Nziga 1
Figure 7.31 CWND – 1Mb – 5ms – DSDV - Reno
Figure 7.32 CWND – 10Mb – 5ms – DSDV - Reno
Figure 7.33 CWND – 1Mb – 2ms – DSDV - Newreno
Figure 7.34 CWND – 1Mb – 5ms – DSDV - Newreno
Figure 7.35 CWND – 10Mb – 5ms – DSDV - Newreno
Figure 7.36 CWND – 1Mb – 2ms – DSDV - Sack
Figure 7.37 CWND – 1Mb – 5ms – DSDV - Sack
Figure 7.38 CWND – 10Mb – 5ms – DSDV - Sack
Figure 7.39 CWND – 1Mb – 2ms – DSDV - Fack
Figure 7.40 CWND – 1Mb – 5ms – DSDV - Fack
Figure 7.41 CWND – 10Mb – 10ms – DSDV - Fack
Figure 7.42 CWND – 5.5Mb – 2ms – DSDV - Vegas
Figure 7.43 CWND – 5.5Mb – 5ms – DSDV - Vegas
Figure 7.44 CWND – 5.5Mb – 10ms – DSDV - Vegas
7.5.5 Round Trip Time of the communication
To keep the size of this thesis at a manageable level, we satisfy the curiosity of the reader by inserting just the RTT graph for each TCP flavor given the following characteristics:
-10Mb of Bandwidth
-5 ms of Link Delay
-40 KM/H for the speed of the Mobile Node.
How to read the graphs? Each data point corresponds to one RTT. The gaps in the RTT graph are caused by retransmissions. The reader can note some similitude among the following TCP flavors:
-Tahoe
-Reno
-Newreno
-Sack
TCP Fack and Vegas have a particular behavior.
Figure 7.45 RTT – 10Mb – 5ms – DSDV - Tahoe
Figure 7.46 RTT – 10Mb – 5ms – DSDV - Newreno
Figure 7.47 RTT – 10Mb – 5ms – DSDV - Reno
Figure 7.48 RTT – 10Mb – 5ms – DSDV - Vegas
Figure 7.49 RTT – 10Mb – 5ms – DSDV - Fack
Figure 7.50 RTT – 10Mb – 5ms – DSDV - Sack
7.5.6: Impact of Speed on TCP Performance
Table 7.3 recapitulates the number of packets received for different speed values of the MN for a better exploitation. The corresponding graphs are posted just below the table. (From Figure 7.51 to Figure 7.62). The reader can notice:
-Some discrepancies
-Some improvement when the MN is moving at 60 KM/H.
Table 7.3 Packet Number received by the Mobile Node
Bandwidth(Mb) / Link
Delay
(msec) / Speed
MN
(KM/H) / TCP
Tahoe
(Num-ber) / TCP
Reno
(Num-
ber) / TCP
Newreno
(Num-
ber) / TCP
Vegas
(Num-
ber) / TCP
Fack
(Num-
ber) / TCP
Sack
(Num-
ber)
10 / 2 / 40 / ?
5000 / ?
5000 / ?
5000 / 5400 / ?
5000 / ?
5000
10 / 5 / 40 / 7000 / 7000 / 7000 / 5000 / ?
5700 / 7000
10 / 10 / 40 / 7500 / 7500 / 7500 / ?
4000 / 7500 / 7500
10 / 2 / 60 / 4800 / 4800 / 4800 / 3700 / 4800 / 9000
10 / 5 / 60 / 9000 / ?
7200 / ?
7200 / 3400 / 8700 / 4750
10 / 10 / 60 / 4700 / 4700 / 4700 / 2800 / 4800 / 4750
10 / 2 / 80 / 4750 / 4750 / 4750 / 3600 / 4750 / 4750
10 / 5 / 80 / 4750 / 4750 / 4750 / 3000 / 4750 / 4750
10 / 10 / 80 / 4750 / 4750 / 4750 / 2700 / 4750 / 4750
From the table 7.3, we observe the following:
1-For the first time, we notice that for some TCP flavors ((Tahoe, Vegas, Fack and Sack at 60 KM/H), (all the flavors at 80 KM/H)), we are certain that whatever the congestion on the link is, the mobile node will not be disconnected.
