Evaluation of 802.11a for Evaluation of 802.11a for Streaming Data - - PowerPoint PPT Presentation

▶
evaluation of 802 11a for evaluation of 802 11a for
SMART_READER_LITE
LIVE PREVIEW

Evaluation of 802.11a for Evaluation of 802.11a for Streaming Data - - PowerPoint PPT Presentation

Jul 23, 2023 387 likes •694 views

Evaluation of 802.11a for Evaluation of 802.11a for Streaming Data in Ad- -hoc hoc Streaming Data in Ad Networks Networks Samip Bararia Bararia, , Shahram Shahram Ghandeharizadeh Ghandeharizadeh, Shyam , Shyam Kapadia Kapadia Samip

slide-1
SLIDE 1

Evaluation of 802.11a for Evaluation of 802.11a for Streaming Data in Ad Streaming Data in Ad-

  • hoc

hoc Networks Networks

Samip Samip Bararia Bararia, , Shahram Shahram Ghandeharizadeh Ghandeharizadeh, Shyam , Shyam Kapadia Kapadia Computer Science Department Computer Science Department University of Southern California University of Southern California Los Angeles 90089 Los Angeles 90089 bararia@usc.edu,shahram@usc.edu,kapadia@usc.edu bararia@usc.edu,shahram@usc.edu,kapadia@usc.edu

slide-2
SLIDE 2

H20 Application class: An example deployment H20 Application class: An example deployment

slide-3
SLIDE 3

H20 Application class: An example deployment H20 Application class: An example deployment

H2O device roles: H2O device roles:

Data producer (source), Data forwarder (router), Data consumer ( Data producer (source), Data forwarder (router), Data consumer (sink) sink)

slide-4
SLIDE 4

Candidate wireless technologies Candidate wireless technologies

Technology Frequency band Spec B/W Typical B/W Radio-range(indoor) Bluetooth 2.4Ghz 1Mbps 700Kbps 30 feet 802.11b 2.4-2.48Ghz 11Mbps 4-5Mbps 300 feet 802.11a 5.725-5.85Ghz 54Mbps 20-25Mbps 40 feet

Note: Note: (1) (1) 802.11a 802.11a turbo provides bandwidths turbo provides bandwidths upto upto 75Mbps (raw) but not supported by all 75Mbps (raw) but not supported by all manufacturers (not a IEEE std) manufacturers (not a IEEE std) (2) (2) Bandwidth required for display of a DVD Bandwidth required for display of a DVD-quality (MPEG quality (MPEG-2) video clip is 4Mbps. 2) video clip is 4Mbps.

slide-5
SLIDE 5

Hypothesis: Hypothesis: IEEE 802.11a IEEE 802.11a may be a feasible option for may be a feasible option for the H20 application class. the H20 application class.

slide-6
SLIDE 6

Dimensions of the empirical study Dimensions of the empirical study

  • Distance between participating devices

Distance between participating devices

  • Number of intermediate H20 devices used to route a stream

Number of intermediate H20 devices used to route a stream from a producing H20 device to a consuming H20 device from a producing H20 device to a consuming H20 device

  • Number of simultaneous senders in the same radio range

Number of simultaneous senders in the same radio range

  • Operating system level versus application level routing

Operating system level versus application level routing

Note: Used INTEL PRO/Wireless 5000 LAN Note: Used INTEL PRO/Wireless 5000 LAN Cardbus Cardbus adapter 802.11a cards at adapter 802.11a cards at 54Mbps ( 54Mbps (Auto data rate control disabled Auto data rate control disabled)

slide-7
SLIDE 7
  • For e.g.

For e.g.

  • (a) 3:1 hop

(a) 3:1 hop transmission transmission

  • (b) 1:3 hop

(b) 1:3 hop transmission transmission

Terminology Terminology

  • In general, any scenario is m transmissions k

In general, any scenario is m transmissions k hops each hops each

  • Denoted as

Denoted as m:k m:k, , m,k m,k>=1 >=1

Movie Movie D Node 1 Node 1 Node 2 Node 2 Node 3 Node 3 Node 4 Node 4 Movie Movie D Movie Movie D Node 1 Node 1 Node 2 Node 2 Node 3 Node 3 Node 4 Node 4 Movie Movie A Movie Movie B Movie Movie C 1 foot 1 foot 1 foot 1 foot 1 foot 1 foot

slide-8
SLIDE 8

Terminology ( Terminology (contd contd) )

TCP/UDP TCP/UDP Data=1GB Data=1GB

ADU ADU ADU ADU ADU ADU

Application Application layer layer

IEEE IEEE 802.11a 802.11a

TCP/UDP TCP/UDP Data=1GB Data=1GB

ADU ADU ADU ADU ADU ADU

Application Application layer layer

Data producer Data producer Data consumer Data consumer

Note: ADU Note: ADU – – Application Data Unit Application Data Unit

slide-9
SLIDE 9

TCP and UDP performance for a 1:3 hop connection TCP and UDP performance for a 1:3 hop connection Bandwidth ( Bandwidth (Goodput Goodput) and loss rate for a 1:3 hop ) and loss rate for a 1:3 hop connection. connection.

