Ethernet, standardized as IEEE 802.3 by the Institute of Electrical and Electronics Engineers (IEEE), refers to a group or family of frame-based computer networking technologies developed specifically for local area networks (LAN). It was conceptualized in 1974 by Robert Metcalfe and David Boggs at Xerox PARC.

It has defined a common addressing format, as well as a set of wiring and signaling standards for the physical layer, via network access at the Media Access Control (MAC) or Data Link Layer. Its name was derived from the physical concept of the luminiferous ether, which is a medium for the propagation of light.

Introduced in the 1990’s, it has replaced competing LAN standards such as token ring, fiber distributed data interface (FDDI) and Attached Resource Computer Network (ARCNET) to become the most widespread wired LAN technology, in both its twisted pair cable versions for connecting end systems to networks, as well as its fiber optic versions used for network backbones, although at present, it is being either augmented to or superseded by IEEE 802.11, the wireless LAN standard.

The early concept of Ethernet was based on the idea of computer communicating over a shared coaxial cable that would serve as a medium for broadcast transmission, a concept similar to that of radio broadcast systems. However, there are fundamental differences between radio broadcast systems and cable broadcast systems, such as collisions being more easily detected in cable broadcast systems.

Since its early concept, several changes and improvements have been made to Ethernet, such as replacing coaxial cable with point-to-point links using twisted pair cabling connected by hubs or switches in order to reduce installation costs, increase reliability and data transmission rate (from 10 Mbps to 1 Gbps and higher) and allow troubleshooting and point-to-point management.

Each Ethernet station is assigned a 48-bit MAC address in order to specify the destination and source of each data packet sent between Ethernet stations above the physical layer.

Ethernet utilized a scheme called carrier sense multiple access with collision detection (CSMA/CD) to control how computers would share a channel. When a computer wants to transmit information, the following procedure would be followed:
• Readying the frame for transmission;
• Ensuring that the medium is not idle, or otherwise waiting until it becomes ready and then waiting the interframe gap period;
• Beginning transmission;
• Detecting collisions, if any;
• Resetting retransmission counters and ending the frame transmission.

If a collision occurs, the collision detected procedure is activated. It is as follows:
Maintenance of transmission until packet time is reached in order for all receivers to be able to detect the collision; 
• Augmentation of the retransmission counter;
• Detecting if the maximum number of transmission attempts has been reached, in which case transmission will be halted;
• Calculation of random backoff period based on number of collisions, and then waiting for the duration of the period;
• Re-entry of the main procedure.

There are some problems faced by simple Ethernet networks containing switches. These problems include single points of failure which can affect several users if it is in a central location, massive amounts of broadcast traffic which can take up a lot of bandwidth, and bandwidth choke points which occur when a lot of traffic is forced down a single link.

To resolve these issues, switches may offer a variety of tools such as the spanning-tree protocol to maintain the active links of the network as a tree as physical loops for redundancy are enabled, port protection features, Virtual LANS (VLAN) to separate the different classes of users over the same physical infrastructure, and link aggregation to add bandwidth to overloaded links.