Multiple Access&

Local Area Networks

Prof. A. Sahoo

KReSIT

2

Data LinkLayer

802.3CSMA-CD

802.5Token Ring

802.2 Logical Link Control

PhysicalLayer

MAC

LLC

802.11Wireless

LAN

Network Layer

Network Layer

PhysicalLayer

OSIIEEE 802

Various Physical Layers

OtherLANs

3

12

34

5M

Shared MultipleAccess Medium

Static Channel Allocation Problem Access Model:

Can be modeled as n independent users (one per node), each wanting to communicate with another user and they have no other form of communication.

Channel Allocation ProblemTo manage a single broadcast channel which must be shared among n uncoordinated users in following manner• efficiently i.e. maximize message throughputmaximize message throughput and • fairly and,• minimize mean waiting time mean waiting time

4

• Ring networks

• Multitapped Bus

5 main contexts:1. Wired Local Area

network

1. Wireless Local Area network

1. Packet Radio network

1. Cellular telephony

1. Satellite Communication

Context of Multiple Access Problem

5

Satellite Channel = fin

= fout

6

Possible Model Assumptions for Channel Allocation Problem

0. Listen property :: (applies to satellites)

The sender is able to listen to sent frame one round-trip after sending it.

no need for explicit ACKs

1. Model consists of n independent stations.

2. A single channel is available for communications.

7

Possible Model Assumptions for Channel Allocation Problem

3. Collision Assumption :: If two frames are transmitted simultaneously, they overlap in time and the resulting signal is garbled. This event is a collision.

4a. Continuous Time Assumption :: frame transmissions can begin at any time instant.

4b. Slotted Time Assumption :: time is divided into discrete intervals (slots). Frame transmissions always begin at the start of a time slot.

8

Possible Model Assumptions for Channel Allocation Problem

5a. Carrier Sense Assumption ::

Stations can tell if the channel is busy (in use) before trying to use it. If the channel is busy, no station will attempt to use the channel until it is idle.

5b. No Carrier Sense Assumption ::

Stations are unable to sense channel before attempting to send a frame. They just go ahead and transmit a frame.

9

Centralized versus Distributed

Two fundamental design choice Centralized : (GSM, CDMA)

One station is a master station Other stations are slaves Master decides when a slave can send data

Distributed: (Wireless LAN,Ethernet) No master station every other station is free to talk to any other

station

10

Circuit versus Packet mode

Circuit mode for smooth continuous traffic (voice) Makes sense to allocate part of the link to to the source for its exclusive use,

avoiding link access protocol repeatedly---> circuit mode(GSM)

Packet mode for bursty data traffic Fixed allocation of resource may cause under-utilization Compete for link access for each packet transmission ( GPRS)

11

Constraints

Design of multiple access scheme is highly constrained by implementation environment

Spectrum Scarcity spectrum is scarce resource FCC allows 902-928 Mhz, 2.40-2.48 Ghz for ISM band

for 1- 10 mile with restriction in transmission power Radio link properties

Fading, multi-path interference, hidden and exposed node problem, near-far problem

12

Constraints

Performance of ‘packet mode’ multiple access scheme heavily depends upon

a = D/TD= maximum propagation delay between any two stations (in seconds)T= Time to transmit any average size packets (seconds)

‘a’ is the number of packets that a station can put in the medium before the first bit is received by the receiver. .01 (wired, wireless LAN, cellular packet radio), 100 satellite

Impact of ‘a’ on collision recover scheme: A determines what happens when two senders transmit simultaneously

With small a packet collide soon and senders can know soon by listening to the medium

With large ‘a’, collision takes longer time after packet is transmitted. So only sensing medium is not sufficient to recover from collision. Some more protocol structure is required.

13

The parameter ‘a’

The number of packets sent by a source before the farthest station receives the first bit

14

15

Performance Metrics

Normalized throughput or Goodput Fraction of link capacity devoted to carrying non-retransmitted packet. E.g capacity 1 mbps, packet size 125 bytes, then it should carry 1000 packets

per second. But due to protocol overhead carries only 250 packets/sec. Goodput= .25. Realistic MA schemes --> .1 to .9

Mean Delay Duration a packet has to wait before it gets transmitted successfully Depends on --load, -- MA scheme, -- media characteristic

16

Contd..