2-Using TCP Reno and TCP Newreno, MN is disconnected for 5 ms of link delay value at 60 KM/H.
3-Using all the 6 TCP flavors, packets received by the mobile node when the connection is not interrupted are bigger at the speed value of 60 KM/H. This leads us to the next observations.
4-TCP Vegas, TCP Fack and TCP Sack may be used as the transport protocol in order to send the maximum number of packets to the mobile node.
5-The value of 60 KM/H seems to be the speed at which the car should be going during the communication in order to receive the maximum number of packets from the fixed correspondent.
6-If the zone is more congested, TCP Tahoe and TCP Fack are advisable.
7-If it is the opposite, we might use TCP Sack.
Figure 7.51 Pkt_id – 10Mb – 2ms, 5ms, 10ms – DSDV – Tahoe – 60 KM/H
Figure 7.52 Pkt_id – 10Mb – 2ms, 5ms, 10ms – DSDV – Tahoe – 80 KM/H
Figure 7.53 Pkt_id – 10Mb – 2ms, 5ms, 10ms – DSDV – Reno – 60 KM/H
Figure 7.54 Pkt_id – 10Mb – 2ms, 5ms, 10ms – DSDV – Reno – 80 KM/H
Figure 7.55 Pkt_id – 10Mb – 2ms, 5ms, 10ms – DSDV – Newreno – 60 KM/H
Figure 7.56 Pkt_id – 10Mb – 2ms, 5ms, 10ms – DSDV – Newreno – 80 KM/H
Figure 7.57 Pkt_id – 10Mb – 2ms, 5ms, 10ms – DSDV – Vegas – 60 KM/H
Figure 7.58 Pkt_id – 10Mb – 2ms, 5ms, 10ms – DSDV – Vegas – 80 KM/H
Figure 7.59 Pkt_id – 10Mb – 2ms, 5ms, 10ms – DSDV – Fack – 60 KM/H
Figure 7.60 Pkt_id – 10Mb – 2ms, 5ms, 10ms – DSDV – Fack – 80 KM/H
Figure 7.61 Pkt_id – 10Mb – 2ms, 5ms, 10ms – DSDV – Sack – 60 KM/H
Figure 7.62 Pkt_id – 10Mb – 2ms, 5ms, 10ms – DSDV – Sack – 80 KM/H
Table 7.4 below recapitulates the number of window’s size reduction based on the speed of the MN. Some selected graphs are posted below the table, two for each TCP flavor. (From Figure 7.63 to Figure 7.74). The reader can notice:
-Some discrepancies
-Some improvement when the MN is moving at 60 KM/H with the pause tending to disappear (pay a careful look to Figures 7.63, 7.69 and 7.71 where the congestion window looks like the one seen for the TCP behavior on wired network).
Table 7.4 Number of window’s size reduction during the connection
Bandwidth(Mb) / Link
Delay
(msec) / Speed
MN
(KM/H) / TCP
Tahoe
(Number) / TCP
Reno
(Number) / TCP
Newreno
(Number) / TCP
Vegas
(Number) / TCP
Fack
(Number) / TCP
Sack
(Number)
10 / 2 / 40 / 3 / 3 / 3 / 7 / 2 / 3
10 / 5 / 40 / 4 / 3 / 3 / 6 / 2 / 4
10 / 10 / 40 / 2 / 2 / 2 / 6 / 2 / 2
10 / 2 / 60 / 2 / 2 / 2 / 5 / 1 / 2
10 / 5 / 60 / 3 / 4 / 5 / 5 / 3 / 3
10 / 10 / 60 / 1 / 1 / 1 / 5 / 1 / 1
10 / 2 / 80 / 2 / 2 / 2 / 5 / 2 / 2
10 / 5 / 80 / 2 / 2 / 2 / 5 / 2 / 2
10 / 10 / 80 / 1 / 1 / 1 / 5 / 1 / 1
Figure 7.63 CWND – 10Mb – 5ms – DSDV – Tahoe – 60 KM/H