Performance Validation of Performance Validation of Mobile IP Wireless NetworksMobile IP Wireless Networks

Presented by

Syed Shahzad Ali

Advisor Dr. Ravi Pendse

Agenda

ObjectiveMotivationIntroductionProblems : IP Addresses, Location, Routing Solution: Mobile IP ProtocolNs-2 test-bed and resultsRouter’s Lab setup Test-bed and resultFuture Work

Objective

Performance validation of mobile IP based wireless networks by collecting results on NS-2 simulator and by working in the LAB.

Validate the effects of mobility on the transport layer

Implement a test bed for the measurement of various parameters that might impact wireless network performance

Motivation

Increasing mobile workforceDramatic improvement in size, weight, and

sophistications of notebook computersTo correctly predict current and future use

of Internet technology

Wireless Networking

Wireless networking is the use of Radio Frequency (RF) technology to connect workstations in a Local Area Network (LAN) or LAN's in a Wide Area Network (WAN).

This technology allows ethernet speed with a limited or no wired connections. It transmits and receives information over the air.

What is Mobile IP?

MIP defines how MN change their Point Of Attachment to the Internet without changing their IP address

Each MN is always identified by its home IP address

HA sends datagrams for MN through a tunnel to the COA

No protocol enhancement is required in other hosts or routers

Routing decisions are based upon the network-prefix portion of the IP Destination Address

What if Node Moves

10.10.0.1/16

130.10.1.0/24

Router

Problem Due to Mobility

Internet

Mobile Node

130.1.1.1130.1.x.x

AA BB CC

DD

?

130.1.1.1

Route Next 130.1.x.x BRoute Next 130.1.x.x B

Route Next 130.1.x.x ARoute Next 130.1.x.x A

CN

Internet

130.1.x.x

A B C

D Route Next 130.1.x.x BRoute Next 130.1.x.x B

Route Next 130.1.x.x ARoute Next 130.1.x.x A

CN

Mobile IP Process

Agent Discovery to find AgentHome agent and foreign agents periodically

advertise agent advertisementsThey also respond to solicitation from mobile

nodeMobile Node selects an agent and uses Care-

Of-Address for further communication

IP Address associated with a mobile node that is visiting a foreign network

It generally changes every time MN moves from one foreign network to another

A FA COA can be any one of the FA’s IP Address

A FA COA can be shared by many MN simultaneously

Care-Of-AddressCare-Of-Address

Mobile IP Process (Contd.)

RegistrationMobile Node registers its COA with home

agent either directly or through foreign agent.Home agent then sends a reply back to mobile

node via FA about successful COA registrationEach mobility binding has a negotiated life time

limitTo continue further working in FN, registration

must be done within lifetime

Overview of Mobile IP Functionality

1. MN discovers agent2. MN obtains COA (Care Of Address)3. MN registers with HA4. HA tunnels packets from CN to MN5. FA forwards packets from MN to CN

HA FA

1. and 2.1. and 2. 3.3.MN

CN

5.5. 4.4.

Encapsulation/ Tunneling

Home Agent Intercepts mobile node’s datagrams and forwards them to COA

Home Agent tells all hosts to send mobile node’s datagram to it

Home Agent then send it to FA via TunnelDecapsulation: Datagram is extracted by

FA and sent to mobile node

Encapsulation/Tunneling

IP Cloud Home Network

New Network/Foreign Network

Home AgentForeign Agent

Mobile Node

Mobile Node

Correspondent Node

TunnelTunnel

Network Simulators

Simulation: a common technique used by researchers

Test scenarios that yet not occurred in real world

Predict performance to aid technology design

Improve validation of the behavior of existing protocols

NS-2 Simulator

Object oriented, discrete event simulatorNS development is supported by DARPA

(Defense Advanced Research Project Agency)NS-2 is written in C++ and Otcl (Object tool

command language)Network scenarios are written into Tcl languageLarge and complex trace files are generatedSoftware package extracts the data from trace fileComplicated and long process

