Ethernet LANs| Coaxial Cable | Twisted-Pair Cable | Advance Ethernet

Ethernet LANs

One of the oldest and by far the most widely used of all LANs is Ethernet. Ethernet, which was developed by Xerox Corporation at Palo Alto Research Center in the 1970s, was based on the Aloha wide-area satellite network implemented at the University of Hawaii in the late 1960s.

In 1980, Xerox joined with Digital Equipment Corporation (now part of Hewlett Packard) and Intel to sponsor a joint standard for Ethernet. The collaboration resulted in a definition that became the basis for the IEEE 802.3 standard. (The Institute of Electrical and Electronics Engineers [IEEE] establishes and maintains a wide range of electrical, electronic, and computing standards. The 802.X series relates to LANs.) Today, there are numerous variants of Ethernet. More than 95 percent of all LANs use some form of Ethernet. Furthermore, Ethernet has grown more capable over the years and is now routinely used in both MANs and WANs.

Topology


The original versions of Ethernet used a bus topology. Today, most use a physical star configuration (see Fig. 12-16). Each node uses a twisted-pair cable to connect to a centrally located hub or switch. The switch is then connected to a router or gateway that provides access to the services needed by each user. Information to be transmitted from one user to another can move in either direction on the bus, but only one node can transmit at any given time.

Encoding

Ethernet uses baseband data transmission methods. This means that the serial data to be transmitted is placed directly on the bus media. Before transmission, however, the binary data is encoded into a unique variation of the binary code known as the Manchester code (Fig. 12-17).

Fig. 12-17(a) shows normal serial binary data made up of standard dc binary pulses. This serial binary data, which is developed internally by the computer, is known as unipolar NRZ. This data is encoded by using the Manchester format by circuits on the NIC. Manchester coding can be alternating or direct current. In ac Manchester, shown in Fig. 12-17(b), the signal voltage is switched between a minus level and a plus level. Each bit, whether it is a binary 0 or binary 1, is transmitted as a combination of a positive pulse followed by a negative pulse or vice versa. For example, binary 1 is a positive pulse followed by a negative pulse. A binary 0 is a negative pulse followed by a positive pulse. During each bit interval, a positive-to-negative or negative-to-positive voltage transition takes place.

Ethernet LANs| Coaxial Cable | Twisted-Pair Cable | Advance Ethernet

This method of encoding prevents the dc voltage level on the transmission cable from building up to an unacceptable level. With standard binary pulses, an average dc voltage will appear on the cable, the value of which depends on the nature of the binary data on the cable, especially the number of binary 1s or 0s occurring in sequence. Large numbers of binary 1 bits will cause the average dc level to be very high. Long strings of binary 0s will cause the average voltage to below. The resulting binary signal rides up and down on the average dc level, which can cause transmission errors and signal interpretation problems. With Manchester encoding, the positive and negative switching for each bit interval causes the direct current to be averaged out. Ethernet uses dc Manchester [Fig. 12-17(c)]. The average dc level is about 21 V, and the maximum swing is about 0 to 22 V. The average direct current is used to detect the presence of two or more signals transmitting simultaneously.

Another benefit of Manchester encoding is that the transition in the middle of each bit time provides a means of detecting and recovering the clock signal from the transmitted data.

Ethernet LANs| Coaxial Cable | Twisted-Pair Cable | Advance Ethernet

Speed


The standard transmission speed for original basic Ethernet LANs is 10 Mbps. The time for each bit interval is the reciprocal of the speed, or 1/f, where f is the transmission speed or frequency. With a 10-Mbps speed, the bit time is 1/10 3 106 5 0.1 3 1026 5 0.1 μs (100 ns) [see Fig. 12-17(a)].

A more widely used version of Ethernet is called Fast Ethernet. It has a speed of 100 Mbps. Other versions of Ethernet run at speeds of 1 Gbps or 10 Gbps, typically over fiber-optic cable but also on shorter lengths of coaxial or twisted-pair cable. The most common version of Ethernet today is Gigabit Ethernet (GE), which is essentially an integral part of every PC and laptop.

