WAP Downlink Performance Evaluation in UMTS Network

1
WAP Downlink Performance Evaluation in UMTS
Network

• Author Pertti Hakkarainen
• Supervisor Prof. Jorma Virtamo
• Instructors MSc Jani Kokkonen, Dr Samuli Aalto
• Work was carried out Nokia Networks, Espoo
• Thesis number 1033 2004
• Presentation date November 9, 2004

2
Table of contents

• Introduction
• Research Problem
• WAP
• Wireless TCP/IP
• Measurement Setup
• Measurement Results
• Conclusion

3
Introduction

• The study is the downlink performance measurement
  in the Universal Mobile Telecommunication
  Services (UMTS) network
• Technical specification of the network is based
  on 3rd Generation Partnership Project (3GPP)
  Release 99

4
Research Problem

• The object of this study is to compare the
  download performance of the Wireless Application
  Protocol (WAP) and TCP/IP protocol WAP
  applications in the 3G network.
• The protocols are designed for different network
  domains
• WAP protocol stack for circuit switched networks
  where the bandwidth is relatively low
• TCP/IP protocols for packet switched networks
  where the bandwidth is considerably higher
• The aim of this work is to study how well
  different protocols can utilize the performance
  of the 3G network.

5
WAP 1/2

• Why WAP?
• The WAP specification is designed to bring
  Internet access to the wireless environment.
• Wireless data networks present a more constrained
  communication environment compared to wired
  networks.
• Handheld wireless devices present a more
  constrained computing environment compared to
  desktop computers (CPU, memory, power supply,
  display, keypad etc.).
• WAP-compatible component communicate with all
  other components in the solution network by using
  the standard methods and protocols defined in the
  WAP specification

6
WAP 2/2

• WAP version 1.2
• WAP 1.2 is based on WAP protocols
• WAP gateway makes the conversion from text to
  binary and vice versa
• The plain text headers of HTTP are translated
  into binary code that significantly reduces the
  amount of data that must be transmitted over the
  air interface.
• Complete end-to-end security cannot be guaranteed
  due to a security gap in the GW

• WAP version 2.0
• WAP 2.0 is based on TCP/IP
• WAP Gateway is acting as a WAP 2.0 proxy
• No security gap, security is comparable to the
  Internet model transaction all the way from WAP
  Device to the Web server will be secured

7
Wireless TCP/IP

• The Transmission Control Protocol/Internet
  Protocol (TCP/IP) forms the basis of the
  Internet. Originally it is designed and optimized
  to provide reliable byte transfer with
  retransmission in a terrestrial environment where
  conditions are relatively stable compared to the
  wireless environment. The assumption is that a
  packet loss is due to congestion in the network.
• The wireless networks have brought new aspects to
  the packet loss.
• More narrow and variable bandwidth
• Higher bit error rate
• More latency
• More transmission delay
• Less connection stability
• Less predictable availability
• Since TCP/IP operates over different kinds of
  link conditions, from stable and fast wireline
  links to delay sensitive wireless links, there
  may be situations that performance degrades due
  to the fact that the optimization is done in a
  different way on different parts of the network
  and packet delays that are treated as congestion
• wireless TCP/IP provides the same functions as
  the normal TCP/IP with some optimization in the
  protocols
• The round-trip time (RTT) is one of the most
  essential issues that has affected to wireless
  TCP optimization.

