Abdul Jabbar Mohammad, Said Zaghloul*,

1
TCP Performance over Multilink PPP in Wireless
Networks Theory and Field Experiences

• Abdul Jabbar Mohammad, Said Zaghloul,
• and Victor S. Frost
• Information and Telecommunication Technology
  Center
• Electrical Engineering Computer Science
• frost_at_eecs.ku.edu, 785-864-4833

Currently working at Sprint in Network Services
Gateway in the Data Network/Network Development
department for Brent Scott
2
Outline

• Motivation
• Technology
• Field Experiments
• Analytic Prediction of TCP over MLPPP with Call
  Drops

3
Motivation

• Polar Radar for Ice Sheet Measurements (PRISM)
• The communication requirements of PRISM field
  experiments in Greenland and Antarctica
• Data telemetry from the field to the University
• Access to University and web resources from field
• Public outreach
• Mainstream communication system for polar science
  expeditions, field camps in Arctic/Antarctic and
  other research purposes
• Government and security use
• Solution
• Implement a multi-link point-to-point Iridium
  communication system to combine multiple links to
  obtain a single logical channel of sufficient
  aggregate bandwidth.

4
Technologies-Iridium
5
Technologies-Protocols
6
Technologies-Multi-Link Point-to-Point Protocol

• Multilink option are negotiated when establishing
  the connection.
• Packets may be fragmented.

Network Layer
Network Layer
MLPPP
MLPPP
MLPPP PDUs reassembled into the original layer 3
packet
MLPPP PDUs
2
4
5
Link 1
Layer 3 Packet
6
1
6
Link 2
MLPPP fragments layer 3 packets
3
7
Link n
MLPPP fragments have non-decreasing sequence
numbers
2
Sender
Receiver
7
Technology- Network Architecture
8
Summer 2003 Field Experiments
Test Location
9
2003 Results Throughput
Method 1 Modem 2 Modems 3 Modems 4 Modems
Iperf 2.1 4.0 7.0 9.6
Iperf 1.9 3.9 7.0 9.3
Iperf 1.7 4.5 6.8 9.7
Ttcp 2.29 4.43 6.6 8.9
Ttcp 2.48 4.40 7.0 8.78
Average 2.1 4.25 6.88 9.26

• Tools used TTCP, IPERF
• Throughput varies to some extend due to RTT
  variation
• Efficiency gt 90

Effective throughputs during large file transfers
File Size (MB) Upload Time (min) Throughput (bits/sec)
0.75 11 9091
3.2 60 7111
1.6 23 9275
2.3 45 6815
1.5 28 7143
2.5 35 9524
(K b p s)
10
Applications Uploads and Downloads

• Files were downloaded to support the science and
  operations of the camp. The importance of each
  file to the user is noted on a subjective scale
  of 1-10,10 being the most valuable.

Title Downloaded/uploaded Size Imp
1 Spectrum Analyzer programmers Manual Download from Agilent.com 7.2MB 9
2 Matlab Programs Download from ITTC 500KB 7
3 Voltage regulator data sheet Download from Fairchild.com 226KB 9
4 GPS software Download 800KB 9
5 Proposal submission Upload 600KB 8
6 Access point manager software Download from Orinoco.com 4.66MB 7
7 Drawing of machine spares to order Upload to University of Copenhagen 1MB 9
8 Video of core, datasheet Upload for press release 2MB 8
9 Pictures, press release of longest core in Greenland Upload to Kangerlussauq for press release 500KB 6
11
Eight Modem Iridium System 2004/5 Field
Experiments

• The Modem/Computer box is a 19 rack mount 5U
  equivalent
• The front panel is 8.72 tall and 19 wide. The
  sides are 8.34 tall and 24 deep.
• Weight approximately 45lbs.
• Reproduction cost 18,000

Iridium Modems
Ethernet
USB
12
Field Experiments System Implementation
8-Channel system in a weather-port at SUMMIT camp
in Greenland, July 2004
13
Field Experiments Antenna Setup
4 ft
10 ft
8 Antenna setup at SUMMIT camp in Greenland, July
2004
14
Results Throughput

• Average throughput efficiency was observed to be
  95
• The above results are from the test cases where
  no call drops were experienced
• In event of call drops the effective throughput
  of the system will be less than the above values

15
Results Throughput
FTP throughput observed during data transfer
between the field camp and KU
Size of file in MB Approx. Upload Time Effective Throughput in Kbps
1.38 01124 16.53
3.77 03542 14.42
5.62 04612 16.61
15.52 23000 14.12
20.6 30000 15.62
35.7 51500 15.47
55.23 90000 13.96

• Average throughput for FTP upload of large files
  was observed to be 15.38 Kbps
• Due to call drops, the efficiency was reduced to
  80

16
Results Round Trip Time

• Average RTT 1.4 sec
• Minimum observed RTT 608 msec
• Mean deviation 800 msec

17
Results Reliability 14th July 12-hr test

• Call drop pattern during 8 Iridium 8 Iridium
  DAV mode test for 12 hrs
• Percentage uptime with full capacity (8 channels)
  is 89 and with at least one modem is 98
• Total number of primary call drops during 12 hrs
  4

