The different media systems use different encoding methods

French | Spanish | Portuguese | Italian | German | Japanese | Chinese | Korean | Russian | Arabic

Transmitting the Bits

The different media systems use different encoding methods in transmitting the data bits or code symbols that represent the data bits. Ten-Mb/s systems use a method called Manchester encoding (Network article 2-18). In a logic 1, the signal is a low voltage or low light level for the first half of the bit period and a high voltage or high light level for the second half of the bit period. A logic 0 is the reverse: the signal is high for the first half of the bit period and low for the second half. The advantage to Manchester encoding is that it guarantees a transition in each bit, and this makes signals easy to synchronize to. The down side is that the maximum rate of transitions is twice the bit rate, and hardware that can handle higher bit rates tends to be more expensive.

In Fast Ethernet systems, the bits in a code symbol transmit using a method of encoding called multi-level transition 3, or MLT-3. Instead of specifying two voltage ranges that correspond to logic 0 and logic 1, MLT-3 uses three voltage ranges. For each bit time, a change from one logic level to the next signifies a logic 1, while no change signifies a logic 0. The data in Fast Ethernet systems is also scrambled before transmitting to provide bit patterns that reduce electromagnetic emissions. In twisted-pair Gigabit Ethernet, instead of serially transmitting the bits in the code symbols, which would take five bit times, the transmissions use a system of 5-level pulse-amplitude modulation. A code symbol is one of five defined voltage levels, and each level represents a 2-bit value. In this way, two bits of data transmit in one bit time on each pair of wires. With each transition representing two bits and four signal pairs transmitting at once, each bit time transmits 8 bits of information. Transmitting a Gigabit per second, the maximum transition rate is 125 transitions per microsecond, which is the same maximum transition rate as in Fast Ethernet. Twisted-pair Gigabit systems also uses error correcting, digital signal processing, and other techniques to ensure signal quality. An Idle symbol transmits when there is no other traffic. Fiber-optic segments for Fast Ethernet transmit the individual bits in a code symbol using a method of encoding called non-return to zero, invert on ones (NRZI). A logic 0 results in no change in light level; if light transmitted during the previous bit’s time period, light continues to transmit for the logic-0 bit that follows. A logic 1 results in a change in light level: if light transmitted during the previous bit’s time period, light doesn’t transmit for the logic-1 bit that follows, and if light didn’t transmit during the previous bit’s time period, light does transmit for the logic-1 bit that follows. Gigabit fiber-optic segments transmit the individual bits in the code symbols using non-return-to-zero (NRZ) encoding. NRZ is the simplest encoding: light transmits to indicate a logic 1 and light doesn’t transmit to indicate a logic 0.

Interfacing to Ethernet Controllers

The different cable types and speeds require different hardware interfaces to Ethernet controller chips. The Ethernet standard defines several types of interfaces for connecting a computer’s media access control (MAC) layer, which manages the sending and receiving of network data, to the physical layer, which contains the components that are specific to a cable type or speed. Depending on the cable type and network speed, the physical layer may contain little more than transceivers, some filtering, and a connector, or the layer may include circuits that encode and decode data and convert between serial and parallel interfaces. Many Ethernet controllers have an on-chip interface for twisted-pair cables and require only filtering circuits and a connection to an RJ-45 connector. Network article 2-19 shows some options for cable connections. In Network article 2-19A, a 10-Mb/s controller uses an attachment unit interface (AUI) that connects to a media attachment unit (MAU) that in turn interfaces to the network cable. The MAU is a separate unit that attaches to the controller’s 15-pin AUI interface. The MAU provides an interface to an Ethernet connector, a collision-detect output, and jabber-detection circuits to prevent a malfunctioning interface from continuously transmitting. With an AUI, you can switch between coaxial, twisted-pair, and fiber-optic cable by swapping the MAU. In Network article 2-19B, a 10- or 100-Mb/s Ethernet controller uses a medium-independent interface (MII) that connects to a physical layer device (PHY), which in turn connects to the network cable. The PHY may be on the same circuit board as the Ethernet controller or a separate unit. The MII converts between the network’s serial data and a 4-bit data bus that connects to the Ethernet controller. Some Ethernet controller chips include a PHY core for twisted-pair interfaces. An Ethernet controller that includes an embedded PHY typically requires external filtering circuits between the PHY and network connector. In Network article 2-19C, a Gigabit Ethernet controller uses a Gigabit medium-independent interface (GMII) that connects to a Gigabit PHY, which connects to the network cable. Because of Gigabit Ethernet’s speed, the GMII can’t use a cable to connect to the Ethernet controller, but must be on the same circuit board or a daughter board. The GMII converts between the network data and an 8-bit data bus that connects to the Ethernet controller. Another option for a Gigabit Ethernet controller used with fiber-optic cable is a Ten-Bit Interface (TBI) (Network article 2-19D), which connects to a Gigabit PHY, which connects to the network cable. The TBI converts between the network data and a 10-bit data bus that connects to the Ethernet controller. Network article 2-19E shows an Ethernet controller that contains an embedded PHY for use with twisted-pair cable.

Using Repeater Hubs, Ethernet Switches, and Routers

Ethernet networks have several options for interconnecting the computers in a network. As Chapter 1 explained, repeater hubs and switches have attachment points for two or more cables that can connect to other interfaces in a local network. Both repeater hubs and switches repeat, or regenerate, traffic received on one port to the other ports. A repeater hub repeats all traffic to all ports, with the exception of some multi-speed repeater hubs that convert traffic between speeds only when necessary. A switch examines the destination of all received traffic and when possible, forwards the traffic only to the port on the path to the destination. A note on terminology: the Ethernet standard uses the terms repeater hub and switching hub to distinguish between the two device types. However, in popular use, hub by itself generally refers only to repeater hubs, while switch refers to switching hubs. To prevent confusion, this networking tutorial uses the standard’s term repeater hub but avoids using switching hub in favor of switch or Ethernet switch. Repeater hubs and switches make it easy to add and remove network interfaces. Each repeater hub or switch has multiple ports. To add an interface, you just attach the interface’s cable to an available port on the repeater hub or switch. With most media systems, you can add more ports by connecting additional repeater hubs or switches to available ports on existing repeater hubs and switches. As explained earlier, networks that use coaxial cable have other options such as T connectors, but twisted-pair and fiber-optic networks must use repeater hubs or switches to connect more than two interfaces. Communicating with other networks, including the Internet, requires an additional piece of equipment: a router. Repeater hubs, switches, and routers are readily available as off-the-shelf products. A network of embedded systems can use the same devices as networks that link PCs. For Internet communications, a local computer may connect via a modem to a router at an Internet Service Provider.

Repeater Hubs

A repeater hub can connect multiple interfaces and helps to ensure reliable communications by detecting collisions, regenerating missing preamble bits, and blocking traffic from failed interfaces. The IEEE 802.3 standard specifies the functions of repeater hubs that support a single speed but doesn’t forbid hubs that support multiple speeds. On receiving traffic from an interface, the repeater hub repeats the traffic, passing it to each of the other attached interfaces. In a network that uses repeater hubs, each interface sees all of the traffic from the other interfaces, with two exceptions. Repeater hubs block traffic from failed interfaces. And a multi-speed hub may convert between speeds only when necessary.