8
Measurement setup 1/2

• The target of the measurement has been to measure
  WAP throughput with different WAP versions, WAP
  1.2 and WAP 2.0, to the download direction and
  evaluate that network is capable to carry traffic
  with nominal throughputs.
• Tools
• Ostrich is a trace interface adapter device that
  provides mechanisms for buffering the tracing and
  debugging data sent by the phone.
• EARP is a PC application for decoding and logging
  trace data during the testing and debugging of
  user equipment.
• Measurement is carried out in the Nokia WCDMA
  test network.
• During the measurement there were not other
  traffic in the network

Test network setup
UE
Gateway
Iu-PS
Iub
Gn
Gi
RNC
3G SGSN
3G GGSN
Firewall
Switch
BS
Ostrich
EARP
RTT
Content Server
9
Measurement setup 2/2

• Measurement was conducted with 13 different file
  sizes, ranging from 2 kilobytes to 160 kilobytes
• Downlink rates are 64 kbps, 128 kbps and 384
  kbps, for uplink 3G Partnership Project Release
  99 defines 64 kbps
• Used phones
• Nokia 6650 for WAP protocol measurement
• Nokia 7600 for TCP/IP measurement

Nokia 7600
Nokia 6650
10
Measurement results 1/4

• WAP 1.2 used segmentation and reassembly (SAR)
  value of 5
• SAR defines a method for a WAP gateway to break a
  large message into small chunks (the
  segmentation) and for the phone to piece it back
  together (the reassembly)
• SAR requires the UE to acknowledge to the WAP
  gateway after every fifth packet
• Operator controlled value and optional
• The measurement starts when the user has selected
  the file for download and presses the button on
  the phone in order to receive it. The connection
  for the download is already created.
• The measurement stops at the reception of the
  last octet of the file.

11
Measurement results 2/4

• Calculation for average and maximum download
  throughput
• Average throughput is the download period from
  the very first octet of the file until the last
  octet of a particular file is received including
  the slow start
• The maximum throughput calculation is done over
  the stabilized transfer part of the download
  (figure)

• WAP protocol has the segment size of 1412 bytes
  gt 7060 bytes requires ACK
• In TCP/IP the Maximum Segment Size (MSS) is 1460
  bytes and UE allows 20 segments per window

12
Measurement results 3/4

• WAP 1.2 shows quite poor download utilization in
  3G network
• The reason is SAR value that requires UE to
  acknowledge to the gateway
• If SAR is not used and there is failure in
  transmission, the whole file must be
  retransmitted, the bigger the file is the more
  retransmission deteriorates the throughput

Max. WAP 1.2 throughput Max. WAP 1.2 throughput Max. WAP 1.2 throughput
Nominal bit rate kilo bits/s 64 128 384
kilo bits/s 41.613 54.654 97.172
kilo bytes/s 5.080 6.672 11.862
Utilization 65.0 42.7 25.3

• WAP 1.2 designed for circuit switched environment
  and hence it can not utilize the packet switched
  environment features

13
Measurement results 4/4

• Wireless TCP/IP utilizes the link connection
  pretty well, the pipe fill is optimal
• Wireless TCP/IP is designed for the packet
  switched environment

Max. WAP 2.0 throughput Max. WAP 2.0 throughput Max. WAP 2.0 throughput
Nominal bit rate kilobits/s 64 128 384
kilobits/s 63.650 125.095 375.694
kilobytes/s 7.770 15.271 45.861
Utilization 99 98 98

• For small packets the WAP 1.2 throughput is
  better than WAP 2.0 throughput
• When using the SAR value five in WAP 1.2, the
  better throughput in favor of WAP 2.0 happens
  between 7 and 10 kilobytes

14
Average throughputs

• Average throughput remains below the maximum
  throughput

• WAP 1.2 average throughputs for different bearer
  rates

WAP 2.0 average throughputs for different bearer
rates
15
Conclusion

• What we found out in our measurement?
• The WAP protocol is beneficial when a small
  amount of data is downloaded, less than 10
  kilobytes
• For larger file size wireless TCP gains better
  performance
• The proportion of the slow start from whole
  downloaded data on the time scale decreases as
  the file size grows
• The transition in favor of wireless TCP depends
  on the algorithms and parameters used in both
  protocols
• In our measurement, the wireless TCP gained
  better download performance around 10 kilobytes
  file
• The 3G network is mature to achieve and maintain
  nominal transfer rates with WAP applications when
  using WAP 2.0 in stable environment