Typically the least expensive high-speed LAN system, Ethernet technology is by far the most popular.

Last month, we introduced the concept of structured cabling in our discussion of computer networks (Feb. 2000, page 56F). As you may recall, structured cabling refers to a network cabling system designed and installed according to preset standards. This month, we focus our attention on local area network (LAN) systems with a particular focus on Ethernet technology.

Ethernet is typically the least expensive high-speed LAN system, and by far the most commonly used. Xerox research is responsible for the early development of Ethernet. In fact, as the holder of the trademark name Ethernet, Xerox Corporation established and published the standards. With more technology, a second generation called Ethernet II became popular. We often call Ethernet from this period "DIX" after its corporate sponsors Digital, Intel, and Xerox.

Obviously, no technology could become an international standard for all sorts of equipment if a single U.S. corporation controlled the rules. So, IEEE developed formal international standards for LAN technology. An IEEE committee devotes its time to looking at Ethernet, token ring, fiber-optic, and other LAN technology. The objective was not just to standardize each LAN individually, but to establish rules that would be global to all types of LANs -- so data could easily move from one network to another.

The IEEE eventually published a set of standards. The most important of these are:

• 802.3: Hardware standards for Ethernet cards and cables;

• 802.5: Hardware standards for token ring cards and cables; and

• 802.2: New format for data on any LAN.

But all of these networking systems require some means of transferring signals from one place to another; and the electrical characteristics of these various cables are a critical component of the networking standards.

Cables for high-speed networks. Normally, wire-pair cables cannot carry high bandwidth signals fast enough for network communications. The wires also radiate like antennas, so they interfere with other electronic devices. However, several developments allowed use of simple wire for high-speed signals.

The first development is using twisted pair cables. The two conductors are tightly twisted (one to three twists per inch) to couple the signals into the pair of wires. Each pair of wires is twisted at different rates to minimize cross coupling. (Cross coupling is essentially the same as crosstalk; the transferring of a signal from one pair of wires to another via electromagnetic induction.)

The next step is to use balanced transmission to minimize electromagnetic emissions. Balanced transmission works by sending equal but opposite signals down each wire. The receiving end looks at each wire and sees a signal of twice the amplitude carried by the pair of wires; this helps to get enough signal through the link. While each wire radiates because of the transmitted signal, they carry the opposite signal so the two wires cancel out the radiated signals and reduce electromagnetic interference.

Using these techniques, and having two pair of wires sending signals in opposite directions, it's possible to adapt unshielded twisted pair (UTP) cables to work with Ethernet and token ring networks. At higher speeds, signals are compressed and encoded, and multiple pairs transmit in each direction to allow operation with networks of over 100 megabits per second.

Typical UTP cables have four twisted pairs in the cable, two of which simultaneously transmit signals in opposite directions, creating a full duplex link. Some of the higher speed networks now use all four pairs to reduce the total bandwidth requirement of any single pair.

We also use more complex cables for LANs, including shielded twisted pair (STP) cable with each pair of wires shielded individually and an overall shield provided, and screened twisted pair cable (ScTP), which has four twisted pairs inside a foil shield.

Devices. Here are a few of the devices the industry commonly uses in Ethernets:

• A repeater receives and then immediately retransmits each bit. It has no memory and does not depend on any particular protocol. It duplicates everything, including collisions. The term collision refers to two electronic signals transmitted into each other (not separated as they should be). When this happens, the clashing voltages add to or diminish each other, usually leaving the signals unintelligible at the receiver.

• A bridge receives an entire message into memory. If a collision or noise damages the message, the bridge discards it. Likewise, if the bridge knows the message is transmitting between two stations on the same cable, it discards it. Otherwise, the message is queued up, and it retransmits on another Ethernet cable. The bridge has no address. Its actions are transparent to the client and server workstations.

• A router acts as an agent to receive and forward messages. The router has an address, and the client or server machines recognize it. Typically, machines directly send messages to each other when they are on the same cable, and they send the router messages addressed to another zone, department, or subnetwork.

• A driver is the software that allows an Ethernet card in a computer to decode packets and send them to the operating system and encode data from the operating system for transmission by the Ethernet card through the network. It provides a device-independent interface to the upper layer protocols.

• Ethernet adapter cards. For a network computer, these cards usually range in price from $60 to $120. They transmit and receive data at speeds of 10 million bits per sec through up to 100 m (90 m horizontal, and 10 m in closet and wall) of cable to a hub device normally stacked in a wiring closet. The hub adds about $50 to the cost of each desktop connection. It's interesting that Ethernet and EIA/TIA Standard 568 are the same. In fact, Ethernet is designed to run over the 568 structure.

Data transmission. We call a block of data transmitted on Ethernet a frame. The first 12 bytes of every frame contain the 6-byte destination address (the recipient) and a 6-byte source address (the sender). Each Ethernet adapter card comes with a unique factory-installed address (called a universally administered address). Use of this address guarantees a unique identity to each card.

The source address field of each frame must contain the unique address (universal or local) assigned to the sending card. The destination field can contain a multicast address representing a group of workstations with some common characteristic.

An Ethernet adapter card receives only frames with a destination address matching its unique address or destination addresses that represent a multicast message. However, you can set most Ethernet adapter cards into "promiscuous" mode, where they receive all frames on the network. If this is a security problem, smart hub devices can filter out frames with private destination addresses belonging to another station.

Troubleshooting the network. Ethernets typically fail in one of three ways:

• A nail or other object breaks one conductor.

• A screw or other object touches one or more of the conductors and short them to an externally grounded metal shield, conduit, or other grounded metal.

• A station on the network breaks down and starts to generate a continuous stream of electronic noise, thus blocking legitimate transmissions.

To find problems in an Ethernet, we use a time domain reflectometer (TDR). It plugs into any attachment point in the cable and sends out its own voltage pulse. The effect is similar to a sonar ping. If the cable breaks, the pulse hits the loose end of the cable and bounces back. It senses the echo, computes the length of the round trip, and calculates the distance of the break.

If an Ethernet cable shorts out, a voltmeter determines proper resistance is missing from the signal and shield wires.

However, there's another type of network problem: the installation of the cabling itself. Networking cabling is more problematic than power wiring, and if not put in under precise standards, it may not work. We'll cover this in detail next month.