DefendingConcealednessinIEEE802cse.iitkgp.ac.in/~sandipc/presentations/COMSNETS_2014_P1.pdf- Channel Bonding: Combine two 20 MHz ... COMSNETS2014 07January,2014 15/20. Performance:

Defending Concealedness in IEEE 802.11n

Sandip Chakraborty12, Subhrendu Chattopadhyay, SuchetanaChakraborty, Sukumar Nandi

Department of Computer Science and EngineeringIndian Institute of Technology Guwahati,

Guwahati 781039 INDIA

07 January, 2014

1This work is supported by TATA Consultancy Services (TCS), INDIA through TCSResearch Fellowship program

2Supported by COMSNETS 2014 Travel GrantSandip Chakraborty (CSE, IITG) COMSNETS 2014 07 January, 2014 1 / 20

Preface: IEEE 802.11n

High speed extension for wireless networks: 600 Mbps

Extensions over IEEE 802.11b/g

- MIMO - Spatial diversity and spatial multiplexing

- Channel Bonding: Combine two 20 MHz channels to one 40 MHzchannel (also knowm as 20/40 semantics)

- MAC Layer: Frame aggregation and block acknowledgements

Source:http://www.eetimes.com/document.asp?doc_id=1278239Sandip Chakraborty (CSE, IITG) COMSNETS 2014 07 January, 2014 2 / 20

http://www.eetimes.com/document.asp?doc_id=1278239

IEEE 802.11 Channel Access

Sandip Chakraborty (CSE, IITG) COMSNETS 2014 07 January, 2014 3 / 20


Two-way handshaking (BasicAccess)




��

��

��

��

��

��

��

��

��

��

��

��

A

B

C

D

E

F




��

��

��

��

��

��

��

��

��

��

��

��

A

B

C

D

E

F

DATA




ACK ��

��

��

��

��

��

��

��

��

��

��

��

A

B

C

D

E

F




��

��

��

��

��

��

��

��

��

��

��

��

A

C

D

E

F

B




��

��

��

��

��

��

��

��

��

��

��

��

A

C

D

E

F

B

Hidden Terminal Problem




��

��

��

��

��

��

��

��

��

��

��

��

A

C

D

E

F

B


Four-way Handshake (RTS-CTSAccess)




��

��

��

��

��

��

��

��

��

��

��

��

A

C

D

E

F

B



��

��

��

��

��

��

��

��

��

��

��

��

A

B

C

D

E

F

RTS




��

��

��

��

��

��

��

��

��

��

��

��

A

C

D

E

F

B



��

��

��

��

��

��

��

��

��

��

��

��

A

B

C

D

E

F

CTS




��

��

��

��

��

��

��

��

��

��

��

��

A

C

D

E

F

B



��

��

��

��

��

��

��

��

��

��

��

��

A

B

C

D

E

F

DATA




��

��

��

��

��

��

��

��

��

��

��

��

A

C

D

E

F

B



��

��

��

��

��

��

��

��

��

��

��

��

A

B

C

D

E

F

ACK




��

��

��

��

��

��

��

��

��

��

��

��

A

C

D

E

F

B



��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

A

B

C

D

E

F




��

��

��

��

��

��

��

��

��

��

��

��

A

C

D

E

F

B



��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

A

B

C

D

E

F




��

��

��

��

��

��

��

��

��

��

��

��

A

C

D

E

F

B



��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

A

B

C

D

E

F

T-Exposed Terminal Problem




��

��

��

��

��

��

��

��

��

��

��

��

A

C

D

E

F

B



��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

A

B

C

D

E

F

T-Exposed Terminal Problem

R-Exposed Terminal Problem


Hidden and Exposed Nodes

Extensive studies have been carried out to mitigate with hidden andexposed nodes problem.

For, IEEE 802.11b/g networks, solutions of hidden and exposed nodesrequire extra controlling overhead, which severe for a low to moderateloaded network.

Four-way access may sometime impact in a negative performance3

Is the scenario similar for IEEE 802.11n high data rates?

