Data-Link Layers

Note: The slides are adapted from the materials from Prof. Richard Han at CU Boulder and Profs. Jennifer Rexford and Mike Freedman at Princeton University,
and the networking book (Computer Networking: A Top Down Approach) from Kurose and Ross.
Broadcast Links: Shared Media

2
 3

Digital adaptors Communicating

• Link layer implemented in adaptor (network interface card)

– Ethernet card, WiFi card or chip
• Sending side:
– Encapsulates datagram in a frame
– Adds error checking bits, reliable data transfer, flow control, etc.
• Receiving side
– Looks for errors, reliable data transfer, flow control, etc.
– Extracts datagram and passes to upper layer at receiving node
 Link-Layer Services
• Encoding
– Representing the 0s and 1s
• Framing
– Encapsulating packet into frame, adding header, trailer
– Using MAC addresses, rather than IP addresses
• Error detection
– Errors caused by signal attenuation, noise.
– Receiver detecting presence of errors
• Error correction
– Receiver correcting errors without retransmission
• Flow control
– Pacing between adjacent sending and receiving nodes
4
Link-Layer Protocols

5
 Outline
• Link-layer protocols
– Encoding, framing, error detection

• Multiple-access links: sharing is caring!

– Strict isolation: division over time or frequency
– Centralized management (e.g., token passing)
– Decentralized management (e.g., CSMA/CD)

6
 Digital -> Analog Encoding
• Signals sent over physical links
– Source node: bits -> signal
– Receiving node: signal -> bits

• Encoding in telegraph
– Morse code: “long”
and “short” signals

7
 Digital -> Analog Encoding
• Signals sent over physical links
– Source node: bits -> signal
– Receiving node: signal -> bits
• Simplify some electrical engineering details
– Assume two discrete signals, high and low
– E.g., could correspond to two different voltages
• Simple approach: Non-return to zero
– High for a 1, low for a 0

8
 Problem With NRZ
• Long strings of 0s or 1s introduce problems
– No transitions from low-to-high, or high-to-low

• Receiver keeps average of signal it has received

– Uses the average to distinguish between high and low
– Long flat strings make receiver sensitive to small change

• Transitions also necessary for clock recovery

– Receiver uses transitions to derive its own clock
– Long flat strings do not produce any transitions
– Can lead to clock drift at the receiver
• Alternatives (see Section 2.2)
– Non-return to zero inverted, and Manchester encoding9
 Protocols with clock-recovery
• Manchester encoding (basic Ethernet)
– clock XOR NRZ: 802.3: H→L (0), L→H (1) : self-clocking

• Efficiency? (read 4B/5B encoding in Sec 2.2)

– Manchester: 2 clock transitions per bit: 50% efficient
– 1 GbE: 8b/10b: 80% efficient
– 10 GbE: 64b/66b: 96.9% efficient 10
Different Encoding Strategies
 Framing
• Break sequence of bits into a frame
– Typically implemented by the network adaptor
• Sentinel-based
– Delineate frame with special pattern (e.g., 01111110)

01111110 Frame contents 01111110

– Problem: what if special patterns occurs within frame?

– Solution: escaping the special characters (stuffing)
• E.g., sender always inserts a 0 after five 1s
• … and receiver always removes a 0 appearing after five 1s
• Byte stuffing (BiSync, PPP) and bit stuffing (HDLC)
– Similar to escaping special characters in C programs 12
 Framing (Continued)
• Counter-based
– Include the payload length in the header
– … instead of putting a sentinel at the end
– Problem: what if the count field gets corrupted?
• Causes receiver to think the frame ends at a different place
– Solution: catch later when doing error detection
• And wait for the next sentinel for the start of a new frame

• Clock-based
– Make each frame a fixed size
– No ambiguity about start and end of frame
– But, may be wasteful 13
 Character/Byte Stuffing Example
• Sentinel X, at both start and end of packet
• Stuff X in data, replace X with escape character
“E” (DLE in textbook) and X, i.e., X -> (E,X)
• PPP uses byte stuffing for PPP data payload
Data: AKLWXIKKEZLXKDKLEXYBE