Transmission Medium

The original transmission medium for Ethernet was coaxial cable. However, today twisted-pair versions of Ethernet are more popular.

Coaxial Cable

The two main types of coaxial cable used in Ethernet networks are RG-8/U and RG-58/U. RG-8/U cable has a characteristic impedance of 53 V and has approximately a 0.405-in diameter. This large coaxial cable, referred to as thick cable, was originally designed for antenna transmission lines in RF systems. It is usually bright yellow. Large type-N coaxial connectors are used to make the interconnections.

Ethernet systems using thick coaxial cable are generally referred to as 10Base-5 systems, where 10 means a 10-Mbps speed, Base means baseband operation, and the 5 designates a 500-m maximum distance between nodes, transceivers, or repeaters. Ethernet LANs using thick cable are also referred to as Thicknet. This original version of Ethernet is no longer used.

Ethernet systems implemented with thinner coaxial cable are known as 10Base-2 or Thinnet systems; here the 2 indicates the maximum 200-m (actually, 185-m) run between nodes or repeaters. The most widely used thin cable is RG-58/U. It is much more flexible and easier to work with than RG-8/U cable.

Coaxial versions of Ethernet are no longer widely used. Most LANs today use twisted-pair cable because it is simpler and less expensive. Some older legacy coaxial systems may still exist, and some special industrial applications may use them because of the interference rejection provided by the coaxial cable.

Twisted-Pair Cable

More recent versions of Ethernet use twisted-pair cable. The twisted-pair version of Ethernet is referred to as a 10Base-T network, where the T stands for twisted-pair. The twisted-pair cable used in 10Base-T systems is standard 22-, 24-, or 26-gauge solid copper wire with RJ-45 modular connectors. The PC nodes connect to a hub, as shown in Fig. 12-18, which provides a convenient way to connect all nodes. Each port contains a repeater that rejuvenates the signal and buffers it for retransmission.

Ethernet LANs| Coaxial Cable | Twisted-Pair Cable | Advance Ethernet

Physically, a 10-Base-T LAN looks like a star, but the bus is implemented inside the hub itself. It is usually easier and cheaper to install 10Base-T LANs than it is to install coaxial Ethernet systems, but the transmission distances are often limited to less than 100 m.

100-Mbps Ethernet

Several 100-Mbps versions have been developed including 100Base-T or 100Base-TX, also called Fast Ethernet, 100VG-AnyLAN, 100Base-T4, and 100Base-FX. All use twisted-pair cable except the FX version, which uses a fiber-optic cable. The TX and FX methods use the carrier sense multiple access with collision detection (CSMA/CD) access method (described later) and have the same packet size. The 100VG-AnyLAN and -T4 versions use unique access methods. Most, if not all, NICs support 10-Mbps, 100-Mbps, and 1-Gbps rates.

By far the most popular version of 100-Mbps Ethernet is 100Base-TX or Fast Ethernet. It uses two unshielded twisted pairs instead of the single pair used in standard 10Base-T. One pair is used for transmitting, and the other is used for receiving, permitting full-duplex operation, which is not possible with standard Ethernet. To achieve such high speeds on UTP, several important technical changes were implemented in Fast Ethernet. First, the cable length is restricted to 100 m of category 5 UTP. This ensures minimal inductance, capacitance, and resistance, which distort and attenuate the digital data.

Second, a new type of encoding method is used. Called MLT-3, this coding method is illustrated in Fig. 12-19. The standard NRZ binary signal is shown at (a), and the MLT-3 signal is illustrated at (b). Note that three voltage levels are used: +1, 0, and -1 V. If the binary data is a binary 1, the MLT-3 signal changes from the current level to the next level. If the binary input is 0, no transition takes place. If the signal is currently at +1 and a 1111 bit sequence occurs, the MLT-3 signal changes from +1 to 0 to -1 and then to 0 and then to +1 again. What this coding method does is to greatly
reduce the frequency of the transmitted signal (to one-quarter of its original signal), making higher bit rates possible over UTP.