Trace File Format

r 68.423159 3 1 tcp 1020 ------- 2 0.0.0.1 2.0.0.1 6319 12983+ 68.423159 1 2 tcp 1020 ------- 2 0.0.0.1 2.0.0.1 6319 12983- 68.423159 1 2 tcp 1020 ------- 2 0.0.0.1 2.0.0.1 6319 12983r 68.423209029 _6_ AGT --- 12976 tcp 1000 [a2 3 1 800] ------- [0:0 4194305:2 28 4194305] [6316 0] 1 0s 68.423209029 _6_ AGT --- 12996 ack 40 [0 0 0 0] ------- [4194305:2 0:0 32 0] [6316 0] 0 0r 68.423346 4 2 ack 60 ------- 2 1.0.1.2 0.0.0.0 6313 12989+ 68.423346 2 1 ack 60 ------- 2 1.0.1.2 0.0.0.0 6313 12989- 68.423346 2 1 ack 60 ------- 2 1.0.1.2 0.0.0.0 6313 12989r 68.424037 0 1 tcp 1000 ------- 2 0.0.0.0 1.0.1.2 6322 12988+ 68.424037 1 3 tcp 1000 ------- 2 0.0.0.0 1.0.1.2 6322 12988- 68.424037 1 3 tcp 1000 ------- 2 0.0.0.0 1.0.1.2 6322 12988+ 68.424559 4 2 ack 60 ------- 2 1.0.1.2 0.0.0.0 6314 12990- 68.424559 4 2 ack 60 ------- 2 1.0.1.2 0.0.0.0 6314 12990r 68.425271 2 1 ack 60 ------- 2 1.0.1.2 0.0.0.0 6311 12984+ 68.425271 1 0 ack 60 ------- 2 1.0.1.2 0.0.0.0 6311 12984- 68.425271 1 0 ack 60 ------- 2 1.0.1.2 0.0.0.0 6311 12984+ 68.425611 4 2 ack 60 ------- 2 1.0.1.2 0.0.0.0 6315 12992- 68.425611 4 2 ack 60 ------- 2 1.0.1.2 0.0.0.0 6315 12992r 68.426802 2 1 ack 60 ------- 2 1.0.1.2 0.0.0.0 6312 12986+ 68.426802 1 0 ack 60 ------- 2 1.0.1.2 0.0.0.0 6312 12986- 68.426802 1 0 ack 60 ------- 2 1.0.1.2 0.0.0.0 6312 12986+ 68.427576 4 2 ack 60 ------- 2 1.0.1.2 0.0.0.0 6316 12996- 68.427576 4 2 ack 60 ------- 2 1.0.1.2 0.0.0.0 6316 12996r 68.428204 1 0 ack 60 ------- 2 1.0.1.2 0.0.0.0 6307 12977+ 68.428204 0 1 tcp 1000 ------- 2 0.0.0.0 1.0.1.2 6327 12997

Network Setup (NS-2)

MN

HA FA1FA2

TCP Source

Test Bed Setup Flow

Mobile IP Scenario File written in TCL Language

The trace file generated by ns2.

A program written in AWK language grab only TCP information out of the ns2 Trace File. Save the information into another file named tcp_info.tr

Another AWK program examines every line of the tcp_info.tr file and generates latency behavior.

A Program in C++ (delay_avg2.cpp), further normalizes the results

A Program (create.sh) control the whole process of calculating delay out of a raw trace file format of ns2 simulator

Finally a program (final_avg.sh) takes the averages.

Another program automatically email the results to me.

while (!all tcl files run )

Results NS-2

BW Link

Delay Queue Length

Complete Avg. Delay

Avg. Delay

HA-FA1

Avg. Delay

FA1-FA2 Tunnel

Overhead MB ms ms ms ms Ms

1 2 50 63.459 65.856 61.282 43.459 1 2 100 63.413 64.507 31.222 53.353 1 2 150 63.109 64.299 30.39 52.849 1 5 50 85.967 60.021 56.399 59.967 1 5 100 66.331 60.021 50.848 53.331 1 5 150 66.368 59.473 50.316 53.368 1 10 50 67.529 55.957 36.857 49.529 1 10 100 67.431 53.642 36.857 49.431 1 10 150 67.513 55.111 36.921 49.513

Result NS-2 (cont.)

BW Link

Delay Queue Length

Complete Avg. Delay

Avg. Delay

HA-FA1

Avg. Delay

FA1-FA2 Tunnel

Overhead MB ms ms ms ms ms 5.5 2 50 29.29 29.322 30.685 9.381 5.5 2 100 29.619 29.322 30.685 9.718 5.5 2 150 29.619 29.322 30.685 9.718 5.5 5 50 32.909 32.036 34.016 10.45 5.5 5 100 32.459 32.049 34.016 9.559 5.5 5 150 32.459 32.036 34.76 9.559 5.5 10 50 37.458 36.998 38.093 21.004 5.5 10 100 37.458 36.998 38.093 21.004 5.5 10 150 36.998 38.09 36.998 20.544

Result NS-2 (Contd.)

