44
17/10/2003 TCP performance over ad- hoc mobile networks. 1 TCP performance over ad-hoc mobile networks. LCCN – summer 2003 Uri Silbershtein Roi Dayagi Nir Hasson

TCP performance over ad-hoc mobile networks

  • Upload
    brand

  • View
    79

  • Download
    0

Embed Size (px)

DESCRIPTION

TCP performance over ad-hoc mobile networks. LCCN – summer 2003. Uri Silbershtein Roi Dayagi Nir Hasson. Presentation content. Purpose of this project. Ad-hoc mobile networks overview. TCP Configurations used. Mobile networks routing protocols. Simulation methodology - PowerPoint PPT Presentation

Citation preview

Page 1: TCP performance over ad-hoc mobile networks

17/10/2003 TCP performance over ad-hoc mobile networks.

1

TCP performance over ad-hoc mobile networks.

LCCN – summer 2003

Uri Silbershtein Roi Dayagi Nir Hasson

Page 2: TCP performance over ad-hoc mobile networks

17/10/2003 TCP performance over ad-hoc mobile networks.

2

Presentation content.

• Purpose of this project.

• Ad-hoc mobile networks overview.

• TCP Configurations used.

• Mobile networks routing protocols.

• Simulation methodology

• Simulation results.

• Conclusions.

Page 3: TCP performance over ad-hoc mobile networks

17/10/2003 TCP performance over ad-hoc mobile networks.

3

Purpose of this project.

• Look for TCP configurations networks in an end to end approach,that improve ad-hoc mobile networks throughput.

• Test TCP configurations over several routing protocols in an ad-hoc mobile network.

• Test TCP configurations over different ad-hoc mobile environment.

• TCP configurations were limited to those which are supported by the ns.

Page 4: TCP performance over ad-hoc mobile networks

17/10/2003 TCP performance over ad-hoc mobile networks.

4

Ad-hoc mobile networks overview.

• Unlike most of portable communication, ad-hoc network does not have fixed base stations.

• The network topology is in general dynamic.• TCP was original designed to static networks. • Each node in a wireless ad hoc network functions

both as a host and as a router.• Ad-hoc mobile networks examples: natural

disasters, military settings.

Page 5: TCP performance over ad-hoc mobile networks

17/10/2003 TCP performance over ad-hoc mobile networks.

5

Tested TCP configurations.

• Selective acknowledgments (SACK) and Delayed ACK.• Explicit Congestion Notification. (ECN)

• Limited Transmit. (LT)

Page 6: TCP performance over ad-hoc mobile networks

17/10/2003 TCP performance over ad-hoc mobile networks.

6

Tested TCP configurations.• Selective acknowledgments (SACK) and Delayed ACK.

– High probability that some of the packets from the same sending window will be corrupted.(high bit error rate)

– By using selective ACK option we can ensure packets that were

received by the destination successfully, will not be send again.

• Explicit Congestion Notification. (ECN)

• Limited Transmit. (LT)

Page 7: TCP performance over ad-hoc mobile networks

17/10/2003 TCP performance over ad-hoc mobile networks.

7

Tested TCP configurations.• Selective acknowledgments (SACK) and Delayed ACK.

• Explicit Congestion Notification. (ECN)• TCP relies on packet drops as the indication of congestion.• ECN idea is that the router will signal the development of the

congestion before a packet has to be discarded.

• In ad-hoc mobile networks, each node serves as a router, and therefore can notify to the sources about its congestion.

• In this way loss of packets because of congestion could be avoided.

• In mobile topology there is high probability to congestion, occurs due to bottleneck nodes.

• Limited Transmit. (LT)

Page 8: TCP performance over ad-hoc mobile networks

17/10/2003 TCP performance over ad-hoc mobile networks.

8

Tested TCP configurations.• Selective acknowledgments (SACK) and Delayed ACK.

