the data link layer goal –as reliable as possible, efficient communication point-to-point...

Post on 03-Jan-2016

216 Views

Category:

Documents

2 Downloads

Preview:

Click to see full reader

TRANSCRIPT

The Data Link Layer

• Goal– As reliable as possible, efficient communication

• Point-to-Point– single connection– bits arrive in order sent

• Not necessarily reliable (unreliable)

Data Link Layer Problems

• Errors occur

• Finite data rate

• Propagation delays– Time required for signal to travel from source to

destination

Data Link Functions

• Well-defined interface to the network layer

• Grouping the bits of the physical layer into frames

• Handling transmission errors

• Regulating the rate of data flow

Services Provided to the Data Link Layer

• Unacknowledged connectionless

• Acknowledged connectionless

• Acknowledged connection-oriented

Unacknowledged Connectionless• Frames are independent

• No acknowledgment

• No connection

• Lost frames are not detected– may be detected in higher level layers

• Best when error rates are low

• Good when late data is not useful– speech, video

• Most LANs use this service

Acknowledged Connectionless(More Reliable)

• Each frame individually acknowledged

• Unacknowledged frames are eventually resent

• Good for unreliable channels– wireless for example

Acknowledged Connection-oriented Service

• Connection is established

• Each frame is numbered

• Every frame sent is guaranteed to be received eventually

• Frames guaranteed to arrive in correct order

• Provides a reliable bit stream

Framing

• Bit streams from physical layer are unreliable

• We would like to check for errors

• Sending a relatively small frame of bits, combined with some kind of data redundancy, allows us to provide some error checking

Methods that Break the Bitstream into Frames

• Character count

• Starting and ending characters with character stuffing

• Starting and ending flags with bit stuffing

• Physical layer coding violations

Big Problem:– How do we recover if we get out of sync?

Character Count

5 T h i s 9 M e s s a g e 2 S

Frames are prefixed by chars/frame

ASCII 5 84

count count countWhat happens if

we lose this character?

104 115101115105 9 77 2 83101115 10397 32

Start and End Characters with Character Stuffing

• Special ASCII characters signal start and end of frame

DLE Data Link Escape (ASCII = 16)

STX Start of text (ASCII = 2)

DLE Data Link Escape (ASCII = 16)

ETX End of text (ASCII = 3)

• But what if these flags occur in the data?– Use two DLE characters to send one data DLE

Examples of Char Stuffing

DLE STX T H I S M DLE ETX

X Y DLE STX DLE ETX 5

DLE STX X Y DLE DLE STX DLE DLE ETX 5 DLE ETX

T S H I M

Start and End Flags With Bit Stuffing

• Choose a bit sequence for start/end flag

• Example: 01111110Every time you see five 1’s in sequence in the

data, stuff a zero into the stream

Final Framing Method - Physical Layer Coding Violations

• Start/End flag consists of sequence that is illegal in the data

• Example:10 is 1

01 is 0

00 or 11 could be used as flags

Error Control

• Error Control provides feedback about the success of data transmission– frames may arrive correctly (receiver sends ack)– frames may arrive corrupted (receiver sends

negative ack)– frames may be lost (no ack is ever received)

• acks may be lost (no ack is ever received)

• No error control in unacknowledged connectionless service

Flow Control

• What happens if the sender can send data faster than the receiver can handle it?– Receiver buffers overflow– Data is lost or requires retransmission

• Solution: Flow control– Receiver specifies the number of frames that

may be transmitted (no more than can be held in its buffers). The sender cannot send more data until the receiver grants permission.

Error Detection and Correction

• Error rates are relatively low on digital networks and LAN’s

• Error rates are high on analog twisted pairs and wireless communication

• When errors occur, they tend to occur in bursts– More data blocks are correct– Errors are harder to detect

Definitions

• Error-correcting Codes – Codes that contain enough redundant

information to correct errors

• Error-detecting Codes– Codes that contain enough redundant

information to detect errors

Parity - a simple error detection mechanism

• The number of “1” bits in a fixed length bit sequence is either always even (even parity) or always odd (odd parity)

• Example (even parity, length = 8)

parity bit0101011001100101

0101111001000101 Any single

bit errors are detected

1101011101001101

Any even number of bit errors is not detected

1101011101001111

Any odd number of bit errors is

detected

Hamming Distance

• Frame codeword contains n = m + r total bits– where

• m is number of message bits

• r is number of redundant bits

• Hamming Distance– The number of bit positions in which two

codewords differ

• Hamming Distance of Complete Code– The minimum Hamming Distance between

codewords

01101000110 11010010100

Hamming distance = 6

Hamming Distance

• To detect d errors you need a Hamming distance (for the complete code) of d+1

• To correct d errors you need a Hamming distance (for the complete code) of 2d+1

Error Correcting Codes

• Consider an encoding scheme where these are the only valid codewords

0000000000 0000011111 1111100000 1111111111

• The Hamming distance of the complete code is 5

• We can correct d-bit errors where 2d+1 = 5

• We can detect d-bit errors where d+1 = 50000000000 with two errors 0000000101 (with < 3 errors must be 0)

0000000000 with three errors 0000011001 could be 0000000000 with three errors or 0000011111 with two errors.

top related