As for the access method, 100Base-T uses CSMA/CD. Most 100-Mbps Ethernet systems use the 100Base-T format described here.

100Base-FX is a fiber-optic cable version of Ethernet. It uses two multimode fiber strands to achieve the 100-Mbps rate. The access method is CSMA/CD. This version of Ethernet is used primarily to interconnect individual LANs to one another over long distances. The full-duplex operation can be achieved at distances up to 2 km.

Gigabit Ethernet

Gigabit Ethernet (1 GE) is capable of achieving 1000 Mbps or 1 Gbps over Category 5 UTP or fiber-optic cable. The 1-Gbps speed is more readily achieved with fiber-optic cable, but it is more expensive. The UTP version of Gigabit Ethernet is defined by the IEEE standard 802.3ab and is generally referred to as 1000Base-T.

The 1-Gbps rate is achieved by transmitting 1 byte of data at a time as if in a parallel data transfer system. This is done by using four UTP cables plus a coding scheme that transmits 2 bits per baud (b/Bd). The basic arrangement is shown in Fig. 12-20.

Ethernet LANs| Coaxial Cable | Twisted-Pair Cable | Advance Ethernet
Ethernet LANs| Coaxial Cable | Twisted-Pair Cable | Advance Ethernet

One byte of data is divided into 2-bit groups. Each 2-bit group is sent to a D/A converter and an encoder that generates a five-level line code. The five levels are +2, +1, 0, -1, and -2 V. Each 2-bit group needs four coding levels. For example, 00 might be +2, 01 = +1, +0 = -1, and +1 = -2 V. The 0-V level is not used in coding but is used in clock synchronization and in an error detection scheme. You may hear this encoding scheme referred to as PAM5. The resulting five-level code is fed to each twisted pair at a 125-MBd rate. Since 2 bits are transmitted for each baud or symbol level, the data rate on each twisted pair is 250 Mbps. Together, the four pairs produce a composite data rate of 4 x 250 = 1000 Mbps or 1 Gbps. The maximum distance is 100 m. Better and more reliable performance is achieved by restricting the distance from PC to hub to 25 m or less. The CSMA/CD access method is used, and communication can be a half or full-duplex.

The fiber-optic cable versions of Gigabit Ethernet are defined under the IEEE 802.3z standard. All data transmissions are serial, unlike transmissions with the 1000Base-T UTP version. The 1000Base-LX version uses a single-mode fiber (SMF) cable with a diameter of 9 μm and a transmitting laser operating at an infrared wavelength of 1310 nm. It can achieve 1 Gbps on a cable length of 10 km. When the larger, less expensive multimode fiber (MMF) is used, the maximum operating distances are 550 m with 50-μm-diameter fiber and 500 m with 62.5-μm-diameter fiber.

A less expensive option is 1000Base-SX, which uses a lower-cost 780-μm laser and MMF cable. The maximum cable length is 550 m with 50-μm-diameter fiber and 275 m with 62.5-μm-diameter fiber. You will learn the details of fiber-optic communication. All the fiber-optic versions of Gigabit Ethernet also use a different data encoding

the scheme called 8B/10B. This method takes each 8-bit byte (also called an octet) of data and translates it into a 10-bit word. The purpose is to ensure that an equal number of binary 0s and 1s are always transmitted. This makes clock recovery and data synchronization easier. The 8B/10B encoding scheme also allows a simple way to implement error detection for 1-bit errors. In addition, it eliminates any dc bias buildup that may occur if the signal is transmitted through a transformer. Finally, the 8B/10B code also implements frame delimiting, providing a way to clearly mark the beginning and end of a frame of data with unique 10-bit codes.

