SFWR 4C03: Computer Networks & Computer Security

Jan 10-14, 2005

Lecturer: Kartik Krishnan

Lecture 4-6

Ethernet: An introduction
Ethernet is a well known and widely used LAN network technology that employs the bus topology consisting of a single long cable (bus/channel/ether) to which multiple computers attach. It was invented in Xerox Corporation in the early 1970s. This is also the IEEE Standard 802.3. Any computer attached to the bus can send a signal down the cable, and all computers attached to the cable receive a copy of the signal. Every computer can send data to every other computer, thus the Ethernet is an example of a broadcast network. A given Ethernet is limited to 500 meters in length, with a minimum separation of 3 meters between each pair of connections. The cable could be extended using repeaters and bridges; the IEEE standard mandates that no more than four repeaters should be used resulting in a maximum length of 2500 meters. Section 2.4 of Comer [1] covers other hardware details of the Ethernet, and Sections 2.5 and 2.6 cover other network technologies for LANs. We will hereafter assume in class that the underlying network technology is the Ethernet bus topology, unless otherwise specied. It is any case the most widely used. operates 100 times as fat, i.e., at 1 Gbps. In network parlance, bandwidth describes the amount of data a network can transport in a certain period of time. In other words, bandwidth is a capacity for rate of transfer used expressed in bits per second. Ethernet hardware operates at a bandwidth of 10 Million bits per second (Mbps); a newer version known as Fast Ethernet operates 10 times faster, i.e., at 100 Mbps. Recently, there has been another version known as Gigabit Ethernet which operates 100 times as fast, i.e., at 1 Gbps. Since the ethernet is a bus topology network multiple computers share access to a single medium. A sender transmits an ethernet frame in the form of an electric signal that propagates from the sender towards both ends of the cable. During the transmission of the frame, the sending computer has exclusive use of the single cable, i.e., other computers must wait. After the end of the transmission the shared medium becomes available to other computers.

4-6-1

Carrier Sense Multiple Access with Collision Detection

An ethernet network does not have a centralized controller that tells each computer how to take turns using the shared cable. Each computer sends out a frame, and when two frames try to occupy the channel at the same time, there is a collision and both frames will be lost. An important parameter that determines the time to detect a collision is the round-trip propagation delay time (2 ), i.e., the time taken by one bit of the frame to go from one end of the cable to the other end and back. For a 10 Mbps Ethernet with a maximum length of 2500 m this round trip propagation delay time is about 50 sec. In order to eectively communicate over the Ethernet, all attached computers participate in a distributed coordination scheme called Carrier Sense Multiple Access with Collision Detect (CSMA/CD). The scheme uses the electric activity on the cable to determine the status of the channel, and is based on the following set of rules for each computer: 1. If the channel is idle, transmit; otherwise, go to Step 2. 2. If the channel is busy, continue to listen until the channel is idle, then transmit immediately. 3. If a collision is detected detected during transmission, abort transmission, wait for the activity to subside, and then try again after a random amount of time. The randomization in Step 4 is performed using the binary exponential backo algorithm. After a collision, time is divided into discrete slots whose length is equal to the round trip propagation time (2 ). After each collision each station waits 0 or 1 slot times before trying again. If two compute frames collide initially, and each computer picks the same random number, their frames will collide again. After the 2nd collision, each computer picks either 0,1,2, or 3 at random and waits that number of slot times. If a third collision occurs (the probability of this happening is 0.25), then the next time the number of slots to wait is chosen at random from the interval 0 to 23 1. In general, after i collisions, a random number between 0 and 2i 1 is chosen, and that number of slots is skipped. After 16 collisions, each computer simply gives up. Further recovery of the message to be sent is then up to the higher layers in the communication protocol suite.

