Home Networking Depot
Bonding the family dwelling through technology


Wal-Mart.com USA, LLC


ETHERNET is currently the 2nd most common form of home networking, the reason is low cost and offers very fast transmission. It is most commonly found in 10Mb or 100Mb speed and requires an adapter card (also known as a Ethernet card) for each PC or similar device you want to connect. Its open architecture lets you mix hardware from different companies, so you are not locked into one vendor's product line alone. And it is easy to expand. Category 5 cable is recommended for fast transmission rates and you are required to have a central hub or switch. You may want to pass wire through walls (similar to phone lines), an undertaking that can be both expensive and time consuming. And although there are many different implementations of Ethernet, they're all based on the same basic logical bus topology, the same system of sending data in relatively small 'packets' or frames which carry the address of both the sending and receiving device, and the same method of allowing a lot of data transceivers to share that common bus.


In 1985, the IEEE working group came out with ‘thin’ Ethernet, also known as ‘cheapernet’ or ‘10Base2’. This specified the use of thinner (5mm diameter) 50W coaxial cable, which still allowed 10Mb/s baseband transmission as before, but with a maximum cable length of 185 (rounded to ‘200’) metres.

Then in 1990 the IEEE 802.3i or ‘10BaseT’ Ethernet standard was released, which opened the door to much cheaper networking because it allowed 10Mb/s transmission over the low cost 100W unshielded twisted-pair or ‘UTP’ Category 3 cabling which had by then become widely used for telephone wiring in buildings. Using this cable also allowed the network to use a ‘star’ configuration or topology, rather than the bus or ‘daisy chain’ topology needed for thick and thin Ethernet (10Base5 and 10Base2). The two different topologies are Daisy Chain Star Topology.


The next big development came in 1995, when the IEEE working group released the 802.3u standard. This became known as ‘100BaseT’ or ‘Fast Ethernet’, because it allowed 100Mb/s baseband transmission over either two pairs of Category 5 100W UTP cabling (100BaseTX), or four pairs of Category 3 cabling (100BaseT4) or two multi-mode fibre-optic cables (100BaseFX). In other words, 10 times the speed of 10BaseT.

Then in 1997 came IEEE 802.3x, which defined full duplex or simultaneous two-way data communication over either 10BaseT or 100BaseT. This effectively doubled the data rate again, because before this Ethernet allowed only half duplex or ‘one way at a time’ communication especially in 10Base5 and 10Base2 coaxial systems.

Then in 1998 and 1999, the IEEE working group released four different implementations of the 802.3z ‘Gigabit Ethernet’ standard, achieving 1Gb/s transmission or another 10-times increase in data transfer rates. The four versions of this are 1000BaseSX, which uses 850nm lasers and a multi-mode fibre-optic cable; 1000BaseLX, which uses a 1300nm laser and either single or multi- mode fibre-optic cable; 1000BaseCX, which uses ‘twinax’ shielded twisted-pair (STP) cable; and 1000BaseT, which uses four pairs of Category 5 UTP cabling.


Without going too deeply into the technicalities, CSMA/CD essentially works by having each device 'node' on the network listen for bus activity (Le., carrier sensing) before it tries to transmit a packet of data. This is like a subscriber on one of the old telephone 'party lines' picking up the receiver to listen if someone is already using the line, before they try to make a call themselves. If a device doesn't detect any bus activity, it begins to transmit the packet of data. But of course it's possible for another device to begin sending its data at much the same time, in which case there'll be a 'collision' - two packets of data are present on the bus at the same time, and the data gets 'mixed up'. While the devices are transmitting their data, they're also sensing the bus so they can monitor for any collisions (this is the collision detection function). If they sense the extra bit transitions produced by a collision, both devices stop transmitting their data packets and the first device sends out a special 'jam sequence' code - basically a short collision alarm message. Then both of the devices that were trying to transmit wait short but random periods of time before listening for bus activity and trying to transmit again. This is known as backing off, and the random back-off delays are to try and ensure that they shouldn't have a collision with their next attempts.