10-Gbit Ethernet

An even faster version of Ethernet is 10-Gbit Ethernet (10GE), which permits data speeds up to 10 Gbps over fiber-optic cable as well as higher grades of twisted-pair. The defining IEEE standard is 802.3ae. As with Gigabit Ethernet, there are several versions as indicated in the table below. All use 8B/10B coding.

Three of the five variations use serial data transmission. The other two use what is called wide-wavelength division multiplexing (WWDM). Also known as coarse wavelength division multiplexing (CWDM), these versions divide the data into four channels and transmit it simultaneously over four different wavelengths of infrared light near 1310 nm. WWDM is similar to frequency-division multiplexing. This technique is described in greater detail.

It is hard to believe that a 10-Gbps pulse signal could be carried over a copper cable, given the huge attenuation and distortion that the cable capacitance, inductance, and resistance can cause. Yet, today there are several copper versions of 10GE now available. One of them is designated as 10GBase-CX4. This version of Ethernet is standardized by the IEEE standard 802.3ak. The cable is a special twin axial coaxial cable (called Twinax) that contains two conductors inside the outer shield. Four of these coaxial cable assemblies are combined to make a cable. The data to be transmitted is divided into four parallel paths that transmit at a 3.125-Gbps rate. The encoding is 8B/10B, meaning that only 80 percent of the bits transmitted are actual data. This gives an actual data rate of
0.8 x 3.125, or 2.5, Gbps. Four paths give an aggregate of 10 Gbps. The range is limited to roughly 15 m or about 50 ft. This is sufficient for connecting several servers, routers, Ethernet switches, and other equipment in wiring closets, data centers, or server farms, such as those in Internet companies where the various pieces of equipment are located close to one another.

Another copper cable version of 10GE has been in development for several years. Designated 802.3an or 10GBase-T, it is designed to use the four pairs of conductors in CAT5e or CAT6 UTP so that existing cabling can be used. The range is 100 m as with most other versions of Ethernet. Because of the severe cross-talk that occurs in UTP at 10 Gbps, extensive DSP filtering and equalization is employed. One final point is in order. Copper cable versions of Ethernet are attractive despite their complexity simply because they are significantly less expensive than fiber-optic versions. While the cost of fiber-optic equipment has declined over the years, it is still 3 to 10 times more expensive than a copper solution.

The primary application of 1-Gbps and 10-Gbps Ethernet is still in LANs. Not only can LANs be faster and handle more users, but also they can increase in size with these forms of Ethernet. Distances up to several kilometers are easily achieved with some versions. The 1-Gbps and 10-Gbps Ethernet technology is used to create a fast LAN backbone that links and aggregates slower LANs. The 1-Gbps and 10-Gbps versions of Ethernet are also used in data centers. A data center is a central point where multiple servers, switches, routers, and related equipment are located along with a connection to the Internet and other networks. The 10GE links are usually used to connect the equipment over distances of less than 100 meters.

The speed and distance capabilities also make 1-Gbps and 10-Gbps Ethernet attractive for MAN applications. The 40-km version even makes 10-Gbps Ethernet appropriate for some WAN applications. It is now replacing the more complex and expensive synchronous optical network (SONET) fi ber-optic equipment now common in most MANs and WANs. (SONET is covered). And that trend continues with 100 Gbps Ethernet.

Wireless Ethernet

Several versions of Ethernet for transmission by radio have been developed. This permits wireless LANs to be created. The cost and complexity of buying and installing cables are eliminated, and nodes can be relocated at any time without regard to where the cable is. Each PC is equipped with a NIC that incorporates a wireless modem transceiver. Wireless LANs using a version of Ethernet called Wi-Fi are described.

Ethernet in the First Mile

Also known as Ethernet Passive Optical Network (EPON), Ethernet in the First Mile (EFM) is the IEEE standard 802.3ah. It is a version of Ethernet designed to be used in fiber-optic networks that connect homes and businesses to high-speed Internet services. The fi rst mile, also called the last mile, is a term used to describe the relatively short connection from a home or office to a local terminal or connection point that distributes data via a fiber-optic link. The EFM system uses the standard Ethernet protocols at a speed of 1.25 Gbps. It permits up to 32 users per connection, and the maximum range is about 20 km. A 10-Gbps version of EPON is now also in service. More details on fi ber optics are given.

