Inthe beginningofthe eahexperimentaltrialthemobileterminaluses thepointof
attahment made available by its dial-uponnetion toexhange WAP tra over
a 9600bps GSM-CSD link. At time 20se the mobile terminal performes a single
Mobile IP hand-o that auses the ommuniation to proeed through the point
of attahment made available by the 11Mbps IEEE 802.11b Wireless LAN. Figure
5.4 illustrates the paket trae of the aforementioned senario. It an be observed
that the paket transmission rate beomes signiantly faster following the Mobile
IP hand-othatswiftlyompleteswithoutany paketloss. Thisisderived fromthe
fat that nopaketretransmission isobserved afterthe hand-o.
The lakof anyow ontrolorwindowing onsiderationsinthe seondWTP
trans-ationservielassprovidesaonstanttransmissionratebeforeandaftertheMobile
IP hand-o. However, due to logsaling of the y-axis this an not be observed in
gure 5.4. For this purpose, the Round Trip Time (RTT) measurements of the
ommuniation are illustrated in gure 5.5. That is, for every WTP data paket
transmitted by the WAP Gateway, the mobileterminal wasfored to respond with
an immediateWTP aknowledgement. The time intervalbetween the transmission
of the data paket and the reeipt of the orresponding aknowledgement was
on-sidered equivalent tothe RTT. It an be seen that the RTT remainsonstant over
GSM-CSD with an average of 2.438se, while the orresponding value over IEEE
802.11b has been found to be 0.006se with sporadi deterioration due to internal
proessing. In any ase, the aeleration of the RTT justies the inrease in the
transmission rate.
An additionalpointof interest is the RTT of the rst WTP transmission following
1 10 100 1000 10000
0 5 10 15 20 25 30 35 40
W T P p a c k e t s e q u e n c e n u m b e rs
Time [sec]
Progress of a WAP communication with a Mobile IP hand-off
WTP data packet
Figure 5.4: Paket Trae of a WAP Communiation during aMobile IP Hand-o
between GSM-CSD and IEEE 802.11b
0.001 0.01 0.1 1 10
0 5 10 15 20 25 30 35 40
R o u n d T ri p T im e [s e c ]
Time [sec]
RTT of a WAP communication with a Mobile IP hand-off
Figure 5.5: RTT of aWAP Communiation duringa Mobile IP Hand-o between
600 650 700 750 800 850 900 950
0 5 10 15 20
R o u n d T ri p T im e [s e c ]
Time [sec]
Round Trip Time of Ping between mobile terminal and WAP Gateway over GSM
Figure 5.6: Round TripTime of Ping between the Mobile Terminal and the WAP
Gateway overGSM-CSD
RTToverGSM-CSDbutstillgreaterthanthatofIEEE802.11b. Thissigniesthat
this paket exhange ourred during the Mobile IP hand-o. That is, the mobile
terminal reeived a WTP data paket through its GSM-CSD onnetion but sent
its aknowledgement through IEEE 802.11b.
InordertodeterminetheindividualdelaysthatauseWTP tosuerextendedRTTs
overGSM,theping appliationwasusedtodeterminetheGSM-CSDtraversaldelay
betweenthemobileterminalandtheWAPGateway. Theresultsofthisinvestigation
are presented in gure 5.6. ping has been found to demonstrate a onstant RTT
with anaverage of 821ms. Considering that at 9600bps GSM-CSD requires 107ms
for the transmission of the ping request and reply messages, the rest of 714ms is
onsidered to be the traversal delay of the GSM network. This interval is roughly
100ms larger than the orresponding delay measured in other investigations [90℄.
However, the traversal delay of GSM-CSD has been found to be network provider
spei [144℄.
For the transmission of a full WTP data paket and its orresponding
aknowl-edgement, GSM-CSD requires 1.2se and 0.025se respetively. Considering an
additional714ms forthe GSM-CSDtraversal delay,fromthe2.438se averageRTT
there isanoutstanding amount ofalmost 0.5se. This an be explainedby the fat
thatthe GSM-CSDtraversal delayhasbeen foundtovaryduringbulkdatatransfer
and takeon values as high as1.2se [144℄.
The Stop&Waitharater oftheseondWTP transationservielassditatesthat
a single WTP data paket may be transmitted per RTT. The throughput of the
WAP ommuniationisdiretlydependent onthis performane. Thethroughputof
the WAP ommuniation is illustrated in gure 5.7. As expeted, it inreases
sig-niantly beyond the MobileIP hand-o,wherethe mobileterminalresumes aess
1000 10000 100000 1e+06 1e+07
0 5 10 15 20 25 30 35 40
T h ro u g h p u t (u s e fu l d a ta /r o u n d tr ip ti m e ) [b it s /s e c ]
Time [sec]
Throughput of a WAP communication with a Mobile IP hand-off
Figure 5.7: Throughput of aWAP Communiation duringa Mobile IP Hand-o
between GSM-CSD and IEEE 802.11b
the ommuniation the shortomings of the seond WTP transation servie lass
start to beome apparent.
Communiation eieny is dened as the throughput of the ommuniation over
the available bandwidth. Considering that over IEEE 802.11b, the WAP Gateway
may transmit a single full WTP data paket every 6ms, then the theoretial
max-imum throughput is about 1.78Mbps. However, the IEEE 802.11b Wireless LAN
provides onnetivity at 11Mbps. From this it is determined that the lak of ow
ontrol in WTP does not enable it to utilise high bandwidth rates when they are
available. This is also veried by results presented ingure 5.8. Similarly,in links
with smallbit rates but long RTTs (e.g. GSM-CSD) WTP still manages a low
ef-ieny. It an be seen that even before the Mobile IP hand-o, WTP managesto
provide eieny short of 50%.