BW Link Delay

Queue Length

Complete Avg. Delay

Avg. Delay

HA-FA1

Avg. Delay

FA1-FA2 Tunnel

Overhead MB ms Ms Ms ms ms 11 2 50 23.076 22.78 24.406 2.954 11 2 100 23.076 22.77 24.321 2.954 11 2 150 23.011 22.67 24.136 2.889 11 5 50 26.027 25.737 27.629 5.3 11 5 100 26.027 25.739 27.631 5.573 11 5 150 26.198 25.322 27.823 5.93 11 10 50 30.984 30.597 31.879 13.418 11 10 100 30.984 30.688 24.548 14.184 11 10 150 30.83 30.597 31.879 11.103

Lab Setup

Server

FA1 FA2

R2

10.x

11.x13.x

12.x

15.x

16.x

17.x

Client

HA

Lab Results

Bandwidth Link Delay Queue Length

Complete Average Delay

Average Delay

HA-FA1

Average Delay

FA1-FA2 Tunnel

Overhead MB ms ms ms ms ms 1 2 50 55.023 50.715 59.156 24.903 1 2 100 64.991 61.795 69.488 34.371 1 2 150 65.87 62.552 69.45 35.75 1 5 50 56.112 50.551 59.767 25.422 1 5 100 65.735 61.93 62.553 35.615 1 5 150 54.427 50.717 58.577 24.287 1 10 50 66.172 61.541 56.612 35.552 1 10 100 66.807 61.913 61.448 36.117 1 10 150 53.948 60.723 64.608 23.828

Lab Results (Cont.)

Bandwidth Link Delay Queue Length

Complete Average

Delay

Average Delay

HA-FA1

Average Delay

FA1-FA2 Tunnel

Overhead MB ms ms ms ms ms 5.5 2 50 42.705 29.398 51.518 26.575 5.5 2 100 44.99 31.164 44.207 28.37 5.5 2 150 44.112 31.125 51.868 27.432 5.5 5 50 47.631 42.736 51.92 31.511 5.5 5 100 45.309 31.37 38.141 29.159 5.5 5 150 43.124 31.124 52.303 27.004 5.5 10 50 42.35 31.062 52.196 25.39 5.5 10 100 46.522 31.18 51.949 30.402 5.5 10 150 35.262 31.017 38.554 19.142

Lab Results (Contd.)

Bandwidth Link Delay Queue Length

Complete Average Delay

Average Delay

HA-FA1

Average Delay

FA1-FA2 Tunnel

Overhead MB ms ms ms ms ms 11 2 50 35.86 39.131 43.389 21.718 11 2 100 34.892 28.824 48.749 20.767 11 2 150 35.02 28.704 37.538 20.895 11 5 50 37.24 39.145 45.621 23.115 11 5 100 34.964 39.257 52.762 20.839 11 5 150 35.891 39.13 54.185 21.765 11 10 50 33.103 39.454 54.328 18.939 11 10 100 37.313 39.068 48.525 23.151 11 10 150 34.187 28.657 52.792 20.062

1MB Bandwidth

0

20

40

60

80

100

1 2 3 4 5 6 7 8 9

Complete AverageDelay LAB ms

Complete Avg.Delay Sim ms

5.5 MB Bandwidth

0

10

20

30

40

50

60

1 2 3 4 5 6 7 8 9

Complete AverageDelay LAB ms

Complete Avg.Delay Sim ms

11 MB Bandwidth

0

10

20

30

40

1 2 3 4 5 6 7 8 9

Complete AverageDelay LAB ms

Complete Avg. DelaySim ms

1 MB Bandwidth

0

10

20

30

40

50

60

70

1 2 3 4 5 6 7 8 9

Tunnel Overhead Lab ms

Tunnel Overhead Simms

5.5 MB Bandwidth

0

5

10

15

20

25

30

35

1 2 3 4 5 6 7 8 9

Tunnel Overhead LABms

Tunnel Overhead Simms

11 MB Bandwidth

0

5

10

15

20

25

1 2 3 4 5 6 7 8 9

Tunnel OverheadLAB ms

Tunnel OverheadSim ms

Conclusion

Overall comparison shows that results are almost same so we can rely on simulator to predict mobile IP wireless networks

In lowest bandwidths, tunnel overhead is more than 50% of the overall network latency

During roaming packets are lost and retransmitted

Conclusion (cont.)

Latencies values in lab setup are higher than collected from NS-2 setup

In lab handoff was achieved between networks by changing SSID of mobile node

In NS-2 handoff is based on received signal power strength

Network performance improved when high bandwidths are used

Conclusion

Strict separation between layer 2 and layer 3 results in increased network latencies

Mobile node may only communicate with a directly connected FA

MN may only begin registration process after layer 2 handoff to a new FA is completed

The registration process takes some time to complete. During this time MN is not able to send or receive IP packets

Future Work

In this research the correspondent node was stationary. A study can be done where CN is also mobile

Study can be done to determine and validate performance with more number of mobile nodes

Real time traffic using RTP protocol may be realized

Thank you very much for attending the presentation

Questions?