3R. Bruno, M. Conti, and E. Gregori. “IEEE 802.11 optimal performances: RTS/CTSmechanism vs. basic access.” IEEE PIMRC, 2002.Sandip Chakraborty (CSE, IITG) COMSNETS 2014 07 January, 2014 4 / 20

IEEE 802.11n: Basic Access vs RTS/CTS Access

Model basic access and RTS/CTS access, considering- High data rates

- Frame aggregation (A-MPDU) : Channel is hold once access issuccessful

- Block acknowledgements (BACK) : Loss in BACKs result in the loss incomplete A-MPDU

- A mesh network scenario: Contention is high

I

DS

DF

P{I|I}

1

P{DF|I}

D

1P{D|I}

P{DS|I}

1

Basic Access

P{I|I}

P{RTSF|I}1

I

1

1

P{RTSS|I}

P{DS|RTSS}

DSP{DF|RTSS}

DF

1

1

NE

E

P{E|I}

P{NE|I}

RS

RF

RTS/CTS AccessSandip Chakraborty (CSE, IITG) COMSNETS 2014 07 January, 2014 5 / 20

IEEE 802.11n: The Optimal Access

Cost of expode nodes is very high: Channel is reserved for a largeamount of time once access is successful.

Source:http://www.eetimes.com/document.asp?doc_id=1278239


http://www.eetimes.com/document.asp?doc_id=1278239


However, access synchronization is crucial,

- Possibility of data-BACK collision

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

A

B

E

F

DATA

DATA

A

B

E

F

DATA

BACK



However, access synchronization is crucial,

- Possibility of data-BACK collision

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

A

B

E

F

DATA

DATA

A

B

E

F

DATA

BACK

Optimal Access:

- No hidden nodes

- No exposed nodes

- DATA-BACK collisions are avoided


Outcomes from the Model: Numerical Results

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

No

rma

lize

d t

hro

ug

hp

ut

Data generation probability

Basic accessFour-way accessOptimal scenario

0.3

0.4

0.5

0.6

0.7

0.8

0.9

8 10 12 14 16 18 20 22 24N

orm

ali

ze

d t

hro

ug

hp

ut

Node density (per 100 mt)

Basic accessFour-way accessOptimal scenario


Key Observations

Can not avoid hidden nodes: effect of data loss is significant at highdata rates, with moderate contention (basic access outperformsRTS/CTS only at very low contention).

Exposed nodes result in severe underutilization.

Keys for the good (near-optimal) performance:

- Avoid hidden nodes (RTS/CTS access)

- Allow transmissions for the exposed nodes (deviation from the originalfour-way access)

- Avoid data-BACK collision whenever possible (opportunistic access)


Opportunistic Four Way Access

Allow transmission to the exposed nodes, but avoid hidden nodes

- Maintain list of active nodes from which RTS and CTS are overheard(RTSact and CTSact).

- NS : Set of one hop neighborhood for node S .

- Node S wants to transmit to node R ; Node S sends a RTS ifCTSact = NULL and ∀RTSact .DST /∈ NS

- CTSact = NULL implies no active receiver in the neighborhood

- RTSact .DST also implies a receiver, so the second condition ensures nopossible receiver in the neighborhood for the next few slots (there existsno such nodes in the neighborhood that has received a RTS, but yet toreply for a CTS)

- Allows transmission for the T-Exposed nodes.


Opportunistic Four Way Access

Node R receives an RTS from node S . It replies back with a CTS(wants to act as a receiver) only if,

- RTSact = NULL, ensures no active transmitter in the neighborhood

- ∀CTSact .DST /∈ NR , ensures no CTS is overheard where transmitter isin the neighborhood (supports communication asymmetry)

- Allows transmission for the R-Exposed nodes.


Opportunistic Access: Interference

��

��

��

��

MPDU MPDUHDR MPDU MPDU

HDR MPDU MPDU MPDUMPDU

A−MPDU

back−offback−off RTS

CTS

txopp

txopp

A−MPDU

back−off

RTS

CTS

back−off

BACK

BACK

DATA−ACK Collision

DATA−CTRL Collision

A

B

D

E


Opportunistic Access: Interference Mitigation

MPDU MPDUHDR

HDR MPDU

A−MPDU

back−off

back−off

A−MPDU

back−off RTS

CTS

CTS

NULL

NULL

txopp

txopp

BACK

back−off

BACK

RTS

MPDU

MPDUMPDU

A

B

D

E


Testbed Evaluation

N6N7

N N2 N1

N

N

N

3

4

5

8

RS Lab 1RS Lab 2

Security Lab

G

Router: RaLink RT-3352 RoC: 2T2R MAC/BBP/PA/RF, 400MHzMIPS24KEc CPU, 64MB of SDRAM and 32MB of FlashIEEE 802.11n: 300 Mbps, channel bondingOpen80211s: http://www.open80211s.orgLinux Kernel 2.8.54TCP (FTP) and UDP (TFTP) using iperf