Movie Movie D Node 1 Node 1 Node 2 Node 2 Node 3 Node 3 Node 4 Node 4 Movie Movie D Movie Movie D

slide-10
SLIDE 10

Observations Observations

  • UDP Loss rate between 15

UDP Loss rate between 15-

  • 30% with a large

30% with a large variance variance

  • Losses occur due to transient bottlenecks at

Losses occur due to transient bottlenecks at intermediate routers intermediate routers

  • k participants competing for the channel

k participants competing for the channel

  • Due to randomness intermediate router is

Due to randomness intermediate router is flooded occasionally and drops data flooded occasionally and drops data

  • TCP performs well even though there is the ACK

TCP performs well even though there is the ACK

  • verhead
  • verhead
  • A protocol with flow control and congestion

A protocol with flow control and congestion control does well in case multiple senders in control does well in case multiple senders in the same radio range the same radio range

  • System may produce data at a slower rate than

System may produce data at a slower rate than available network bandwidth available network bandwidth

  • Introduce a delay between successive

Introduce a delay between successive ADUs ADUs

slide-11
SLIDE 11

Terminology ( Terminology (contd contd) )

TCP/UDP TCP/UDP Data=1GB Data=1GB

ADU ADU ADU ADU ADU ADU

Application Application layer layer

IEEE IEEE 802.11a 802.11a

TCP/UDP TCP/UDP Data=1GB Data=1GB

ADU ADU ADU ADU ADU ADU

Application Application layer layer

Data producer Data producer Data consumer Data consumer

Note: ADU Note: ADU – – Application Data Unit Application Data Unit Wait Wait-

  • time

time

slide-12
SLIDE 12

Data Flow Control Data Flow Control

Bandwidth and loss rate with UDP for a 1:3 hop connection with w Bandwidth and loss rate with UDP for a 1:3 hop connection with wait ait-

  • time.

time.

slide-13
SLIDE 13

Data Flow Control Data Flow Control

Bandwidth and loss rate with UDP for a 1:3 hop connection with w Bandwidth and loss rate with UDP for a 1:3 hop connection with wait ait-

  • time.

time.

0ms wait 0ms wait-time: Time=961s time: Time=961s 1ms wait 1ms wait-time: Time=1106s time: Time=1106s

slide-14
SLIDE 14

TCP and UDP performance for TCP and UDP performance for 3:1 hop connection 3:1 hop connection

Node 1 Node 1 Node 2 Node 2 Node 3 Node 3 Node 4 Node 4 Movie Movie A Movie Movie B Movie Movie C 1 foot 1 foot 1 foot 1 foot 1 foot 1 foot

slide-15
SLIDE 15

Observations Observations

  • TCP and UDP bandwidth drops by 1/3 as compared to 1:1

TCP and UDP bandwidth drops by 1/3 as compared to 1:1

  • 3 senders contending for the medium

3 senders contending for the medium

  • Loss rate for UDP is about 0.2%

Loss rate for UDP is about 0.2%

  • Allocation of bandwidth is approximately fair

Allocation of bandwidth is approximately fair

slide-16
SLIDE 16

Distance experiments Distance experiments

  • Carried out

Carried out experiments with a 1:1 configuration at USC track field, university housing (indoor experiments) and Marina-del-Rey beach

Node 1 Node 1 Node 2 Node 2 Movie Movie A d feet d feet

slide-17
SLIDE 17

Exposed node limitation Exposed node limitation

  • Related work has shown that exposed node[6]

degrades the performance of 802.11 severely

  • Experimental setup
  • Two pairs of nodes spaced d feet apart
  • 100 MB of data with ADU size of 1KB

1:1 1:1 1:1 1:1

Node 1 Node 1 Node 2 Node 2 Node 3 Node 3 Stream Stream 1 Stream Stream 2 10 feet 10 feet d feet d feet Node 4 Node 4 10 feet 10 feet

slide-18
SLIDE 18

Results Results

17.79747 18.8331 500 17.06635 17.34653 450 16.95107 17.80064 400 14.04708 16.23252 300 14.01428 14.09932 250 13.65804 13.02289 200 12.5572 12.2803 150 12.99614 12.24468 100 Bandwidth Bandwidth d (feet) Session 2 Session 1