Send: XAKLWEXIKKEEZLEXKDKLEEEXYBEEX

Receiver: AKLWXIKKEZLXKDKLEXYBE

14
 Bit Stuffing Example
• Similar to byte stuffing, except bit stuffing is not confined to byte
boundaries
• HDLC denotes beginning and end of a packet/frame with
“01111110” flag
• Since “01111110” may occur anywhere (across byte boundaries)
in data, then “stuff” it:
– At sender, after 5 consecutives one, insert a “0”
– At receiver, “0111110” -> stuffing, so destuff, “01111110” -> end of frame,
“01111111” -> error

Data: 0110111111100111110111111111100000

Send: 0111110011011111011001111100111110111110000000111110

Receiver: 0110111111100111110111111111100000
15
 Error Detection
• Errors are unavoidable
– Electrical interference, thermal noise, etc.

• Error detection
– Transmit extra (redundant) information
– Use redundant information to detect errors
– Extreme case: send two copies of the data
– Trade-off: accuracy vs. overhead

16
 Probability of Packet Error
• Send N bits. Probability of bit error = Pb
• Assume independent bit errors. What is the
probability of packet error?
– Prob[packet error] = Prob[at least 1 bit is corrupt]
= 1- Prob[every bit is clean] = 1-(1-Pb)N
– If Pb= 10-6, and N=10000 bits, then
Prob[packet error] = 9.95 * 10-3 ~= 1%
• Optical links have much lower probabilities of bit
error: 10-12
• Wireless links have much higher probabilities of bit
error: 10-3
– Bit errors correlated, not independent
17
 Error detection
EDC: error detection and correction bits (e.g., redundancy)
D: data protected by error checking, may include header fields

datagram datagram Error detection not 100%

otherwise reliable!
all
bits in D’ N ▪ protocol may miss
OK detected some errors, but rarely
? error
d data bits ▪ larger EDC field yields
D EDC D’ EDC’ better detection and
correction
bit-error prone link
 Parity checking
single bit parity: two-dimensional bit parity:
▪ detect single bit ▪ detect and correct single bit errors
errors
 Internet checksum (review)

Goal: detect errors (i.e., flipped bits) in transmitted segment

sender: receiver:
▪ treat contents of UDP ▪ compute checksum of received
segment (including UDP header segment
fields and IP addresses) as
sequence of 16-bit integers ▪ check if computed checksum equals
▪ checksum: addition (one’s checksum field value:
complement sum) of segment • not equal - error detected
content • equal - no error detected. But maybe
▪ checksum value put into errors nonetheless? More later ….
UDP checksum field
 Cyclic redundancy check
• more powerful error-detection coding
• view data bits, D, as a binary number
• choose r+1 bit pattern (generator), G
• goal: choose r CRC bits, R, such that
– <D,R> exactly divisible by G (modulo 2)
– receiver knows G, divides <D,R> by G. If non-zero remainder:
error detected!
– can detect all burst errors less than r+1 bits
• widely used in practice (Ethernet, 802.11 WiFi, ATM)

Link Layer and LANs 6-21

 CRC example
want:
D.2r XOR R = nG
equivalently:
D.2r = nG XOR R
equivalently:
if we divide D.2r
by G, want
remainder R to
satisfy:
D.2r
R = remainder[ ]
G

Link Layer and LANs 6-22

 Error Correction
• Correct an error, rather than just detecting an error
• Simple technique: Send 2 copies of data with the
data (repetition coding), and then do majority logic
decoding at the receiver
– Original data: 0100 Corrupt
Receiver: 0100
– Send: 0100 0100 0100 0110
Bit
• Problems 0100
– In efficient Decode: 0100
– It cannot correct 2 errors in the same bit

23
 Error Correction (Continued)
• Forward Error Correction (FEC)
– Many types, e.g., Reed-Solomon coding
– Add K bits of redundancy to N bits, to form a (N+K)-bit
long packet, or vector

– N dimensions -> N+K dimensions

– 2N patterns or vectors mapped into 2N+K possibilities
– Spread out these vectors as far away from neighbors as
possible in (N+K)-dimensional space 11 11111
– N=2, K=3, N+K = 5 10 01010
01 10101
00 00000
– Receive 01111 – closet to 11111, so decode “11” and
correct one bit error 24
Sharing the Medium