Stability When offered load is low medium/ MA scheme can carry the load When load increases, collision increases and systen throughput tend to

decrease. Almost every transmission is a collision. Stable system --> Throughput does not decrease when load increases beyond

certain threshold Proper MA scheme can achieve this by dynamically decreasing the load

when overload is detected if infinite number of uncontrolled stations share a link, then instability is

guaranteed but if sources reduce load when overload is detected, can achieve stability

Fairness No starvation Equal share of bandwidth. max-min fair share: will study later

17

Carrier Sense Multiple Access

Listen before you speak Check whether the medium is active before sending a

packet (i.e carrier sensing) If medium idle, then transmit If collision happens, then detect and resolve

18

CSMA: Distributed, Packet mode scheme

Carrier Sense and its variants: Use of carrier sensing capability to know if someone else is using

the medium 1 persistent

If medium busy, keep sensing If medium Idle send immediately

p persistent If medium busy, keep sensing If medium Idle,

send with probability p, in case of no-send (1-p), wait for 1 time slot, and begin medium

sensing again Non persistent

If medium busy, wait for random time before sensing again If medium Idle send immediately

19

Choice of p in p-persistent system

Time slot is usually set to the maximum propagation delay.

as p decreases, stations wait longer to transmit, but the number of collisions decreases

if np > 1: secondary transmission likely So p < 1/n Large n needs small p which causes delay

20

Collision detection (CSMA/CD)

All aforementioned scheme can suffer from collision Device can detect collision

Listen while transmitting Wait for 2 * propagation delay

21

How to avoid/recover collision

1 persistent and non persistent results in guaranteed collisions when two node decided to transmit simultaneously

Rescue: p persistence or Exponential Back off P persistent:

p= .5, .1, .8 reduces the chance of collision Choice of p:

Trade-off between performance under heavy load and mean packet delay Under heavy load mean number of stations that will send packet is np, if

np > 1 then collision is likely. Therefore choose p < 1/n ( dynamic adjustment??)

Under heavy load mean message delay increase in order to ensure stable throughput.

22

Contd..

Advantages: Dynamically adjusts p, no need to choose p optimally. In high load, backs -off drastically, and reduces load on network

to give stable throughput

Disadvantages Delay increases ?? Fairness??

23

Exponential backoff

On detecting 1st collision for packet x

station A chooses a number r between 0 and 1.

waits for r * slot time and transmit.

Slot time is 2 * propagation delay

On detecting kth collision for packet x

choose r between 0,1,..,(2k –1) When value of k becomes high (10), give up. Randomization increase with larger window, but

delay increases.

24

CSMA/ CD (802.3,Ethernet)Performance

understanding the Ethernet's distributed contention scheme, under high load

transmission interval: is that during which the Ether has been acquired for a successful

packet transmission. contention interval:

is that composed of the retransmission slots Idle interval

No activity

25

CSMA contention interval

Station A transmits. Just before its signal reaches B, station B senses channel is idle and starts transmitting resulting in collision

The longer the propagation delay, the worse the performance of the protocol

26

Contention Interval - 2D

A B

A B

A B

A B

t = 0: A begins transmission

t = D - : packet almost at BB begins transmission

t = D: B detects collision, stops transmitting

t = 2D - : A detects collision

27

CSMA/ CD (802.3,Ethernet)Performance

Metcalfe and Boggs approach:Try to find mean duration of contention intervalOrmean number of contention slots in a contention interval k node is contending in contention period Assume each node transmits with a fixed probability p in any slot What is the probability (‘A’) that some station acquires the channel

in a slot ? A= p(1-p)^k-1 + p(1-p)^k-1 +… k times

= k p(1-p)^k-1 (A is maximized when p =1/k, with A -> 1/e as k tends to infinity.

28

Contd..

what is the probability (Pj) that a contention interval has exactly j slots?