  • Results show that each stream observes a bandwidth of 12.2 –

14.4 Mbps up to 250 feet.

slide-19
SLIDE 19

Related work Related work

  • [5] studies the feasibility of IEEE 802.11b as a viable candidat

[5] studies the feasibility of IEEE 802.11b as a viable candidate e for wireless ad hoc networks for wireless ad hoc networks

  • TCP one

TCP one-

  • hop unfairness problem

hop unfairness problem

  • Simulation study verified with empirical deployment

Simulation study verified with empirical deployment 1:1 1:1 1:2 1:2

slide-20
SLIDE 20

No Dropped connections No Dropped connections

  • Experimental setup
  • Even with the 1-

hop flow running

  • n UDP, TCP does

not drop connections.

  • Allocation of

bandwidth is fair across UDP and TCP flows

Movie Movie A Movie Movie A 50 feet 50 feet 50 feet 50 feet 50 feet 50 feet 50 feet 50 feet Node 5 Node 5 Movie Movie B Node 4 Node 4

1:1 1:1

Node 3 Node 3 Node 2 Node 2 Node 1 Node 1

1:2 1:2

slide-21
SLIDE 21

Differences between IEEE 802.11a and Differences between IEEE 802.11a and IEEE 802.11b IEEE 802.11b

  • IEEE 802.11a

IEEE 802.11a

  • Has 12 channels (compared to 3 for 802.11b)

Has 12 channels (compared to 3 for 802.11b)

  • 8 for indoor and 4 for outdoor use

8 for indoor and 4 for outdoor use

  • Lower co

Lower co-

  • channel interference

channel interference

  • Allows for higher user densities and higher system

Allows for higher user densities and higher system data throughput data throughput

  • Higher bandwidth 54Mbps as compared to

Higher bandwidth 54Mbps as compared to 11Mbps for 802.11b 11Mbps for 802.11b

  • Higher system capacity

Higher system capacity

slide-22
SLIDE 22

Related work Related work

  • Does not contradict [3,4] using TCP

Does not contradict [3,4] using TCP-

  • ELFN and TCP

ELFN and TCP-

  • ECN

ECN

  • [6] does an empirical study with IEEE 802.11b

[6] does an empirical study with IEEE 802.11b

  • [7] MIT

[7] MIT Roofnet Roofnet project project

  • [8] Microsoft Research

[8] Microsoft Research Meshnet Meshnet project project

  • [9] IEEE 802.11a paper by

[9] IEEE 802.11a paper by Atheros Atheros

  • Comparison between IEEE 802.11b and IEEE 802.11a in an

Comparison between IEEE 802.11b and IEEE 802.11a in an

  • ffice environment
  • ffice environment
slide-23
SLIDE 23

Conclusions Conclusions

  • IEEE 802.11a is feasible for the class of

applications such as H20

  • Bandwidth and Loss rate observed in

experiments across the different dimensions were sufficient for DVD quality display

  • A protocol with flow control and congestion

control is needed for streaming in the H20 environment

  • The allocation of bandwidth among multiple

competing 1-hop TCP and UDP flows is fair

  • Exposed node limitation does not affect 802.11a

severely

  • No one-hop unfairness observed with 802.11a
slide-24
SLIDE 24

Future work Future work

  • A simulation and analytical model to capture the behavior
  • Streaming issues
  • Hiccups and start-up latency
  • Pre-fetching/Buffering
  • Experimentation with
  • Different variants of TCP
  • 802.11e cards (when they become available)
  • Data placement and statistical admission control
  • Mobility
  • C2P2 (Car-to-Car Peer-to-Peer) Networks
slide-25
SLIDE 25

References References

  • [1] V.Bush,

[1] V.Bush, As We May Think As We May Think. . The Atlantic Monthly, 127(6):101

The Atlantic Monthly, 127(6):101-108, July 1945 108, July 1945

  • [2]

[2] J.Gemmell J.Gemmell, B.Gordon, , B.Gordon, R.Lueder R.Lueder, , S.Drucker S.Drucker, and C.Wong, , and C.Wong, MyLifeBits:Fullfilling MyLifeBits:Fullfilling the the Memex Memex Vision.

  • Vision. In ACM Multimedia, December 2002

In ACM Multimedia, December 2002

  • [3] G.Holland and

[3] G.Holland and N.H.Vaidya N.H.Vaidya, , “Analysis of TCP performance over Mobile “Analysis of TCP performance over Mobile Ad Hoc Networks Ad Hoc Networks”, ”, Mobile Computing and Networking, pages 219

Mobile Computing and Networking, pages 219-230, 1999. 230, 1999.

