Understanding Ethernet: Standards, Frames, and Protocols

We’re all familiar with Local Area Networks (LANs), which connect multiple workstations in a small area. These LANs usually support around 10 to 100 users, but no more than 1000. They’re commonly used to connect office departments within a building or to link different floors.

Ethernet is the physical layer used in a LAN. Over 85% of LANs use Ethernet due to its speed, cost-effectiveness, and ease of installation. Robert Metcalfe, an engineer at Xerox, invented Ethernet in 1973 to interconnect workstations and printers. The initial system detected collisions when two devices transmitted simultaneously. These collisions were later avoided with the CSMA/CD protocol, which we’ll discuss below.

This Ethernet tutorial covers standards, Ethernet frame formats (DIX, IEEE802.3), the CSMA/CD protocol, Ethernet topologies, test protocols, and more.

Ethernet Standards

Xerox collaborated with DEC and Intel, resulting in the IEEE 802.3 standard. Ethernet standards have evolved to support data speeds from 10 Mbps to 100 Mbps and even 1000 Mbps.

• 10Mbps: Referred to as legacy Ethernet and specified in 802.3.
• 100Mbps: Referred to as Fast Ethernet and specified in 802.3u.
• 1000Mbps: Referred to as Gigabit Ethernet and specified in 802.3z and 802.3ab based on the transmission media.

Here’s a table summarizing Ethernet standards:

Topology

Topology refers to the network’s shape, i.e., how nodes and connecting lines are arranged. The four basic topologies used by LANs are Bus, Ring, Star, and Tree. Ethernet uses these topologies to connect nodes. The connecting medium can be coaxial cable (large or small), twisted pair, fiber optic, or a wireless medium. These different mediums give rise to the various Ethernet versions we see.

CSMA/CD Protocol

This protocol ensures efficient use of the bus between transmitting and receiving nodes. CSMA/CD stands for Carrier Sense Multiple Access with Collision Detection.

Let’s say node A wants to send data to node B on an Ethernet network. The software on node A sends the data to the Network Interface Card (NIC) to build a packet. The packet is stored in the NIC’s RAM, and node A monitors the bus. If no carrier is present, node A transmits the packet.

The data is broadcast to all PCs on the Ethernet bus. Each receiving node compares the destination Ethernet address to its own. If there’s a match, the node accepts the packet; otherwise, it discards it.

Whenever a node transmits, all PCs detect a “carrier,” which is simply data transmitted over the bus at a specific rate according to the Ethernet standard. If a collision is detected when two or more nodes try to transmit simultaneously, the nodes wait for a random period before retransmitting. This entire process is the CSMA/CD protocol.

Ethernet Frames

Ethernet has two main layers as per the OSI model: the physical layer and the data link layer. These layers are involved in forming the Ethernet frame, which packages the data transported between nodes. NICs process these Ethernet frames in both transmitting and receiving devices.

There are two main frame formats: DIX and 802.3. There’s a slight difference between the fields in these frames, as shown below. Therefore, a device using DIX and one using IEEE802.3 need to adapt to each other’s frame formats to communicate.

Fig.1: DIX Ethernet Frame

As shown in the DIX frame, it consists of preamble, destination address, source address, type, data payload, and FCS fields. Source and destination addresses identify the originating and terminating NICs, respectively. Ethernet addresses are 6 bytes each. The preamble is used for synchronization in the receiving Ethernet card. FCS (Frame Check Sequence) is used for error detection and is 4 bytes long. The data payload carries the actual information and ranges from about 46 to 1500 bytes. The “type” field signifies the higher-layer protocol used (e.g., TCP/IP or UDP/IP).

Fig.2: Ethernet frame

As shown in Fig.2, the IEEE 802.3 Ethernet frame includes a preamble, Start Frame Delimiter (SFD), destination and source addresses, length, data, and FCS fields. The SFD marks the start of the frame and usually contains the bit sequence 10101011. The “length” field indicates the number of bytes in the data field. If the data field is less than 46 bytes, padding bytes (with a value of zero) are appended. The key difference between the two frame types is that DIX has a “type” field, while 802.3 has a “length” field in the same position.

Encoding

In Ethernet, digital data is directly transmitted over the cable without modulation. Manchester encoding maps data according to a Non-Return-to-Zero (NRZ) format, using both positive and negative voltage in a single bit. A ‘1’ bit is represented by a positive pulse followed by a negative pulse, and a ‘0’ bit is represented by a negative pulse followed by a positive pulse. These signals are AC, resulting in zero voltage on the Ethernet cable. The positive-to-negative and negative-to-positive transitions at the center help derive the clock signal directly from the data, enabling synchronous communication without a separate clock signal. Most Ethernet implementations use a DC version of Manchester encoding.

Ethernet Network

Fig.3: Ethernet Network

Fig.3 shows a simple Ethernet network connecting workstations and a server using Ethernet hardware devices like switches, hubs, routers, and gateways.

Ethernet Protocols

The following protocols are useful for Ethernet testing. ARP and RARP are already part of the Internet protocol suite in most Ethernet devices.

• ARP (Address Resolution Protocol): Converts IP addresses (logical addresses) to physical (Ethernet) addresses.
• RARP (Reverse Address Resolution Protocol): Converts Ethernet addresses to IP addresses.
• Ping: Tests the path between a source and destination node. It sends packets from the source to the destination and back. If all bytes are received without packet loss, the path is clear. A successful ping test confirms the ability to transmit information over the Ethernet network.

Ethernet Test Products and Services

Several companies provide Ethernet test solutions and services, including IXIACOM, EXFO, Agilent Technologies, Spirent, and Tektronix. They offer test solutions for 10Mbps (10BaseT), 100Mbps (100BaseTX), 1Gbps (1000BaseT), and other Ethernet standards.

These companies manufacture Ethernet test solutions for product testing across various transmission mediums like copper, twisted pair, and fiber optic. Key test parameters include throughput (maximum bandwidth), round trip latency, lost frames, and back-to-back frames, as defined in RFC2544. Ideally, there should be zero lost frames. Back-to-back frame tests determine the maximum buffering capacity of the Ethernet.

Networking is the term used for a network of computers or internet-enabled devices connected together to form a data communication network. This network is used to transfer data from one device to another with the use of TCP/IP or UDP/IP packets. TCP (Transmission Control Protocol) is mainly used for connection-oriented networks, while UDP (User Datagram Protocol) is primarily used for connectionless networks.