(i.e after the collision, the least size of back off interval chosen by some node is j slots or, what is the probability that some station transmits only at j th slot and not in previous j-1 no slots?)

p(Not transmitted in 1 st slot)= ( 1-A) p(Not transmitted in 1 st slot and 2nd slot)= ( 1-A)(1-A)

Pj= A(1-A) ^j-1 -----(1)

Over a long period of time, the mean number of slots per contention interval is given by

Sumj=0j=inf (j* A(1-A)j-1 ) = 1/A (= mean of geometric

distribution given in (1) )

29

Contd…

Since each slot is 2*Tprop, Mean length of contention interval (w) is

2* Tprop /A Now if Ttrans is time needed for mean packet size then

Channel efficiency (w) is = time required to transmit in absence of collision/

time required to transmit in presence of collision

w= Ttrans / (Ttrans + 2* Tprop /A) So longer the propagation time Tprop( cable length!), the longer the

contention interval. The longer the contention interval, the lesser the

efficiency/throughput

30

Example: Ethernet (IEEE 802.3)

CSMA/CD with jamming Ethernet Address (48 bits)

Example: 08:00:0D:01:74:71

Ethernet Frame Format SFD : Start of frame delimiter (1 bit) FCS : Frame Check Sequence (checksum) L : length of the Data field

31

Minimum frame size

A B

A B

A B

A B

t = 0: A begins transmission

t = D - : packet almost at BB begins transmission

t = 1: B detects collision, stops transmitting

t = 2D - : A detects collision

32

Minimum frame size

It takes A a complete RTT (2D) to detect collision When B detects collision (gets more power than it is

putting out) it generates 48-bit noise burst (“Jam” bits) to warn all other stations

Min. frame size equal to number of bits transmitted during one RTT

33

Minimum frame size

slotTime: number of bits transmitted by a source during the max. RTT (2D = 51.2 sec) for any Ethernet network

Collisions must be detected by sources while still transmitting

All frames must be at least 1 slot (on 10Mbps, this is 512 bits)

34

Ethernet The most widely used LAN Standard is called IEEE 802.3 Uses CSMA/CD with exponential backoff Also, on collision, place a jam signal on wire, so that all

stations are aware of collision and can increment timeout range

‘a’ small =>time wasted in collision is around 50 microseconds

Ethernet requires packet to be long enough that a collision is detected before packet transmission completes (a <= 1) packet should be at least 64 bytes long for longest allowed

segment Max packet size is 1500 bytes

prevents hogging by a single station

35

More on Ethernet

Ethernet types are coded as <Speed><Baseband or broadband><physical medium> Speed = 3, 10, 100 Mbps Physical medium:

“2” is cheap 50 Ohm cable, upto 200 meters “T” is unshielded twisted pair (also used for telephone

wiring) “36” is 75 Ohm cable TV cable, upto 3600 meters

10base5: 500m segment, 100 node, Original, bus topology 10base 2: 185m segment, 30 node, cheap 10baseT: twisted pair; goes to a hub star topology 10baseF : Optical Fiber Ethernet

36

Recent developments

Switched Ethernet each station is connected to switch by a separate UTP wire line card of switch has a buffer to hold incoming packets fast backplane switches packet from one line card to others simultaneously arriving packets do not collide (until buffers

overflow) higher intrinsic capacity than 10BaseT (and more expensive)

37

Fast Ethernet variants

Fast Ethernet (IEEE 802.3u) same as 10BaseT, except that line speed is 100 Mbps spans only 205 m big winner most current cards support both 10 and 100 Mbps cards (10/100

cards) for about $80 ( Old data, probably cheaper now) 100VG Anylan (IEEE 802.12)

station makes explicit service requests to master master schedules requests, eliminating collisions not a success in the market

38

Fast Ethernet variants

Gigabit Ethernet aims to continue the trend still undefined, but first implementation will be based on fiber

links

39

Gigabit Ether Switch

40

KReSIT internet connection

41

Evaluating Ethernet

Pros easy to setup requires no configuration robust to noise

Problems at heavy loads, users see large delays because of backoff nondeterministic service doesn’t support priorities big overhead on small packets

But, very successful because problems only at high load can segment LANs to reduce load

42

Centralized access schemes

One station is master, and the other are slaves slave can transmit only when master allows

Natural fit in some situations wireless LAN (Point coordination function mode, PCF), where

base station is the only station that can see everyone cellular telephony, where base station is the only one capable of

high transmit power Pros

simple master provides single point of coordination

Cons master is a single point of failure

need a re-election protocol master is involved in every single transfer => added delay