18
Results Reliability 22nd July 32-hr test
Uptime

• Call drop pattern during 8 Iridium 8 Iridium
  DAV mode test for 32 hrs
• Percentage uptime with full capacity (8 channels)
  is 85 and with at least one modem is 96
• Total number of primary call drops during 32 hrs
  24

19
Results Reliability 19th July 6-hr test

• Call drop pattern during 8 Iridium 8 PSTN data
  mode test for 32 hrs
• Percentage uptime with full capacity (8 channels)
  is 67 and with at least one modem is 90
• Total number of primary call drops during 6 hrs
  9

20
Results Mobile tests
Iridium antennas
Experiments monitored from another vehicle
through 802.11b link
Iridium system mounted in an autonomous vehicle
(MARVIN)
21
Results Mobile tests

• Call drop pattern during 8 Iridium 8 Iridium
  DAV mode test for 2 hrs
• Percentage uptime with full capacity (8 channels)
  is 65 and with at least one modem is 92
• Average time interval between call drops is 45
  mins
• Average throughput 18.6 Kbps, Average RTT 2
  sec

22
2004 Applications

• Summer 2004 field experiments
• Communications data upload up to 40 MB files
• Radar data uploads up to 55 MB files
• Text chat with PRISM group at KU
• Video conference - real time audio/video
• Individual audio or video conference works with
  moderate quality with the commonly available
  codecs
• Outreach Use
• Daily Journal logs uploaded
• Daily Pictures uploaded
• Video clips uploaded
• Held video conference with science teachers/
  virtual camp tour
• Wireless Internet access

23
2005 WAIS Field Experiments
WAIS- West Antarctica Ice Sheet
24
2005 WAIS Field Experiments
Item Size Importance
Component data sheets 2.2 MB 8
Oscilloscope lab measurements 2 MB 9
Modified code for SAR measurements 500 KB 10
C IDE 10 MB 5
GIS scripts 2 MB 7
GPS troubleshooting manual 10 MB 10
PICO editor 4 MB 4
Outreach pictures, journal and weather data upload 500 KB/day 10
Video conference Variable 6
Virtual dashboard application 50 MB 9
Critical data internet search by the drilling team Variable 10
Internet/email access to all field personnel at WAIS camp Variable 7
Remote ssh access to field programs from KU Variable 9

• Applications Uploads and Download
• Files were downloaded to support the science and
  operations of the camp. The importance of each
  file to the user is noted on a subjective scale
  of 1-10,10 being the most valuable.

25
Analytic Prediction of TCP over MLPPP with Call
Drops

• A call drop is the event of losing an established
  connection suddenly
• Connections are automatically re-established
• It was observed that a call drop results in TCP
  timeouts
• Various reasons that might lead to call drops,
• Low signal level
• Failure of the inter-satellite handovers
• Goal Predict the throughput as a function of
  drop rate and other system parameters
• First step Call Drop model

26
Call Drops - Distribution

• 394 call-drop measurements were collected in the
  field
• Call drop pdfexponential

• The single link ICTD is a Poisson random process
  with a rate b
• A N Link bundles ICTD is a Poisson process with
  a dropping rate of

ICTD PDF based on GreenlandKansas measurements.
Estimated exponential distribution
(0.02exp(-0.02t)) passes the chi-square
goodness-of-fit test (5 significance level and
14 bins)
lN b
27
Call Drops - Distribution

• KS-Greenland call dropping rate per link is 1/50
  min-1
• KS-KS call dropping rate per link is 1/52 min-1

28
TCP Performance Model

• TCP transfer latency for fs bytes given the MSS
  is
• To estimate TCP throughput (B) in packets/sec
• Evaluate the throughput if no timeouts take place
  
• Extend the no timeout throughput using the
  empirical call drops PDF to include timeouts
• Main Assumptions
• Packet losses are due to ARQ failures (no
  timeouts)
• Timeouts are caused by call drops only

29
Methodology

• Modify exiting results to account for call drops
  J. Padhye, V. Firoiu, D. Towsley, and J.
  Kurose, Modeling TCP Reno performance a simple
  model and its empirical validation, IEEE/ACM
  Trans. Networking, vol. 8, pp. 133-145, Apr.
  2000.
• For details of the modification see Modeling TCP
  Long File Transfer Latency over Long Delay
  Wireless Multilink PPP, Said Zaghloul, Victor
  Frost, Abdul Jabbar Mohammad IEEE Communications
  Letters, Vol. 9, No. 11 November 2005, pp.
  988-990.

30
Model Validation with Experiential data

• File Transfers from Greenland to the University
  of Kansas (Summer 2004),T0 60s, p 5E-4, ?
  1/50 min-1, MSS 1448, RTT 19s, Wmax 47.9KB

31
Case Study Increased Dropping Rates

• A software module was built and added to the
  developed link management software to increase
  real drop rate
• The added module generates call drops according
  to a Poisson process for any given dropping rate

32
Case Study Effect of Wireless Errors

• A wireless error refers to the errors that the
  physical layer ARQ could not handle
• Effect is amplified for low bandwidth-long delay
  connections (ex. Iridium)
• An efficient ARQ mechanism minimizes wireless
  errors

• Inmarsat GEO, bundle 4
• RTT 0.61 s, BW 128 kbps
• MSS1 KB and Wmax40 KB
• A slight increase of the packet loss probability
  results in approximately 25 min increase in the
  transfer time

33

Information and Telecommunication Technology
Center