• Explicit Congestion Notification. (ECN)

• Limited Transmit. (LT)• LT algorithm: when two consecutive ACKs for the same segment

are received, a new segment is transmitted if the ordinary conditions fulfilled.

• The main idea is to trigger the Fast Retransmit algorithm.

• This is helpful in cases where the congestion window is too small for the sender to receive three duplicate ACKs - Avoid Retransmission Timeouts (RTOs).

 

Page 9: TCP performance over ad-hoc mobile networks

17/10/2003 TCP performance over ad-hoc mobile networks.

9

Mobile networks routing protocols.

• DSR and DSDV were the main routing protocols used over the project.

• DSR considered to have the second best performance over ad-hoc networks, according to research done by IBM labs.

• Another routing protocols that was tested are AODV,DSDV and TORA.

• Our main problem was the ns limits – it does not support SSA routing protocol and some routing protocols do not support ECN feature.

Page 10: TCP performance over ad-hoc mobile networks

17/10/2003 TCP performance over ad-hoc mobile networks.

10

Simulation methodology.

• We fixed crucial environment parameters, that effect dramatically the TCP performance over ad-hoc mobile networks.

• Those environment parameters are:– area size.area size.

– total number of nodes.total number of nodes.

– number of sources. number of sources.

– nodes velocity.nodes velocity.

Page 11: TCP performance over ad-hoc mobile networks

17/10/2003 TCP performance over ad-hoc mobile networks.

11

Simulation methodology (cont).

• The following environment parameters were used over our scenarios:

– Area size: 500m*400m, 600m*400m,1500m*1000m.

– Total number of nodes: 25 and 10.

– Number of sources: 10, 6 and 1.

– Nodes’ velocity: 30m\s, 15m\s and 5m\s (108km\h, 54km\h, 18km\h).

• Scenario run time was 200 sec and 100sec.

• Because we want to test ECN in mobile network, we had to use ns 2.26 version which is very sensitive and limiting regarding the value of the environment parameters.

Page 12: TCP performance over ad-hoc mobile networks

17/10/2003 TCP performance over ad-hoc mobile networks.

12

Simulation methodology (cont).

• Automatic Script Generator (ASG). In order to have large amount of results over random scripts, an automatic script generator was developed. Using that tool we were able to run over total of 1000 random scenarios.

• ..\vb\Project1.1.2.exe

Page 13: TCP performance over ad-hoc mobile networks

17/10/2003 TCP performance over ad-hoc mobile networks.

13

Simulation methodology (cont).

Throughput Calculation(in general)

The formula for calculation is the sum of the total TCP packets (in bytes), which were received by every destination node or router divided by the scenario run time.

Page 14: TCP performance over ad-hoc mobile networks

17/10/2003 TCP performance over ad-hoc mobile networks.

14

Simulation methodology (cont).

Throughput Calculation(in details)

The generated tcl files (scenarios)

produce traces file during their run.

The relevant lines, in trace file,consist

in their first column “r” (received),

in the seventh “tcp” and

In the eighth column the size of the TCP

packet in bytes. 

Page 15: TCP performance over ad-hoc mobile networks

17/10/2003 TCP performance over ad-hoc mobile networks.

15

Simulation ResultsIn the following slides each graph has the same environmentConfigurations. The changes are in the TCP configurationson TCP NewReno.

The TCP configurations are:• BASE: Basic configuration.• SACK: Selective acknowledgment and delayed Ack.• LT: Limited transmit.• ECN: Explicit congestion notification.

Note: Not all routing protocols in the ns support the ECN option.

Page 16: TCP performance over ad-hoc mobile networks

17/10/2003 TCP performance over ad-hoc mobile networks.

16

Dynamic Source Routing

Page 17: TCP performance over ad-hoc mobile networks

17/10/2003 TCP performance over ad-hoc mobile networks.

17

Configurations

• Environment configurations: area size 500m*400m, total nodes, 25 and one source

• In this configuration there is no significant difference in the throughput. The reason for that is when one source node transmits, no series congestion can be produced.