43

Polling and probing

Centralized packet-mode multiple access schemes Polling (PCF mode of wireless LAN )

master asks each station in turn if it wants to send (roll-call polling)

inefficient if only a few stations are active, overhead for polling messages is high, or system has many terminals

Probing Putting some intelligence to simple polling stations are numbered with consecutive logical addresses assume station can listen both to its own address and to a set of

multicast addresses master does a binary search to locate next active station

44

Reservation-based schemes

When ‘a’ is large, can’t use a distributed scheme for packet mode (too many collisions and waste of bandwidth) mainly for satellite links

Instead master coordinates access to link using reservations

Some time slots devoted to reservation messages can be smaller than data slots => minislots

Stations contend for a minislot (or own one) Master decides winners and grants them access to link Packet collisions are only for minislots, so overhead on

contention is reduced

45

Reservation Based: Reservation Based: Basic Bit-Map ProtocolBasic Bit-Map Protocol((Collision-Free ProtocolsCollision-Free Protocols))

N stations with addresses 0 to N-1 N one-bit contention slots. If a station i has a frame to send, it sends a one

during contention slot i. Once all stations indicated frame availability, ready

frames are transmitted in address order.

N = 8

46

Bit-Map ProtocolBit-Map Protocol

Representative of reservation protocols (where each station broadcasts its desire to transmit before actual transmission).

Efficiency per frame: With low load = d/(N + d) With high load = d/(1 +

d)

d = number of bits in one frame

47

Bit Map Protocol

Issues Stations' access to the network is unfair: That is, if station

i and station j both want to transmit, and i < j, then station i always first to transmit.

low numbered stations have to wait longer than high numbered stations for the reservation to complete.

Efficiency: at low load, the protocol efficiency is low.

48

Limited Contention Protocol

None of full reservation and full contention mechanisms are not suitable for extreme condition of load.

Reservation scheme is appropriate for high load situation, contention for low load situation

A mix of reservation and contention may be an adaptive and optimal approach

We have seen that probability of successful transmission by some node ( among n contending nodes) is maximized when p <1/n. Decreases drastically with increase of n

Heavily dependent on n or size of contention domain Protocol which can make smaller subgroup of nodes ( and thus

limiting the contention) and then give chance to each group-- Splitting Algorithms

49

Limited contention - Adaptive Tree Walk

Combines best properties of contention and collision free protocols. Divides stations into groups. In first contention slot, everyone is allowed to compete. If collision, then only those in group 1 are allowed to compete, and so on

down the hierarchy.

AA CCBB DD

44

3322

11

HHGGFFEE

776655

50

Decentralized polling

Just like centralized polling, except there is no master Each station is assigned a slot that it uses

if nothing to send, slot is wasted Also, all stations must share a time base

51

Token passing

In distributed polling, every station has to wait for its turn

Time wasted because idle stations are still given a slot What if we can quickly skip past idle stations? This is the key idea of token ring Special packet called ‘token’ gives station the right to

transmit data When done, it passes token to ‘next’ station

=> stations form a logical ring No station will starve

52

Token Ring operation

Frames flow in one direction: upstream to downstream special bit pattern (token) rotates around ring

The ring should be able to hold the token IEEE 802.5---> 1 bit holding buffer

must capture token before transmitting During normal operation, copy packets ( token and data) from input

buffer to output If packet is a token, check if data packets ready to send If not, forward token If so, delete token, and send packets Receiver copies packet and sets ‘ack’ flag Sender removes packet and deletes it When done, reinserts token If ring idle and no token for a long time, regenerate token

53

Logical rings

Can be on a non-ring physical topology

54

Ethernet vs. Token Ring: Ethernet

Dominance

Open standard Proprietary

platforms “forced” to support standards or lose value

FDDI Market $220M 1997, $40M 2001.