25
 Collisions

71-65-F7-2B-08-53 1A-2F-BB-76-09-AD

0C-C4-11-6F-E3-98

• Single shared broadcast channel

– Avoid having multiple nodes speaking at once
– Otherwise, collisions lead to garbled data
26
 Multiple Access Protocol
• Single shared broadcast channel
– Avoid having multiple nodes speaking at once
– Otherwise, collisions lead to garbled data

• Multiple access protocol

– Distributed algorithm for sharing the channel
– Algorithm determines which node can transmit
• Classes of techniques
– Channel partitioning: divide channel into pieces
– Taking turns: passing a token for the right to transmit
– Random access: allow collisions, and then recover27
 Channel Partitioning: TDMA
TDMA: time division multiple access
• Access to channel in "rounds"
– Each station gets fixed length slot in each round
• Time-slot length is packet transmission time
– Unused slots go idle
• Example: 6-station LAN with slots 1, 3, and 4

28
 Channel Partitioning: FDMA
FDMA: frequency division multiple access
• Channel spectrum divided into frequency bands
– Each station has fixed frequency band (Wifi channels 1-11)
• Unused transmission time in bands go idle
• Example: 6-station LAN with bands 1, 3, and 4
frequency bands

29
 Channel Partitioning: FDMA
FDMA: frequency division multiple access
• Channel spectrum divided into frequency bands
– Each station has fixed frequency band (Wifi channels 1-11)
• Unused transmission time in bands go idle
• Example: 6-station LAN with bands 1, 3, and 4

WDM: Wavelength division multiplexing

• Multiple wavelengths λ on same optical fiber

30
 “Taking Turns” MAC protocols
Polling Token passing
• Primary node “invites” • Control token passed from
secondary nodes to one node to next sequentially
transmit in turn • Token message
• Concerns: • Concerns:
– Polling overhead – Token overhead
– Latency – Latency
– Single point of failure – Single point of failure (token)
(primary)

31
 Random Access Protocols
• When node has packet to send
– Transmit at full channel data rate R.
– No a priori coordination among nodes

• Two or more transmitting nodes ➜ “collision”

• Random access MAC protocol specifies:
– How to detect collisions
– How to recover from collisions (e.g., via delayed
retransmissions)
• Examples of random access MAC protocols:
– ALOHA, slotted ALOHA
– CSMA, CSMA/CD, CSMA/CA
32
 Key Ideas of Random Access
• Carrier Sense (CS)
– Listen before speaking, and don’t interrupt
– Checking if someone else is already sending data
– … and waiting till the other node is done
• Collision Detection (CD)
– If someone else starts talking at the same time, stop
– Realizing when two nodes are transmitting at once
– …by detecting that the data on the wire is garbled
• Randomness
– Don’t start talking again right away
– Waiting for a random time before trying again
33
 Slotted ALOHA

assumptions: operation:
▪ all frames same size ▪ when node obtains fresh
▪ time divided into equal size frame, transmits in next slot
slots (time to transmit 1 • if no collision: node can send
frame)
new frame in next slot
▪ nodes start to transmit only
slot beginning • if collision: node retransmits
▪ nodes are synchronized frame in each subsequent
slot with probability p until
▪ if 2 or more nodes transmit in success
slot, all nodes detect collision
randomization – why?
 Slotted ALOHA

node 1 1 1 1 1

2 2 2
C: collision
node 2
S: success
3
node 3 3 3 E: empty
C E C S E C E S S

Pros: Cons:
▪ single active node can ▪ collisions, wasting slots
continuously transmit at full rate ▪ idle slots
of channel
▪ nodes may be able to detect collision
▪ highly decentralized: only slots in in less than time to transmit packet
nodes need to be in sync
▪ simple ▪ clock synchronization
 Slotted ALOHA: efficiency
Efficiency: long-run fraction of successful slots (many
nodes, all with many frames to send)
• suppose: N nodes with many frames to send, each
transmits in slot with probability p
– prob that given node has success in a slot = p(1-p)N-1
– prob that any node has a success = Np(1-p)N-1
– max efficiency: find p* that maximizes Np(1-p)N-1
– for many nodes, take limit of Np*(1-p*)N-1 as N goes to infinity, gives:

max efficiency = 1/e = .37

• at best: channel used for useful transmissions

37% of time!
 Pure ALOHA
▪ unslotted Aloha: simpler, no synchronization
• when frame first arrives: transmit immediately
▪ collision probability increases with no synchronization:
• frame sent at t0 collides with other frames sent in [t0-1,t0+1]

will overlap will overlap

with start of with end of
i’s frame i’s frame

t0 - 1 t0 t0 + 1

▪ pure Aloha efficiency: 18% !

 ALOHA vs. Slotted ALOHA

0.37

0.18

G is a measure of the aggregate offered load

38
 CSMA (Carrier Sense Multiple Access)
• Collisions hurt the efficiency of ALOHA protocol
– At best, channel is useful 37% of the time
• ALOHA: transmit before listen
• CSMA: listen before transmit
– If channel sensed idle: transmit entire frame
– If channel sensed busy, defer transmission
• CSMA/CD: CSMA with collision detection
– collisions detected within short time
– colliding transmissions aborted, reducing channel wastage
– collision detection easy in wired, difficult with wireless
– human analogy: the polite conversationalist
39
CSMA (Carrier Sense Multiple Access)
CSMA: Listen before transmit spatial layout of nodes

Collisions can still occur:

propagation delay means
two nodes may not hear
each other’s transmission

Collision: entire packet

transmission time wasted

40
 CSMA/CD Collision Detection
• Detect collision
– Abort transmission
– Jam the link

• Wait random time

– Transmit again

• Hard in wireless
– Must receive data
while transmitting

41
 Ethernet CSMA/CD algorithm

1. NIC receives datagram from network layer, creates frame

2. If NIC senses channel:
if idle: start frame transmission.
if busy: wait until channel idle, then transmit
3. If NIC transmits entire frame without collision, NIC is done with frame !
4. If NIC detects another transmission while sending: abort, send jam signal
5. After aborting, NIC enters binary (exponential) backoff:
• after mth collision, NIC chooses K at random from {0,1,2, …, 2m-1}. NIC waits
K·512 bit times, returns to Step 2
• more collisions: longer backoff interval
 Cable access network: FDM, TDM and
random access!
Internet frames, TV channels, control transmitted
downstream at different frequencies

cable headend

CMTS
…

splitter cable
cable modem
… modem
ISP termination system

▪ multiple downstream (broadcast) FDM channels: up to 1.6 Gbps/channel

▪ single CMTS transmits into channels
▪ multiple upstream channels (up to 1 Gbps/channel)
▪ multiple access: all users contend (random access) for certain
upstream channel time slots; others assigned TDM
 Cable access network:

MAP frame for

Interval [t1, t2]

Downstream channel i
CMTS
Upstream channel j

cable headend
t1 t2 Residences with cable modems

Minislots containing Assigned minislots containing cable modem

minislots request frames upstream data frames

DOCSIS: data over cable service interface specification

▪ FDM over upstream, downstream frequency channels
▪ TDM upstream: some slots assigned, some have contention
• downstream MAP frame: assigns upstream slots
• request for upstream slots (and data) transmitted random
access (binary backoff) in selected slots
 Three Ways to Share the Media
• Channel partitioning MAC protocols:
– Share channel efficiently and fairly at high load
– Inefficient at low load: delay in channel access, 1/N
bandwidth allocated even if only 1 active node!

• “Taking turns” protocols

– Eliminates empty slots without causing collisions
– Vulnerable to failures (e.g., failed node or lost token)

• Random access MAC protocols

– Efficient at low load: single node can fully utilize channel
– High load: collision overhead
45
Comparing the Three Approaches
• Channel partitioning is
(a) Efficient/fair at high load, inefficient at low load
(b) Inefficient at high load, efficient/fair at low load

• “Taking turns”
(a) Inefficient at high load
(b) Efficient at all loads
(c) Robust to failures
• Random access
(a) Inefficient at low load
(b) Efficient at all load
(c) Robust to failures
46