Page 18: TCP performance over ad-hoc mobile networks

17/10/2003 TCP performance over ad-hoc mobile networks.

18

GraphsVelocity 30 m/s

0

100000

200000

300000

1 2 3 4 5 6 7 8 9 10

senario

thro

ug

hp

ut

BASE

SACK

LT

Velocity 15 m/s

0

100000

200000

300000

1 2 3 4 5 6 7 8 9 10

Scenario

Th

rou

gh

pu

t

BASE

SACK

LT

Page 19: TCP performance over ad-hoc mobile networks

17/10/2003 TCP performance over ad-hoc mobile networks.

19

GraphsVelocity 30 m/s

0

100000

200000

300000

1 2 3 4 5 6 7 8 9 10

senario

thro

ug

hp

ut

BASE

SACK

LT

Velocity 15 m/s

0

100000

200000

300000

1 2 3 4 5 6 7 8 9 10

Scenario

Th

rou

gh

pu

t

BASE

SACK

LT

Page 20: TCP performance over ad-hoc mobile networks

17/10/2003 TCP performance over ad-hoc mobile networks.

20

Configurations

• Environment configurations: area size500m*400m, total nodes 25, 10 sources.

• In this configuration we are witnessing to the greater impact sack has to the throughput performance impact. Limited transmit does not improve TCP performance.

Page 21: TCP performance over ad-hoc mobile networks

17/10/2003 TCP performance over ad-hoc mobile networks.

21

GraphsVelocity 30m\s

170000

180000

190000

200000

210000

220000

1 2 3 4 5 6 7 8 9 10

scenario

thro

ug

hp

ut

BASE

SACK

LT

Velocity 15m/s

160000

170000

180000

190000

200000

210000

220000

1 2 3 4 5 6 7 8 9 10

scenario

thro

ug

hp

ut

BASE

SACK

LT

Page 22: TCP performance over ad-hoc mobile networks

17/10/2003 TCP performance over ad-hoc mobile networks.

22

Graphs (cont).Velocity 5m/s

180000185000190000195000200000205000210000215000220000

1 2 3 4 5 6 7 8 9 10

scenario

thro

ug

hp

ut

BASE

SACK

LT

Page 23: TCP performance over ad-hoc mobile networks

17/10/2003 TCP performance over ad-hoc mobile networks.

23

Graphs (cont).•As can seen in the following graph, SACK has a greater impact over high velocities.

% average throughput improvment

0.96

0.98

1

1.02

1.04

1.06

30 15 5

Velocity

% t

hro

ug

hp

ut

imp

rovm

en

t

SACK

Limited transmit

Page 24: TCP performance over ad-hoc mobile networks

17/10/2003 TCP performance over ad-hoc mobile networks.

24

Configurations

• Environment configurations: area size1500m*1000m, total nodes 25, 10 sources.

• Throughput in general has increased due to the area size changes.Velocity 30 m/s

0

100000

200000

300000

400000

500000

600000

700000

800000

1 2 3 4 5 6 7 8 9 10

senario

thro

ug

hp

ut

BASE

SACK

LT

Page 25: TCP performance over ad-hoc mobile networks

17/10/2003 TCP performance over ad-hoc mobile networks.

25

GraphsVelocity 15 m/s

0

100000200000

300000

400000

500000600000

700000

1 2 3 4 5 6 7 8 9 10

senario

thro

ug

hp

ut

BASE

SACK

LT

Velocity 5 m/s

0

100000

200000

300000400000

500000

600000

700000

1 2 3 4 5 6 7 8 9 10

senario

thro

ug

hp

ut

BASE

SACK

LT

Page 26: TCP performance over ad-hoc mobile networks

17/10/2003 TCP performance over ad-hoc mobile networks.

26

Graphs (cont).