Power over Ethernet

Power over Ethernet (PoE) is an addition to Ethernet LANs that is used to deliver dc power to remote devices connected to the network. The related IEEE standards are 802.3af and 802.3at. Specifically, it supplies about 48 V of unregulated direct current over two of the twisted pairs in a CAT5 UTP cable. This eliminates the need for some devices on the LAN to have their own power supply, and it eliminates the need for that remote device to be near a 120- or 240-V ac outlet.

Some examples are wireless access points used to extend the LAN and Voice over Internet Protocol (VoIP) telephones, which are rapidly replacing standard switched analog phones. There are numerous industrial applications as well, such as video surveillance cameras. Fig. 12-21 shows a common PoE arrangement. On the left, a 48-V dc power supply is connected to the center taps of the I/O transformers in the Ethernet NIC. These transformers carry the serial Ethernet data. Both wires in each pair carry the direct current. The wires in the twisted pairs are effectively in parallel for direct current. The direct current does not interfere with the data.

On the receiving end, transformers accept the signal and pass it along to the NIC circuitry in that device as usual. The dc voltage is captured from the center taps. This dc voltage is then translated to another dc level by a dc-dc converter or a voltage regulator. Voltages of 24, 12, 6, 5, and 3.3 V are common. This voltage powers the interface circuits at that end of the cable, thereby eliminating the need for a separate ac line or power supply.

The choice of 48 V was based on the fact that the wires in a CAT5 cable are very small, usually no. 28. Smaller wires have higher dc resistance and so can produce a rather large voltage drop along the cable. By keeping the voltage high, the line current is less for a given amount of power consumption in the load, thereby producing much less of a voltage drop. In practice, the maximum range is 100 m, and the voltage can usually be anything from 44 to 57 V as the dc supply is unregulated.

The maximum allowed current is 550 mA, although the current is usually held to a value of 350 mA or less. At 48 V, this translates to a maximum current consumption of 16.8 W. The standard states that the maximum desirable load is 15.4 W. Because some power is dissipated in the cables, the maximum load is 12.9 W. Most loads consume much less than that. A newer version of the standard 802.3 uses larger wire in the CAT5 or CAT6 cable to cut losses and uses all four pairs in the UTP cable to deliver a maximum current to 600 mA and a maximum power of 25.5 W.

Power over Ethernet is designed to work with all UTP versions of Ethernet, including 10-, 100-, and 1000-Mbps systems. Only two pairs are used. The dc power is applied to the cable with a separate piece of equipment called an injector. Sometimes the direct current is supplied inside a hub or switch. Different versions of the standard vary with the pairs defined to carry the direct current and which pins on the RJ-45 connectors are used. Some companies offer variations that supply 12 V instead of 48 V.

Access Method

Access method refers to the protocol used for transmitting and receiving information on a bus. Ethernet uses an access method known as carrier sense multiple access with collision detection (CSMA/CD), and the IEEE 802.3 standard is primarily devoted to a description of CSMA/CD.

Whenever one of the nodes on an Ethernet system has data to transmit to another node, the software sends the data to the NIC in the PC. The NIC builds a packet, a unit of data formed with the information to be transmitted. The completed packet is stored in RAM on the NIC while the sending node monitors the bus. Because all the nodes or PCs in a network monitor activity on the bus, whenever a PC is transmitting information on the bus, all the PCs detect (sense) what is known as the carrier. The carrier in this case is the Ethernet data being transmitted at a 10-Mbps (or 100-Mbps) rate. If a carrier is present, none of the nodes will attempt to transmit.

When the bus is free, the sending station initiates the transmission. The transmitting node sends one complete packet of information. In a sense, the transmitting node “ broadcasts’’ the data on the bus so that any of the nodes can receive it. However, the packet contains a specific binary address defining the destination or receiving node. When the packet is received, it is decoded by the NIC, and the recovered data is stored in the computer’s RAM.