Fast Ethernet $150/port

FDDI $750/port

55

Various Token ring networks

Token Ring Networks PRONET: 10Mbps and 80 Mbps rings IBM: 4Mbps token ring 16Mbps IEEE 802.5/token ring 100Mbps Fiber Distributed Data Interface (FDDI)

56

Evaluating token ring Token Holding Time (THT)

10 msec upper bound (ieee 802.5) Token Rotation Time <= ActiveNodes * THT + Ring Latency (RL)

RL significantly large for MANs Pros

medium access protocol is simple and explicit no need for carrier sensing, time synchronization or complex protocols

to resolve contention guarantees zero collisions can give some stations priority over others ( see later) At high loads can effectively become TDM

57

Evaluating token ring

Cons token is a single point of failure

lost or corrupted token trashes network need to carefully protect and, if necessary, regenerate token

all stations must cooperate network must detect and cut off unresponsive stations

stations must actively monitor network ( see next slide) usually elect one station as monitor

58

Token Ring Maintenance Need for a monitor ( increased complexity)

Token may be lost, corrupted Data frame may be orphaned

Monitor sends periodic beacon Monitor election

In absence of beacon, any node transmits ‘claim token’ control packet. The node which can do it first, and seen by everybody, becomes the monitor.

Lost Token: Monitor waits for TimeOut = NoOfStations * THT + RL If no token, insert token

59

Token Ring Maintenance

Garbled frame Check for checksum, and remove

Orphan frame Delete from the ring Use of monitor bit ( if set and seen twice by Monitor, delete it)

Maintaining the length of ring If less than token size, insert delay

60

Priority Operation in Token Ring Used for time critical application

Provides a mechanism that higher priority frames across the nodes will be transmitted first

Same priority frames will have same access right to the ring Priorities are for traffic classes. Token Contains 3 bit priority field. 6 priority level possible

Operation: Token has a certain priority n initially If received packet is token:

A node X, having a packet with priority p, seizes the token if token priority is n <= p

transmits data and, and passes token with its previous priority value n Else passes on the token

In case X received packet is a data and n <= p( reservation) Set priority p and reservation bit Remembers (stacks) the old and new value of p ( Po, Pn) Token holder S, when releasing the token,

sets this p ( received in data frame) in the token ( this ensures that low priority nodes will not be able to seize the token)

61

Priority ( contd..)

In token priority value is not lowered, low priority packet may starve

The node which raised the priority value in the token is responsible for lowering the value ( here X, stacking station, old priority Po, new priority Pn) This is done when X, sees the token coming back with a higher

than/same value as Pn , indicating all higher priority packets are serviced, lowers the ring service value to Po.

Eventually lowest priority packets get serviced. May be extended delay, due to high load of higher

priority pakets

62

Single and double rings

With a single ring, a single failure of a link or station breaks the network => fragile

With a double ring, on a failure, go into wrap mode Used in FDDI

63

Hub or star-ring

Simplifies wiring Active hub is predecessor and successor to every station

can monitor ring for station and link failures Passive hub only serves as wiring concentrator

but provides a single test point Because of these benefits, hubs are practically the only

form of wiring used in real networks even for Ethernet

64

Fiber Distributed Data Interface

FDDI is the most popular token-ring base LAN Dual counterrotating rings, each at 100 Mbps Uses both copper and fiber links Supports both non-realtime and realtime traffic

token is guaranteed to rotate once every Target Token Rotation Time (TTRT) ( see later)

station is guaranteed a synchronous allocation within every TTRT

Supports both single attached and dual attached stations single attached (cheaper) stations are connected to only one of

the rings

65

Physical Properties of FDDI

Dual Ring Configuration

Single and Dual Attachment Stations

(a) (b)

Downstream

Neighbor

UpstreamNeighbor

Concentrator

SAS

SAS

SAS

SAS

SAS

66

Characteristics

Each station imposes a delay (e.g., 50ns) Maximum of 500 stations Upper limit of 100km (200km of fiber) Can be implemented over copper (CDDI) FDDI uses 4b/5b NRZI (Non-Return to Zero Invert

on ones) with 100 Mb/s data rate 10BaseT Ethernet uses Manchester encoding , 10

Mb/s data rate

67

Timed Token Protocol

Target Token Rotation Time (TTRT): agreed-upon upper bound on TRT.

There are 2 types of traffic for FDDI: Asynchronous transmission

Given lower priority Synchronous transmission

is given higher priority, needs upper bound on delay. Sum of all symmetric data transmission time <

TTRT Token Rotation Time (TRT): how long it takes the

token to traverse the ring.TRT <= ActiveNodes x THT + RingLatency

68

Algorithm each node measures TRT between successive arrivals of the token if measured TRT >= TTRT, then token is late so don't send non-

real time ( asynchronous ) data if measured TRT < TTRT, then token is early so OK to send data Always send synchronous data, which has fixed size. define two classes of traffic

synchronous data: can always send. Has fixed size. asynchronous data: can send only if token is early

Example: TRT =100 ms ( I.e. since last time the node has seen the token),

TTRT =200 ms. Suppose node X has 20 ms synchronous data. There for it can send non real time data for 200 - 120 ms. If it consumes 80 ms, then next node do not have any time to send

asynchronous data.