  • [4] S.Floyd and

[4] S.Floyd and K.Ramakrishnan K.Ramakrishnan, , “TCP with ECN: The Treatment of “TCP with ECN: The Treatment of Retransmitted Data Packets Retransmitted Data Packets”, ”, 2000

2000

  • [5] S.

[5] S. Xu Xu and T. and T. Saadawi Saadawi, , “Does the IEEE 802.11 MAC Protocol Work Well “Does the IEEE 802.11 MAC Protocol Work Well in in Multihop Multihop Wireless Ad Hoc Networks? Wireless Ad Hoc Networks?”, ”, IEEE Communications Magazine,

IEEE Communications Magazine, pages 130 pages 130-137, June 2001 137, June 2001

  • [6] G.

[6] G. Anastasi Anastasi, E. , E. Borgia Borgia, , M.Conte M.Conte, , E.Gregori E.Gregori, , “IEEE 802.11 Ad Hoc “IEEE 802.11 Ad Hoc Networks: Performance Measurements Networks: Performance Measurements”, ”, 1999

1999.

.

  • [7]

[7] http:// http://www.pdos.lcs.mit.edu/roofnet www.pdos.lcs.mit.edu/roofnet/ /

  • [8]

[8] http://research.microsoft.com/sn/mesh/ http://research.microsoft.com/sn/mesh/

  • [9]

[9] Atheros Atheros Communications, “ Communications, “Measured Performance of 5 Measured Performance of 5-

  • GHz

GHz 802.11a Wireless LAN Systems 802.11a Wireless LAN Systems”, 2001, ”, 2001,

http://epsfiles.intermec.com/eps_files/eps_wp/AtherosRangeCapaci http://epsfiles.intermec.com/eps_files/eps_wp/AtherosRangeCapacityPaper.pdf tyPaper.pdf

slide-26
SLIDE 26

Questions Questions

THANK YOU THANK YOU

slide-27
SLIDE 27

Data Flow Control Data Flow Control

Bandwidth and loss rate with UDP for Bandwidth and loss rate with UDP for a 1:3 hop connection with wait a 1:3 hop connection with wait-

  • time.

time.

slide-28
SLIDE 28

Observations Observations

  • Loss reduces significantly with wait

Loss reduces significantly with wait-

  • time

time

  • Data is sent out at a slower rate

Data is sent out at a slower rate

  • With ADU of 1KB bandwidth observed with a wait

With ADU of 1KB bandwidth observed with a wait-

  • time of 1ms

time of 1ms is higher than that observed with 2ms is higher than that observed with 2ms

  • With 1ms wait

With 1ms wait-

  • time the transmission time eclipses the wait

time the transmission time eclipses the wait-

  • time

time

  • With 2ms wait

With 2ms wait-

  • time exceeds the transmission time

time exceeds the transmission time

  • Network remains idle giving lower bandwidth

Network remains idle giving lower bandwidth

  • Execution times for 0ms,1ms and 2ms are 961,1106 and 2187

Execution times for 0ms,1ms and 2ms are 961,1106 and 2187 seconds. seconds.

  • For ADU size > 2KB bandwidth and loss for wait

For ADU size > 2KB bandwidth and loss for wait-

  • time=1ms

time=1ms and wait and wait-

  • time=2ms is almost identical

time=2ms is almost identical

  • With 2KB minimum transmission time with 1ms and 2ms wait is

With 2KB minimum transmission time with 1ms and 2ms wait is 524s and 1048s respectively 524s and 1048s respectively

  • With wait

With wait-

  • time = 0ms taken to complete experiment = 976s

time = 0ms taken to complete experiment = 976s

  • With a wait

With a wait-

  • time bandwidth increases with ADU size

time bandwidth increases with ADU size

  • Delay causes network to remain idle but idle time reduces with

Delay causes network to remain idle but idle time reduces with ADU size ADU size

slide-29
SLIDE 29

Observations ( Observations (contd contd) )

  • Large losses seen in 1:k configuration

Large losses seen in 1:k configuration

  • Trends seen are similar in 1:2, 1:4, 1:5 configurations

Trends seen are similar in 1:2, 1:4, 1:5 configurations

  • Loss has a high variance

Loss has a high variance

  • To investigate losses further we used routing at the operating

To investigate losses further we used routing at the operating system level and 2 network cards per computer system level and 2 network cards per computer

slide-30
SLIDE 30

Application and Operating system level routing Application and Operating system level routing results of UDP for ADU size = 1KB results of UDP for ADU size = 1KB