Although a node will not transmit if a carrier is sensed, at times two or more nodes may attempt to transmit at the same time or almost the same time. When this happens, a collision occurs. If the stations that are attempting to transmit sense the presence of more than one carrier on the bus, both will terminate transmission. The CSMA/CD algorithm calls for the sending stations to wait a brief time and then attempt to transmit again. The waiting interval is determined randomly, making it statistically unlikely that both will attempt retransmission at the same time. Once a transmitting node gains control of the bus, no other station will attempt to transmit until the transmission is complete. This method is called a contention system because the nodes compete, or contend, for use of the bus.

In Ethernet LANs with few nodes and little activity, gaining access to the bus is not a problem. However, the greater the number of nodes and the heavier the traffic, the greater the number of message transmissions and potential collisions. The contention process takes time; when many nodes attempt to use a bus simultaneously, delays in transmission are inevitable. Although the initial delay might be only tens or hundreds of microseconds, delay times increase when there are many active users. Delay would become an insurmountable problem in busy networks were it not for the packet system, which allows users to transmit in short bursts. The contention process is worked out completely by the logic in the NICs.

Packet Protocols

Fig. 12-22(a) shows the packet (frame) protocol for the original Ethernet system, and Fig. 12-22(b) shows the packet protocol defined by IEEE standard 802.3. The 802.3 protocol is described here.

The packet is made up of two basic parts: (1) the frame, which contains the data plus addressing and error detection codes, and (2) an additional 8 bytes (64 bits) at the beginning, which contains the preamble and the start frame delimiter (SFD). The preamble consists of 7 bytes of alternating 0s and 1s that help to establish clock synchronization within the NIC, and the SFD announces the beginning of the packet itself with the code 10101011.

The destination address is a 6-byte, 48-bit code that designates the receiving node. This is followed by a 6-byte source address that identifies the sending node. These are the MAC addresses. Next is a 2-byte field that specifies how many bytes will be sent in the data field. Finally, the data itself is transmitted. Any integer number of bytes in the range from 46 to 1500 bytes can be sent in one packet. Longer messages are divided up into as many separate packets as required to send the data.

Finally, the packet and frame end in a 4-byte frame check sequence generated by putting the entire transmitted data block through a cyclical redundancy check (CRC). A resulting 32-bit word is a unique number designating the exact combination of bits used in the data block. At the receiving end, the NIC again generates the CRC from the data block. If the received CRC is the same as the transmitted CRC, no transmission error has occurred. If a transmission error occurs, the software at the receiving end will be notified. Sometimes when the data length is short, 1 or more padding bytes (octets) are added between the data and CRC.

It is important to point out that data transmission in Ethernet systems is synchronous: The bytes of data are transmitted end to end without start and stop bits. This speeds update transmission but puts the burden of sorting the data on the receiving equipment. The clocking signals to be used by the digital circuits at the receiving end are derived from the transmitted data itself to ensure proper synchronization and counting of bits, bytes, fields, and frames.

Example 12.2 How fast can a 1500-byte block of data be transmitted on a 10-Mbps Ethernet (IEEE
802.3) packet? [Note: For the Ethernet IEEE 802.3 format, use Fig. 12-22(b)].

Ethernet and the OSI Layers


It is helpful to understand how Ethernet relates to the OSI model discussed. Ethernet actually defines the first two layers; layer 1 is the physical layer (PHY) and layer 2 is the data link layer. The physical layer defines all the hardware that handles the transfer of bits from one place to another. It identifies the medium type, connectors, encoding and decoding, and all the related signaling functions. Any network interface card or chip includes the physical layer components. Repeaters and hubs are also layer 1 devices.

The data link layer is actually divided into to sublayers called the media access control (MAC) sublayer and the logical link control (LLC) sublayer. The MAC layer handles all the encapsulation of the data to be transmitted, which includes the wrapping of the preamble, SOF delimiter, destination and source addresses, and CRC around the data. Hardware within the interface card or chip handles these chores.