69

Token Maintenance Lost Token

no token when initializing ring bit error corrupts token pattern node holding token crashes

Generating a Token (and agreeing on TTRT) execute when join ring or suspect a failure each node sends a special claim frame that includes the

node's bid for the TTRT when receive claim frame, update bid and forward if your claim frame makes it all the way around the ring:

your bid was the lowest everyone knows TTRT you insert new token

70

Monitoring for a Valid Token should see valid transmission (frame or token) periodically maximum gap = ring latency + max frame <= 2.5ms set timer at 2.5ms and send claim frame if it fires

71

Ethernet vs. Token Ring: Media Access Control Methods

Contention (Ethernet) performs better than token passing on low utilization

LANs high utilization - collisions and retransmission when 2

stations try to communicate simultaneously Token passing

high utilization - superior performance, no collisions QoS – multimedia preference to some applications used to control bus in USB, Firewire, and other

emerging shared media technologies

72

Comparisons

Each station imposes a delay (e.g., 50ns) Maximum of 500 stations Upper limit of 100km (200km of fiber) Can be implemented over copper (CDDI) FDDI uses 4b/5b NRZI (Non-Return to Zero Invert

on ones) with 100 Mb/s data rate 10BaseT Ethernet :10 Mb/s data rate, no of station,

reach: much lower

73

Ethernet vs. Token Ring:

Response Time vs. Load

74

ALOHA (“free for all”)

Stations transmit whenever they have data to send

Detect Collision or Wait for an acknowledgment

If no acknowledgment (or collision), try again after a random waiting time

Collision: If more than one node transmit at the same timeIf there is a collision, all nodes have to re-transmit packets

75

Vulnerable Window

For a given frame, the time when no other frame may be transmitted if a collision is to be avoided.

Assume all packets have same length (L) and require Tp seconds

for transmission Each packet vulnerable to collisions for time Vp = ??

Suppose packet A sent at time to

If pkt B sent any time between to – Tp and to end of packet B

collides with beginning of packet A If pkt C sent any time between to and to + Tp start of packet

C will collide with end of packet A Total vulnerable interval for packet A is 2Tp

t

Packet CPacket B

Tp

Packet ATp

76

Throughput of Pure ALOHA

Based on several assumptions:

1. Traffic Model: transmission attempts follows a Poisson distribution

2. Fixed packet size Poisson Distribution:

Pr[i customers arrive in t time interval] = P[n(t) =i] = (t i e -t)/i!

Where Average arrival rate ( no of arrival per unit tine)

Assume, G = Average number of transmissions in time interval T= T, where T =frame time Probability of i transmission attempts per frame time is

poisson with mean G per frame time Probability that i frames are generated during a frame time

is P[i] = Gi * e-G / i!

77

ALOHA Throughput Let S = number of successful packet transmissions per frame

time (equals channel utilization)

G (= in previous expression )= average number of attempted transmissions per packet time(user load+retransmissions).Then,with Poisson distribution traffic model

Probability that i frames are generated during time interval 2T (vulnerable window) is

P[n(2T) =i] = (G i e –G)/i!

(In an interval two frame times long, the mean number of frames generated will be 2G)

78

ALOHA Throughput

Now Throughput (S) is = offered load G * Prob that no frame suffers collision

S= G * Pr( that there is no transmission in 2T time interval

i.e Pr( n(2T)=0 ) )

Pr( n(2T)=0 )= e - 2G

S= G* e - 2G

79

ALOHA and Slotted ALOHAThroughput versus Load

Peaks at G=.5 ---> max(Throughput) = 1/2e ~ 0.18 ALOHA can achieve maximum throughput of 18.4%

dS/dG = e-2G – 2Ge-2G = 0

Gmax = 1/2 Smax = 1/(2e) = 0.184

0

0.05

0.1

0.15

0.2

0.25

0.3

0.35

0.4

0.01

563

0.03

125

0.06

25

0.12

5

0.25 0.5 1 2 4 8

Ge-G

Ge-2G

G

0.184

0.368

S

80

Slotted ALOHA

Time is divided into slots (i.e., slot = one packet transmission time at least) Master station generates synchronization pulses for time-slots. (e.g., use