% average throughput improvment

0.9

0.95

1

1.05

1.1

1.15

30 15 5

Velocity

% a

vg

. th

rou

gh

pu

t

imp

rovm

en

t

SACK 500*400SACK 1500*1000LT 500*400LT 1500*1000

A Comparison between throughput improvements over two area sizes – 500m*400m and 1500m*1000m,both with 10 sources out of total 25 nodes.

Page 27: TCP performance over ad-hoc mobile networks

17/10/2003 TCP performance over ad-hoc mobile networks.

27

Destination Sequenced Distance Vector

Page 28: TCP performance over ad-hoc mobile networks

17/10/2003 TCP performance over ad-hoc mobile networks.

28

Configurations

• Environment configurations: area size600m*400m, total nodes 11, 6 sources.

Velocity 30m/s

150000

155000

160000

165000

170000

175000

180000

185000

1 2 3 4 5 6 7 8 9 10

scenario

thro

ug

hp

ut BASE

SACK

LT

ECN

Page 29: TCP performance over ad-hoc mobile networks

17/10/2003 TCP performance over ad-hoc mobile networks.

29

GraphsVelocity 15m/s

150000

155000

160000

165000

170000

175000

180000

185000

190000

1 2 3 4 5 6 7 8 9 10

scenario

thro

ug

hp

ut

BASE

SACK

LT

ECN

Velocity 5m/s

172000174000176000178000180000

182000184000186000188000190000

1 2 3 4 5 6 7 8 9 10

senario

thro

ug

hp

ut

BASE

SACK

LT

ECN

Page 30: TCP performance over ad-hoc mobile networks

17/10/2003 TCP performance over ad-hoc mobile networks.

30

Graphs (cont).nodes Throughput average improvement comparison over DSDV.

DSDV avg. throughput improvement

0.980.9850.99

0.9951

1.0051.01

1.0151.02

1.0251.03

30 15 5

Velocity

% t

hro

ug

hp

ut

imp

rovm

en

t

SACK

LT

ECN

Page 31: TCP performance over ad-hoc mobile networks

17/10/2003 TCP performance over ad-hoc mobile networks.

31

Ad-hoc On-demand Distance Vector

Page 32: TCP performance over ad-hoc mobile networks

17/10/2003 TCP performance over ad-hoc mobile networks.

32

Configurations

• Environment configurations: area size600m*400m, total nodes 11, 6 sources.

Velocity 30m/s

145000

150000

155000

160000

165000

170000

1 2 3 4 5 6 7 8 9 10

senario

thro

ug

hp

ut AODV_BASE_30

AODV_SACK_30

AODV_LR_30

AODV_ECN_30

Page 33: TCP performance over ad-hoc mobile networks

17/10/2003 TCP performance over ad-hoc mobile networks.

33

GraphsVelocity 15m/s

150000

155000

160000

165000

170000

175000

1 2 3 4 5 6 7 8 9 10

senario

thro

ug

hp

ut AODV_BASE_15

AODV_SACK_15

AODV_LR_15

AODV_ECN_15

Velocity 5m/s

150000

155000

160000

165000

170000

175000

1 2 3 4 5 6 7 8 9 10

senario

thro

ug

hp

ut AODV_BASE_5

AODV_SACK_5

AODV_LR_5

AODV_ECN_5

Page 34: TCP performance over ad-hoc mobile networks

17/10/2003 TCP performance over ad-hoc mobile networks.

34

SACK improvement comparison.Graphs - SACK improvement over different routing

protocols.

All with the same topology: 600m*400m, 6 sources

out of total 11 nodes.

 

TCP throughput with SACK over routing protocols.

50000

100000

150000

200000

250000

velocity 30 velocity 15 velocity 5

velocity

% t

hro

ug

hp

ut

avg

. im

pro

vmen

t

DSR

AODV

DSDV

Page 35: TCP performance over ad-hoc mobile networks

17/10/2003 TCP performance over ad-hoc mobile networks.

35

SACK improvement comparison.Graphs (cont).

