The Physics of Ethernet Networks
Ethernet is a networking standard engineered by the Institute of Electronic Engineers. It’s designed to allow efficient intercommunication between computers and related devices that are physically close to each other. Any device connected to this type of network can communicate with any other device in the network. In general and for the purpose of this paper, a copper wire is connected to every device within the network. The wire is attached to network card, which consists of a variety of transistors. The information is passed from computer to computer in binary form through electrical pulses. Each electrical pulse corresponds to one bit. The speed of the network is given directly in the amount of bits sent through the medium. A modern Ethernet network operates at a speed of 100 Mbps (100,000,000 bits per second). [1]
Bits are actually sent through the wire by a wave of electrical voltage. When the signal it received it’s terminated by a resistor at the end of the wire, which removes the single, preventing it from bouncing back. All information is carried in packets with is a binary encoded information that contains the sender’s network address, the recipients network address, how many packets the information is broken up into, that specific packet number, the information itself, and then error correction and then data to show end of information. [3]
Twisted pair cabling, the connected wiring for Ethernet uses a separate conductor for receiving and transferring information. The two twisted copper pairs (2 pairs of 2 wires that consist of 2 twisted wires each, 8 individual wires) are sometimes wrapped in an insulator to block out interference and other electrical signals. Two of the pairs are setup to transmit data in either direction, but only one direction at a time. The other two are setup to transmit in one (opposite to each other) direction only. The single direction channel in the appropriate direction is used for collision detection.
Collisions occur when two devices start transmitting at the same time. Any device on the network can begin transmitting data if the network is “quiet”, meaning there isn’t any current data being transferred along the network. When one device sends a packet to another device on the network, all the other connected devices see the packet, but only the intended device processes it. This is where network addressing becomes important, which will be discussed later on in this work. Any device can begin transmitting when there isn’t data already being transferred. If two devices begin transmitting at the same time, the electrical pulses collide with each other in the wire and send an unrecognizable signal to all connected stations. When this happens both stations wait a random amount of time before they begin retransmission. This is what’s called CSMA/CD. Which is the acronym for carrier-sense multiple access with collision detection. [2]
Original Ethernet sends data at a rate of 10 megabits per second. Which gives a time delay of 100 nanoseconds per bit being sent. Light and electricity travel about one foot per nanosecond. When a station has transmitted it’s first bit, that bit travels approximately 100 feet down the wire, before that station has begun to send the second bit. If the bit has a collision with another bit farther down the wire, the device farther down will receive the collision notification earlier than the original sending device, thus the second device will wait a random amount of time and then start resending, theoretically before the first device is even aware of the collision. To prevent this from happening, there is a frame “delay” or a minimum size number for each transmission which relates to an amount of time equal to the longest path through the network. As Ethernet (remember it’s just a standard for networking) is normally referred to, this path is around 50 microseconds. In an example of a 10 Base T Ethernet Network the signaling speed would be 10 million bits per second, this would be enough time to transmit 500 bits. Converting that to 8 bits per byte, the total amount of minimum information is barley less than 64 bytes. Ethernet networks would therefore require any station transmitting to transmit for the required 50 microseconds, which is equal to about 64 bytes of data. If the transmitting device had less data than 64 bytes, then it would simply add empty or extra 0s into its data.
Every device in the network is assigned an individual network address based on the protocol being used. Every device also has a physical address called the MAC address and consists of a 48 bit binary identification number, of which half is assigned by the IEEE and the other half comes from the vendor. These addresses are burned into read only memory called ROM and copied into RAM when the card is initialized. MAC address is the only permanent location of a device connected to the network. The network server maps the network address to each devices MAC address.
As electrical signals travel through wire, they gradually decrease in strength and are also susceptible to interference. As these electrical signals move through large amounts of wire, they loose energy from resistance in the wire, interference, and other factors. This is one reason why traditional Ethernet networks are limited in size due to restrictions in the length of copper wire that can be used. The wire itself is constructed to aid in an effect called cancellation. This is where two opposite magnetic fields are created from the flow of current through two opposing wires on the same electrical circuit. These magnetic fields thus cancel themselves out and cancel out any outside magnetic interference. Twisting the wires together further enhances this and provides a self-shield for the wires in the media. CSMA/CD only allows a single device to communicate at any given time. These factors create limitations to the physical size and amount of devices that can be connected to a single strand connected Ethernet network.
To counteract the size limitations, multiple devices have been introduced into Ethernet computer networking. A repeater is a device that simple repeats a given single. It allows networks to span over larger areas by repeating a network messages including collisions over a large area or a different part of a network. A bridge works exactly like a repeater except that can identify the recipient of a network packet and only repeats messages if the message is addressed to one of the devices in the area it’s repeating to. A switch has multiple network connections and is able to switch through different messages to different parts of the network at the same time. It works similar to the old telephone switchboards. It identifies multiple messages from different parts of the network and switches them to the correct recipients.
Ethernet is a simple standard for information sharing between computing devices that operates on the ability for any device to start transmitting and allows for collision detection. Electrical pulses are the medium through which binary information is sent via copper wire. Repeaters, bridges and switches allow for Ethernet networks to become larger than their physical restrictions.