“pip” from a satellite) Station waits till beginning of slot to send packet. Stations transmit ONLY at the beginning of a time slot Collisions will occur because more than one frame can send in the slot n But collision probability reduces as Vulnerability Window reduced from 2T

to T;

81

Slotted ALOHA

Goodput doubles. Average no of packet in T (Vulnerable window) time=

P[n(T) =i] = (G i e –G)/i!

So S= G * Pr( that there is no transmission in 2T time interval

i.e Pr( n(T)=0 ) )

Pr( n(T)=0 )= e - G

S= G* e - G

Peaks at G=1 max(Throughput) = 1/e ~ 0.36

dS/dG = e-G – Ge-G = 0

Gmax = 1.0 Smax = 1/e ~ 0.368

82

Aloha (contd..)

Aloha performance :- not dependent on a Reservation Aloha for Satellite ( Kesav 152, Gallgher313)

Observations: ALOHA is an unstable protocol

If G increases to greater than 1 due to fluctuation in offered load, S will decrease

Reduction in throughput means fewer successful packet transmissions and more collisions

Number of retransmissions increases, backlogging messages to be transmitted and traffic load G

This in turn decreases S Results in operating point moving to right and S 0

Random access protocols can be made stable using backoff parameters

83

Comparison

With small p better s but longer delay

1.0

0.9

0.8

0.5

0.4

0.3

0.2

0.1

01 2 3 4 5 6 7 8 90

0.6

0.7

S (

thro

ughp

ut p

er p

acke

t tim

e)

G (attempts per packet time)

PureALOHA

SlottedALOHA

1-persistentCSMA

0.1-persistent CSMA

0.5-persistent CSMA

Nonpersistent CSMA

0.01 persistent CSMA

84

Difference Between Wired and Wireless

If both A and C sense the channel to be idle at the same time, they send at the same time.

Collision can be detected at sender in Ethernet. Half-duplex radios in wireless cannot detect collision

at sender.

A B C

A

B

C

Ethernet LAN Wireless LAN

85

A and C cannot hear each other. A sends to B, C cannot receive A. C wants to send to B, C senses a “free” medium (CS fails) Collision occurs at B. A cannot receive the collision (CD fails). A is “hidden” for C.

Hidden Terminal Problem

BA C

86

Exposed Terminal Problem

A starts sending to B. C senses carrier, finds medium in use and has to wait

for A->B to end. D is outside the range of A, therefore waiting is not

necessary. A and C are “exposed” terminals

A B

CD

87

CSMA: Distributed, Packet mode scheme

Carrier Sense and its variants: Use of carrier sensing capability to know if someone else is using

the medium 1 persistent

If medium busy, keep sensing If medium Idle send immediately

p persistent If medium busy, keep sensing If medium becomes Idle after continuous sensing,

send with probability p, (wait IFS time for 802.11, then do a random back-off)

in case of no-send (1-p), wait for 1 time slot, and begin medium sensing again

If medium is free for IFS period, transmit packet.

88

CSMA: Distributed, Packet mode scheme

Non persistent If medium busy, wait for random time before sensing again If medium Idle send immediately

89

Summary of CSMA schemes

90

802.11 - MAC layer

Traffic services Asynchronous Data Service (mandatory) – DCF Time-Bounded Service (optional) - PCF

Access methods DCF CSMA/CA (mandatory)

collision avoidance via randomized back-off mechanism ACK packet for acknowledgements (not for broadcasts)

DCF w/ RTS/CTS (optional) avoids hidden/exposed terminal problem, provides

reliability PCF (optional)

access point polls terminals according to a list

91

t

medium busy

DIFSDIFS

next frame

contention window(randomized back-offmechanism)