SACK improvment over routing protocols

0.991

1.01

1.021.031.041.05

1.061.071.08

velocity 30 velocity 15 velocity 5

velocity

% t

hro

ug

hp

ut

avg

. im

pro

vmen

t

DSR

AODV

DSDV

Notice the SACK improvement over DSR routing protocol in compare to the improvement over other routing protocols

Page 36: TCP performance over ad-hoc mobile networks

17/10/2003 TCP performance over ad-hoc mobile networks.

36

Temporally Ordered Routing Algorithm

Page 37: TCP performance over ad-hoc mobile networks

17/10/2003 TCP performance over ad-hoc mobile networks.

37

TORA Problematic Character• The run results shows that this routing protocol performance is highly sensitive to topology.

•Therefore conclusive conclusions could not be made regarding TCP performance using this routing protocol.

• The following graph demonstrates the performance sensitivity to topology in TORA routing protocol.

Page 38: TCP performance over ad-hoc mobile networks

17/10/2003 TCP performance over ad-hoc mobile networks.

38

Velocity 15m\s

0

20000

40000

60000

80000

100000

120000

140000

1 2 3 4 5 6 7 8 9

scenario

thro

ug

hp

ut BASE

SACK

LT

ECN

TORA Problematic Character.Graph -

1500mx1000m, 25 nodes, 10 sources.

Page 39: TCP performance over ad-hoc mobile networks

17/10/2003 TCP performance over ad-hoc mobile networks.

39

Conclusions

Page 40: TCP performance over ad-hoc mobile networks

17/10/2003 TCP performance over ad-hoc mobile networks.

40

Selective ACK conclusions• The selective ACK showed the best TCP performance

improvement. Its performance out-goes over all routing protocols and topologies that were examined in this project.

• In ad-hoc mobile networks there is a high probability for packets loss due to high bit error rate, and frequent topology changes. SACK is mainly helpful during multi packets loss in the same sending window.

• SACK had shown up to 12% performance improvement while using high velocities. In high velocities frequent topology changes occur which causes more packets loss.

Page 41: TCP performance over ad-hoc mobile networks

17/10/2003 TCP performance over ad-hoc mobile networks.

41

Limited Transmit conclusions• Limited transmit had poorly throughput performance

in our research (similar to base and some times even lower).

• It’s happened because LT nature is quite aggressive which may contribute to creation of fast congestion.

• Studies has shown that LT is more effective when the duration of the connection is short (browser use http 1.0) because the sender can not probe the bandwidth due to small amount of data to be sent, and we used long and heavyconnections (FTP).

Page 42: TCP performance over ad-hoc mobile networks

17/10/2003 TCP performance over ad-hoc mobile networks.

42

Explicit Congestion Notification conclusions

During our work on the project we have runenormous scenarios and no a real improvementusing ECN was viewed. There are some reasonsthat could cause to that situation:

•RED is strong enough for signaling the sources for congestion and therefore ECN can not contribute dramatically.

Page 43: TCP performance over ad-hoc mobile networks

17/10/2003 TCP performance over ad-hoc mobile networks.

43

Explicit Congestion Notification conclusions (cont).

•A quite high loss packets environment as in the ad hoc mobile network causes loss of TCP ACK packets with the ECN field set which reduce the power of ECN.

• RED seems to be difficult to tune, and adding ECN to RED in mobile ad hoc networks were the topology is very dynamic make it more difficult to use it in an optimal manner (we tried to tuned RED queue parameters and still no significant result).

Page 44: TCP performance over ad-hoc mobile networks

17/10/2003 TCP performance over ad-hoc mobile networks.

44

Topology Size conclusions.

As seen on run results, the throughput depends on network dimensions. On the contrary to our first assumption the throughput increased when network dimensions grew, however the throughput behavior is not linear

Throughput as function of area size.

0

50000

100000

150000

200000

400*500 1500*1000 2500*2000 5000*4000

area size

thro

ug

hp

ut