The LLC assists the MAC sublayer in dealing with the transmitted or received data in the upper layers of the OSI stack. In the data field shown in Fig. 12-22(b), several additional fields are appended to the beginning of the data. These fields control the handling of the data, identify services available, and define the protocol to be used in the upper layers. There are several variations of these additional fields, which are beyond the scope of this text.

Advanced Ethernet

As old as Ethernet is, it continues to evolve. Today it is faster than ever and offers multiple new features and characteristics. Here is a summary of the most recent developments.

40 Gbps and 100 Gbps Ethernet

As semiconductor and optical technologies have progressed, it has been possible to push Ethernet speeds higher and higher. Over the years, speeds have increased by factors of 10 from 10 to 100 to 1000 Mbps (1GE) and today 10 Gbps (10GE). Now 40 Gbps (40GE) and 100 Gbps (100GE) versions are available and already deployed. It was initially thought that the next decade increment from 10 Gbps to 100 Gbps would be too difficult and impractical, so a 40-Gbps option was created. As it turned out, the 100-Gbps level was achievable. The 40GE/100GE version of Ethernet was standardized in 2010 as IEEE 802.3ba. With these speed levels, Ethernet can compete with MAN and WAN services as well as maintain its dominance in the LAN arena.

The goal of the 40GE/100GE development was to create a number of different versions that could be used not only in LANs but also in WAN, MAN, data centers, and equipment backplanes. Other objectives were to keep the standard frame format and to achieve a BER of at least 10−12. Those goals were reached, and the result is summarized in Table 12-1.

Most of the options in Table 12-1 are fiber-optic cables. Yet some copper cable options are available. A backplane is basically a printed circuit board (PCB) used as a base to interconnect other integrated circuits and PCB assemblies with connectors. The copper line on the PCB replaces the cables. The length of the copper lines is less than 1 meter. KR and KP designate copper backplanes. The 4 means four lanes or paths. The copper cable is the Twinax coax as described earlier. The designation CR4 means four parallel cables. A new version of unshielded twisted-pair designated CAT8 is used in a 40GBASE-T version. CAT8 cable or connectors are not yet available as of this writing. All of the other versions are fiber-optic cables. These copper versions are intended for use in data centers to interconnect servers, switches, routers, and other equipment.

To achieve the 40- and 100-Gbps speeds, a variety of methods are used. First a 64B/66B version of the Reed-Solomon (R-S) forward error correction (FEC) code is used to ensure a good BER. Each 64-bit block of data is converted to a 66-bit code that permits error detection and correction.

Next either 4 or 10 parallel lanes or paths are used. For 40 Gbps, four 10-Gbps paths are used. Each path may be a separate fiber or a different light “color” or wavelength on a single fiber. This latter approach is a form of frequency division multiplexing called wavelength division multiplexing (WDM).

To deliver 100 Gbps of data, several methods are used. One of them is to use four 25-Gbps paths on separate fibers or wavelengths of light. Alternately, 10 paths at 10 Gbps are used. One version uses multilevel PAM4 to deliver 2 bits per baud per path. A single serial path is used in the 40GBASE-FR version over SMF. Note that 40GBASE-SR4 and 100GBASE-SR10 use different types of MMF cable. OM4 has less loss and attenuation than OM3. You will learn more about fiber cables.

In Table 12-1 the terms SR means short-range or reach, LR means long-range or reach, FR means far range or reach, and ER means extended range or reach. The numbers 4 and 10 indicate the number of paths used. Also, with the overhead of extra bits added by using the R-S 64B/66B FEC, that actual line speed is higher. For example, for 10-Gbps paths, the actual rate is 10.3125 Gbps producing a full rate of either 41.25 Gbps or 103.125 Gbps. For the versions using four paths for 100 Gbps, the line rate per path is 25.78125 Gbps giving a total rate of 103.125 Gbps. In all cases, the actual data rates after FEC processing are 40 Gbps or 100 Gbps.

