AIMD Based Protocols Evaluation

Figure 3.3, Figure 3.4 and Figure 3.5 show the performance results of TCP, WCP and TCP-AP for the mesh scenarios. The presented results are for the 4x4 grid mesh nodes and variable number of mobile nodes. From the results it is possible to observe that WCP has the best performance results. WCP improved AIMD mechanism makes it perform more efficiently. TCP-AP also obtains better overall results than TCP, but with more mobile nodes in the network its results become similar to standard TCP results. This is due to the fact that the link capacity evaluation of TCP-AP becomes more inaccurate when the number of nodes increases. WCP has also lower performance when the number of nodes increases. As WCP signals all flows in a neighborhood of congestion and sets the control interval to the maximum round trip time (RTT), this strategy, based on the AIMD standard scheme, makes WCP to be very conservative not using efficiently the medium. With respect to the number of received packets, it is possible to observe that both WCP and TCP have better results than TCP-AP. This is due to the fact that the four hop propagation delay of TCP-AP is very conservative, making it receive less packets for the same throughput. With the previous results, it is also possible to conclude that TCP-AP is not fair, since throughput values are not obtained with good received packets values. WCP fairness decreases with the increase in mobility.

To understand how the variation of the number of mesh nodes affects network perfor- mance, Figure 3.6, Figure 3.7 and Figure 3.8 show the results for mesh scenarios with a fixed number of 7 mobile nodes and a variable number of mesh nodes (5, 9, 12 and 15 mesh nodes).

The obtained results show that WCP has the best results. WCP presents, however, a very irregular behavior as opposed to TCP when the number of mesh nodes increases. This is due to the fact that, with the increased number of mesh routing messages, WCP is not capable to correctly infer and monitor the round-trip time of every flow. With more mesh nodes and more flows in the network, WCP becomes inefficient and less accurate. With the increase of nodes and flows, the routing messages in the network also increase, and WCP does not evaluate correctly the interference range of the congested link, becoming more aggressive and obtaining poor performance.

TCP-AP, while presenting better overall throughput results than TCP, obtains lower average results for delay and received packets. TCP-AP is based on the estimation of the current four hop propagation delay and in the coefficient of variation of recently measured RTTs. The increase on the number of mesh routing messages in the network causes TCP- AP rate control mechanism to be excessively proactive making it send few packets, resulting

3.4 Performance Evaluation Results 100 200 300 400 500 600 700 800 3 3.5 4 4.5 5 5.5 6 6.5 7 Throughput(Kbps)

Number of Mobile Nodes TCP Avg. Throughput

TCP-AP Avg. Throughput WCP Avg. Throughput

Figure 3.3: AIMD Based Protocols Average Throughput - 16 Mesh Nodes, Variable Num- ber of Mobile Nodes.

10 100 1000 10000 3 3.5 4 4.5 5 5.5 6 6.5 7 Delay (ms)

Number of Mobile Nodes

TCP Avg. Delay TCP-AP Avg. Delay WCP Avg. Delay

Figure 3.4: AIMD Based Protocols Average Delay - 16 Mesh Nodes, Variable Number of Mobile Nodes.

in less received packets and irregular throughput values. Another important conclusion is that TCP-AP with multiple flows and higher number of mesh routing messages is not able to correctly infer available bandwidth assuming that the bandwidth at each node is identical.

TCP is using more fairly and more efficiently the medium when the number of mesh nodes, and consequently the number of mesh routing messages, increases. Its standard congestion control mechanisms, while very conservative in the slow start phase at the

0 2000 4000 6000 8000 10000 12000 14000 3 3.5 4 4.5 5 5.5 6 6.5 7 Number of Packets

Number of Mobile Nodes TCP Avg. Recv. Packets

TCP-AP Avg. Recv. Packets WCP Avg. Recv. Packets

Figure 3.5: AIMD Based Protocols Average Received Packets - 16 Mesh Nodes, Variable Number of Mobile Nodes.

100 120 140 160 180 200 220 240 260 4 6 8 10 12 14 16 Throughput(Kbps)

Number of Mesh Nodes TCP Avg. Throughput

TCP-AP Avg. Throughput WCP Avg. Throughput

Figure 3.6: AIMD Based Protocols Average Throughput - Variable Number of Mesh Nodes, 7 Mobile Nodes.

beginning, it improves then the network performance with the retransmit/fast recovery phase. Thus, TCP shows a very regular and stable behavior even when the number of mesh nodes increases.

The results of ad-hoc scenarios are shown in Figure 3.9, Figure 3.10 and Figure 3.11. WCP obtains the best average throughput values, while TCP and TCP-AP show very similar results. However, in terms of delay results, TCP gets the best results. Fairness can be outperformed with throughput and bandwidth allocation. Thus, combining Figure

3.4 Performance Evaluation Results 100 1000 10000 4 6 8 10 12 14 16 Delay(ms)

Number of Mesh Nodes

TCP Avg. Delay TCP-AP Avg. Delay WCP Avg. Throughput

Figure 3.7: AIMD Based Protocols Average Delay - Variable Number of Mesh Nodes, 7 Mobile Nodes. 1000 1500 2000 2500 3000 3500 4000 4500 4 6 8 10 12 14 16 Number of Packets

Number of Mesh Nodes TCP Avg. Recv. Packets

TCP-AP Avg. Recv. Packets WCP Avg. Recv. Packets

Figure 3.8: AIMD Based Protocols Average Received Packets - Variable Number of Mesh Nodes, 7 Mobile Nodes.

3.9 and Figure 3.10, it is possible to conclude that TCP has a more fair behavior than both WCP and TCP-AP, allowing to use more efficiently the medium when the network is heavily utilized. In terms of number of received packets, WCP gets the best values and TCP-AP the worse values. In terms of number of received packets, TCP results are not very far way from the ones obtained by WCP. WCP modified AIMD mechanism tries to explicitly recognize and account for congestion within a neighborhood; in high mobility and congested networks WCP is not very accurate introducing higher packet losses, thus,

0 200 400 600 800 1000 1200 1400 1600 1800 2000 2200 0 20 40 60 80 100 120 140 Throughput(Kbps)

Number of Simultaneous Flows TCP Avg. Throughput

TCP-AP Avg. Throughput WCP Avg. Throughput

Figure 3.9: AIMD Based Protocols Average Throughput - Ad-Hoc Scenario.

10 100 1000 10000 0 20 40 60 80 100 120 140 Delay(ms)

Number of Simultaneous Flows

TCP Avg. Delay TCP-AP Avg. Delay WCP Avg. Delay

Figure 3.10: AIMD Based Protocols Average Delay - Ad-Hoc Scenario.

obtaining very conservative received packets results. Regarding TCP-AP, its results show that it uses very conservative mechanisms, and that being an hybrid mechanism, when considering high density and high mobility networks, its estimation mechanism together with the AIMD strategy makes it behave very inefficiently and inaccurately, thus, obtaining moderate evaluation results, sometimes even worse than TCP. Surprisingly, TCP shows a very stable and fair behavior.

3.4 Performance Evaluation Results 2000 4000 6000 8000 10000 12000 14000 16000 18000 20000 22000 0 20 40 60 80 100 120 140 Number of Packets

Number of Simultaneous Flows TCP Avg. Recv. Packets

TCP-AP Avg. Recv. Packets WCP Avg. Recv. Packets

Figure 3.11: AIMD Based Protocols Average Received Packets - Ad-Hoc Scenario.