Ethernet Hardware Address and Frame format

Each computer attached to the Ethernet is assigned a unique 48 bit number called its Ethernet or physical address. Usually, the Ethernet address is xed in machine readable form in the host machine interface hardware. Thus, replacing a hardware interface that has failed changes the machines physical address. The host interface hardware on each computer examines frames and determines the frames to be sent to the host. It uses the destination address eld in the frame as a lter ignoring those that are meant for other machines,and passing to the host only those frames that are meant for it. The host interface

4-6-2

operates independently of the computers central processor, thereby preventing the trac on the Ethernet from slowing down the processing on the host computer. A 48 bit address could also specify more than a single destination computer. For example, the broadcast address (all 1s) is reserved for sending to all computers simultaneously. The data transmitted between two computers attached to the Ethernet is encapsulated in an Ethernet frame. These frames are of variable length with no frame smaller than 64 bytes (octets) or larger than 1518 bytes. The format of a frame is shown in Figure 2.7 of Comer [1]. Each frame contains the following elds in the order specied: 1. Preamble Field: The bits transmitted over the channel are encoded as electric signals using Manchester encoding. With Manchester encoding, each bit period is divided into two equal intervals. A binary 1 bit is sent by having the voltage set high during the rst interval and low in the second one. A binary 0 is just the reverse: rst low and then high. This scheme ensures that every bit period has a transition in the middle, making it easy for the receiver to synchronize with the sender. Every frame starts with a Preamble of 8 bytes, each containing the bit pattern 10101010. The Manchester encoding of this pattern produces a 10 MHz square wave for 6.4 sec to allow the receivers clock to synchronize with the senders. They are required to stay synchronized for the rest of the frame, using the Manchester encoding to keep track of the bit boundaries. 2. Destination Address: It contains the 48 bit address of the intended recipient. 3. Source Address: It contains the 48 bit address of the source computer. 4. Frame Type: It identies the type of data being carried in the frame. The operating system on the machine uses the frame type to nd out which software module should process the frame. 5. Frame Data: The data to be transmitted. A lower bound on the frame data length is 46 bytes, and an upper bound is 1500 bytes. The factors determining these lower and upper bounds are mentioned below. 6. Cyclic Redundancy Check (CRC): The 32 bit CRC helps the interface layer detect errors in transmission: the sender computes the CRC as a function of the data in the frame. The receiver computes the CRC, compares it with the one sentto verify that the frame has been received intact. The upper bound on the frame length was set arbitrarily at 1500 bytes, mostly based on the fact that the network interface hardware requires enough RAM to hold an entire frame and RAM was expensive in the 1970s. Another contention was that all computer have to share the channel, and thus an upper bound on the frame length does not allow a computer to monopolize the entire channel for more than a specied period of time. The reason for having a lower bound on the frame length is more interesting, and it is to prevent a computer from completing the transmission of a short frame before the rst bit of the frame has reached the far end of the cable, where it may collide with another frame. Consider the following problem: At time 0, station A at one end of the network 4-6-3

sends out a frame intended for station B at the other end of this LAN network. Let us call the propagation time for the frame to reach the other end of the cable as (this is one half of the round trip propagation time 2 discussed earlier). Lets say that just before this frame gets to the other end, i.e., at time , the most distant station B starts transmitting. When B detects the leading edge of As frame it aborts its transmission (Step 3 in the CSMA protocol). The frames sent by computers A and B collide and the result is some garbled noise. At about time 2 , station A seems this noise burst and aborts its transmission too. Now, if the station were transmitting a very short frame, it is conceivable that a collision occurs, but the transmission is completed before the acknowledgement (noise burst) gets back at 2 . The station A may then incorrectly conclude that the frame was successfully sent. On a 10 Mbps LAN it takes 107 sec to transmit each bit of the frame. Keeping in mind that 2 = 50 sec, 5000 bits is the smallest frame that is guaranteed to work. To add some margin of safety, this number was rounded to 52 bits or 64 bytes. Frames with fewer than 64 bytes are padded out to 64 bytes with the Pad eld.

Suggested Readings
1. Sections 2.4-2.6 of Comer [1], Chapter 14 of Stallings [2] and Section 4.3 of Tanenbaum [3] for a review of various LAN technologies including Ethernet.

References
[1] D.E. Comer, Internetworking with TCP/IP: Principles, Protocols, and Architectures, 4th edition, Prentice Hall, NJ, 2000. [2] W. Stallings, Data and Computer Communications, 6th edition, Prentice Hall, NJ, 2002. [3] A. Tanenbaum, Computer Networks, 4th edition, Prentice Hall, NJ, 2003.

4-6-4