The 40GE and 100GE versions are being used right now. However, new versions for even higher data rates are in development. The next increments are 200 Gbps and 400 Gbps. The next decade increment of 1 terabit (1T) Ethernet is still to come. More details on fiber-optic systems are included.

Carrier Ethernet

With the 40-Gbps and 100-Gbps speeds, Ethernet can potentially be used to replace existing technologies in the MANs and WANs. For example, Ethernet could replace multiple TDM T1 or T3 carrier systems in MANs or Sonet (Synchronous optical network) in WANs. Furthermore, Ethernet can provide communication links in the Internet backbone networks or mobile backhaul connections. Internet connectivity is now being referred to as the cloud, where the cloud represents a mixture of networks that connect users with their smartphones, tablets, laptops, PCs, or even individual devices like sensors or appliances. Cloud computing means that users access storage or software at remote sites rather than apply those available locally on the user’s device. For these applications, more than just a high data rate is needed.

Ethernet has been only a “best-effort” connection where BER was not particularly high or necessary. For commercial MANs and WANs, a higher degree of reliability and better BER is essential. In addition, Ethernet packet or frame delivery times can vary widely depending on the nature of the media involved and the circuitry. In many systems, there is considerable latency or delay, which makes the time of delivery highly variable or unpredictable. For some applications, such as real-time control in industrial applications of cellular network service, Ethernet must be deterministic or provide some form of predictable timing and synchronization over distance. Finally, Ethernet for WAN, MAN or cloud computing needs greater security.

All of these needs are now being met by an enhancement to Ethernet called Carrier Ethernet (CE). CE uses additional hardware and software with standard Ethernet to provide the more reliable, deterministic, and secure service demanded by banks, other businesses, and industry. Now traditional telecommunications and Internet service companies can begin to use Ethernet to replace existing systems for improved service. In the meantime, Ethernet can still be repackaged and transported over other technologies, such as Sonet, that may still be used or the new Optical Transport Network (OTN). These are discussed later.

The key feature of Carrier Ethernet is guaranteed Quality of Service (QoS), which provides the essential reliable and timely delivery of any data (voice, video, etc.). Next, is a feature called Operation, Administration, and Maintenance (OAM), which permits a carrier to provide the service remotely as well as monitor and control the network from another site. OAM allows online troubleshooting as well as a quick response to customers who want to expand their service.

Carrier Ethernet is a large and complex addition to Ethernet created and managed by the Metro Ethernet Foundation (MEF). MEF also certifies systems that meet their rigid standards and specifications. With CE, Ethernet offers a lower-cost but effective networking option to those building MANs and WANs.

Watch video

Protocols | OSI Model | Error Detection | Redundancy | Convolutional ( Ethernet LANs| Coaxial Cable | Twisted-Pair Cable | Advance Ethernet )

Spread Spectrum | Wideband Modulation | Broadband Modem ( Ethernet LANs| Coaxial Cable | Twisted-Pair Cable | Advance Ethernet )

Multiplexer | Demultiplexer | FDM | TDM | PAM | Applications ( Ethernet LANs| Coaxial Cable | Twisted-Pair Cable | Advance Ethernet )

Digital Codes | Hartley’s Law | ASCII | Asynchronous | Encoding Methods ( Ethernet LANs| Coaxial Cable | Twisted-Pair Cable | Advance Ethernet )

Atomic Structure | Voltage Source | Maximum Power Transfer Theorem ( Ethernet LANs| Coaxial Cable | Twisted-Pair Cable | Advance Ethernet )

Click Here to Learn More ( Ethernet LANs| Coaxial Cable | Twisted-Pair Cable | Advance Ethernet )

Click Here to Learn ( Ethernet LANs| Coaxial Cable | Twisted-Pair Cable | Advance Ethernet )

Reference : Electronic communication by Louis Frenzel

Leave a Reply

%d bloggers like this: