Ethernet

Ethernet is the de facto standard for local area networks (LAN). Figure 1 shows the ethernet frame structure according to IEEE 802.3. Every frame starts with a 7 byte preamble, having the bit sequence 101010. The manchester coding of this bit sequence generates a 5.6 us lasting 10MHz oscillation. This is used to enable the receiver to synchronize. The following frame delimiter 10101011 marks the beginning of the frame. The frame contains a destination address and a source address. The standard mentions 2 byte and 6 byte address schemes, but the 10 Mbit/sec ethernet version uses the 6 byte (48 bit) address scheme only. The MSB (bit 47) of the address is used for multicast and broadcast services, bit 46 is used to distinguish between local and global addresses. Local addresses are assigned by the network operator and have no meaning outside that particular network in question. Global addresses are given by IEEE in order to ensure that they are unique in the whole world. The address range of the usable 48-2=46 bits provides about 7x10^13 unique addresses. Every ethernet station in the world can address every other ethernet station with its unique ethernet address. The routing of ethernet frames between subnetworks is done by network layer protocol. In the TCP/IP protocol world, the address resolution protocol (ARP) does the mapping between IP and ethernet address. In the OSI world, the ethernet address may be part of the NSAP.

The data field of the frame may in theory be of length zero. In order to avoid unnoticeable frame collisions on a LAN, the padding field ensures that a valid ethernet frame has a length of at least 64 bytes, counted from the destination address field to the checksum field. The last field of the ethernet frame contains the 32 bit checksum for the cyclic redundancy error detection algorithm.

Figure 1 - Ethernet Frame Structure according to IEEE 802.3

For more information on ethernet check this site.