802.11 - CSMA/CA

station which has data to send starts sensing the medium (Carrier Sense based on CCA, Clear Channel Assessment)

if the medium is free for the duration of an Inter-Frame Space (IFS), the station can start sending (IFS depends on service type)

if the medium is busy, the station has to wait for a free IFS plus an additional random back-off time (multiple of slot-time)

if another station occupies the medium during the back-off time of the station, the back-off timer stops (fairness)

slot timedirect access if medium is free DIFS

92

802.11 DCF – basic access

If medium is free for DIFS time, station sends data receivers acknowledge at once (after waiting for SIFS) if the

packet was received correctly (CRC) automatic retransmission of data packets in case of transmission

errors

t

SIFS

DIFS

data

ACK

waiting time

otherstations

receiver

senderdata

DIFS

contention

93

Solution to Hidden/Exposed Terminals

A first sends a Request-to-Send (RTS) to B On receiving RTS, B responds Clear-to-Send (CTS) Hidden node C overhears CTS and keeps quiet

Transfer duration is included in both RTS and CTS Exposed node overhears a RTS but not the CTS

D’s transmission cannot interfere at B

A B C

RTS

CTS CTS

DATA

D

RTS

94

802.11 - Reliability Use acknowledgements

When B receives DATA from A, B sends an ACK If A fails to receive an ACK, A retransmits the DATA Both C and D remain quiet until ACK (to prevent collision of

ACK) Expected duration of transmission+ACK is included in

RTS/CTS packets

A B C

RTS

CTS CTS

DATA

D

RTS

ACK

95

802.11 –RTS/CTS If medium is free for DIFS, station can send RTS with reservation parameter

(reservation determines amount of time the data packet needs the medium) acknowledgement via CTS after SIFS by receiver (if ready to receive) sender can now send data at once, acknowledgement via ACK other stations store medium reservations distributed via RTS and CTS

t

SIFS

DIFS

data

ACK

defer access

otherstations

receiver

senderdata

DIFS

contention

RTS

CTS

SIFS SIFS

NAV (RTS)NAV (CTS)

96

802.11 - Carrier Sensing In IEEE 802.11, carrier sensing is performed

at the air interface (physical carrier sensing), and at the MAC layer (virtual carrier sensing)

Physical carrier sensing detects presence of other users by analyzing all

detected packets Detects activity in the channel via relative signal

strength from other sources Virtual carrier sensing is done by sending MPDU

duration information in the header of RTS/CTS and data frames

Channel is busy if either mechanisms indicate it to be Duration field indicates the amount of time (in

microseconds) required to complete frame transmission Stations in the BSS use the information in the duration

field to adjust their network allocation vector (NAV)

97

802.11 - Collision Avoidance If medium is not free during DIFS time.. Go into Collision Avoidance: Once channel

becomes idle, wait for DIFS time plus a randomly chosen backoff time before attempting to transmit

98

802.11 - Collision Avoidance

For DCF the backoff is chosen as follows: When first transmitting a packet, choose a backoff interval in

the range [0,cw]; cw is contention window, nominally 31 Count down the backoff interval when medium is idle Count-down is suspended if medium becomes busy When backoff interval reaches 0, transmit RTS If collision, then double the cw up to a maximum of 1024

Time spent counting down backoff intervals is part of MAC overhead

99

Example - backoff

data

waitB1 = 5

B2 = 15

B1 = 25

B2 = 20

data

wait

B1 and B2 are backoff intervalsat nodes 1 and 2

cw = 31

B2 = 10

100

802.11 - Priorities

defined through different inter frame spaces – mandatory idle time intervals between the transmission of frames

SIFS (Short Inter Frame Spacing) highest priority, for ACK, CTS, polling response SIFSTime and SlotTime are fixed per PHY layer (10 s

and 20 s respectively in DSSS) PIFS (PCF IFS)

medium priority, for time-bounded service using PCF PIFSTime = SIFSTime + SlotTime

DIFS (DCF IFS) lowest priority, for asynchronous data service DCF-IFS: DIFSTime = SIFSTime + 2xSlotTime

101

802.11 - Congestion Control

Contention window (cw) in DCF: Congestion control achieved by dynamically choosing cw

large cw leads to larger backoff intervals small cw leads to larger number of collisions

Binary Exponential Backoff in DCF: When a node fails to receive CTS in response to its

RTS, it increases the contention window cw is doubled (up to a bound cwmax =1023)

Upon successful completion data transfer, restore cw to cwmin=31