IF you have questioned why the Ethernet standards have a specification for the maximum cable length between any two device 'nodes', it's because of the way the CSMA/CD collision detection scheme works. It always takes a certain time for signals to 'propagate or transmit' along a cable, and this delay time is proportioned directly to the cable's length in which is working. But for the collision detection scheme to work, each device must be able to receive the collision warning signal from any other device before it's too late to respond. So the maximum total 'round-trip' propagation time must be less than the Ethernet's CSMA/CD slot time. Since cable delay time (in each direction) is the main contributor to total propagation time, this sets a limit to the maximum cable length between any two PC nodes on the network.


10BaseT Ethernet uses low cost 'Category 5' (or better) four-pair UTP cabling of nominal loon impedance to connect up the computers and other equipment as a network. All connections are normally made via 8-pin RJ45 modular plugs and sockets, which are very much like the 4-pin RJ 11 and 6-pin RJ 12 modular connectors now used for telephones, only slightly larger. The basic connections used for IOBaseT UTP cabling and the RJ45 connectors are shown in figure below As you can see only two pairs of wires are actually used; the other two pairs are connected, but not used.Each computer


connects to the network via a network Interface adaptor card or 'NIC, which is basically a plug-in card containing a controller to manage the data interface between that computer and the network, and a data transmitter/receiver combination or 'transceiver'. In most modern NICs the controller and transceiver are combined in a single LSI chip, which makes the card verylow in cost. (By the way some of the latest model PCs have a network interface built directly into the motherboard, so they don't need the addition of a separate NIC. They have all the hardware necessary for networking already present.) An Important point to remember Is that regardless of whether the PC's network Interface Is In the form of a NIC or built into the motherboard, it has a fixed and unique network or 'MAC' (media access control) address which Is hardwlred Into It during manufacture. This address Is a six-byte (i.e., •• a-bit) binary number, which Is used by the network to Identify the node at which that PC Is connected. The interface or NIC will only 'accept' data packets which carry this address In their dlstlnatlon address field (FIG), and will always include the same address code In the source address field of any data packets It transmits out to the network.


Often diagnostic software will represent the NICs MAC address as a 12-digit hexadecimal number, such as 054F 17B3A8. Not that you normally need to worry
about your PC's NIC address, of course - the networking software looks after all that. To allow the NICs controller to manage the exchange of data between the computer's processor and all other computers on the network, you have to install networking 'driver' software in the computer. NICs usually come with a CD-ROM which provides a range of software drivers, to suit different PC operating systems and/or network operating systems. The simplest possible 10BaseT network is shown in FIG.6. As you can see it's simply a pair of PCs, each fitted with a NIC and with the two NIC ports connected together with a Category 3 or higher rated UTP 'patch lead'. If the two PCs are running virtually any modern operating system (i.e., Windows XP or later) it's simply a matter of installing the appropriate software drivers in each computer and they'll be operating as a fully functional 'peer to peer' network. Data can be transferred between the two in either direction at 10Mb/s, simultaneously if necessary.


10 megabits per second is quite fast, of course. Fast enough for most home and small office networks, in fact. But if you do need even faster networking, this can be achieved quite easily by using 100BaseT (strictly 100BaseTX) instead. This is very similar to 10BaseT, and still uses four-pair UTP cable; the only difference is that the connecting cable(s) must now be Category5 rated - so that they can cope with 100BaseT data rates of up to 100Mb/s. The actual cable connections are still exactly the same. For 100BaseT you also need to use NICs which are rated for 100Mb/s operation, as well. Luckily most currently available low cost NICs are capable of operating at either the 10BaseT or 100BaseT rates, so this isn't likely to be a problem. Most of these so-called' I 0/1 OOBaseT' cards are designed to plug into a PCI bus card slot in the computer, to take advantage of the higher speed.These are the only two real differences between I OBaseT and I OOBaseTX - the latter needs Category 5 cabling and NICs capable of the higher speed. So the simple two-PC network shown in FIG.6 could operate at either 10Mb/s or 100Mb/s in either direction, depending on the rating of the cables and NICs. The maximum length of cable between the two PC's would be 100 metres