(http://iperf.sourceforge.net/)Semi-indoor environment, Tx Power 16dBm, Rx Sensitivity 0 dBm (45-55mt in indoor)


http://www.open80211s.org

http://iperf.sourceforge.net/‎

Performance: Throughput

0

20

40

60

80

100

N1 N2 N3 N4 N5 N6 N7 N8

Tra

ns

mis

sio

n T

hro

ug

hp

ut

(Mb

ps

)

Node Number

Basic AccessFour-way Access

Opportunistic Access

High Load

0

5

10

15

20

N1 N2 N3 N4 N5 N6 N7 N8

Tra

ns

mis

sio

n T

hro

ug

hp

ut

(Mb

ps

)

Node Number



Low Load


Performance: Forwarding Delay

0

2

4

6

8

10

N1 N2 N3 N4 N5 N6 N7 N8

En

d-t

o-e

nd

De

lay

(m

s)

Node Number



High Load

0

0.5

1

1.5

2

2.5

3

N1 N2 N3 N4 N5 N6 N7 N8

En

d-t

o-e

nd

De

lay

(m

s)

Node Number



Low Load


Performance: Data Loss

0

100

200

300

400

500

600

N1 N2 N3 N4 N5 N6 N7 N8

Da

ta L

os

s (

Kb

ps

)

Node Number



High Load

0

10

20

30

40

50

60

70

80

N1 N2 N3 N4 N5 N6 N7 N8

Da

ta L

os

s (

Kb

ps

)

Node Number



Low Load


General Performance Metrics

0

1

2

3

4

5

6

1 2 3 4 5 6 7 8 9 10

Tra

ns

mis

sio

n T

hro

ug

hp

ut

(Mb

ps

)

Traffic Generation Rate (Mbps)

Basic AccessFour way Access


Throughput

0

0.5

1

1.5

2

2.5

3

3.5

4

4.5

5

1 2 3 4 5 6 7 8 9 10

En

d-t

o-e

nd

De

lay

(m

s)




Forwarding Delay


General Performance Metrics

0

50

100

150

200

250

300

1 2 3 4 5 6 7 8 9 10

Da

ta L

os

s (

Kb

ps

)




Data Loss

0

0.2

0.4

0.6

0.8

1

1 2 3 4 5 6 7 8 9 10

Fa

irn

es

s I

nd

ex




Fairness


Conclusion with Some Comments

Explores the severity of hidden and exposed terminal problems overIEEE 802.11n

Proposed an opportunistic solution to improve MAC performance overIEEE 802.11n

- Distributed as well as localized- Less control overhead- Improved spatial reuse

This paper has considered IEEE 802.11n with A-MPDU aggregation.However the compatibility of this solution with the existing basic andfour way access is further required to be investigated.

- How does the opportunistic solution perform in mix network of basic access,four-way access and opportunistic access?

- How does the proposed solution scale with number of nodes in the network?


Conclusion with Some Comments

Explores the severity of hidden and exposed terminal problems overIEEE 802.11n

Proposed an opportunistic solution to improve MAC performance overIEEE 802.11n

- Distributed as well as localized- Less control overhead- Improved spatial reuse

This paper has considered IEEE 802.11n with A-MPDU aggregation.However the compatibility of this solution with the existing basic andfour way access is further required to be investigated.

- How does the opportunistic solution perform in mix network of basic access,four-way access and opportunistic access?

- How does the proposed solution scale with number of nodes in the network?

Thank You


Documents

DefendingConcealednessinIEEE802cse.iitkgp.ac.in/~sandipc/presentations/COMSNETS_2014_P1.pdf- Channel Bonding: Combine two 20 MHz ... COMSNETS2014 07January,2014 15/20. Performance: