one of the main challenges in optical networks. Ideally all the functions inside an optical node should be performed in the optical domain. However, packet contention in switches can be processed only in electronic devices. Several technological factors bring restrictions to the design of an optical switch. The expansion of a fixed- size switch to higher scales is also a noticeable challenge. Electronic switches have much better flexibility of integration than do optical devices. Some performance factors other than switching speed should be considered in order to prove the suitability of a switch for optical networks. As do regular switches, optical switches face the challenge of contention resolution within their structures. Contention resolution is of the following three types:
Another important factor in optical switches is insertion loss , which is the fraction of power lost in the forward direction. This factor should be kept as small as possible in optical switches. Sometimes, the dynamic range of the signals must be increased in switches just because the level of loss may not be acceptable for the corresponding particular connection. The factor of cross talk in switches must be minimized. Cross talk is the ratio of output power from a desired input to the output power from all other inputs. At issue is the amount of energy passing through from adjacent channels. This energy can corrupt system performance. The cost of optical devices, especially WDMs, is high compared to that of electronic modules. The production of all-fiber devices can also be a way to reduce the cost. |