Gugrajah_Yuvaan_ Ramesh_2003.pdf
Gugrajah_Yuvaan_ Ramesh_2003.pdf
Gugrajah_Yuvaan_ Ramesh_2003.pdf
You also want an ePaper? Increase the reach of your titles
YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.
Simulation ofa Load Balancing Routing Protocol<br />
3.6. Traffic Model<br />
Chapter 3<br />
Two packet buffers are maintained at each node, each being buffer_length packets<br />
long. One buffer contains data packets and the other routing information packets<br />
(route requests, route replies and route error packets). Routing information packets<br />
have the higher priority. Packets are dropped if they are in a queue for longer than<br />
pkCtimeout seconds.<br />
[Broch98], [Johannson99] and [DasOl] used constant bit rate (CBR) sources and<br />
varied the number of sources for different simulation runs in order to determine the<br />
effects of traffic load. The source nodes selected to be CBR sources were the only<br />
nodes throughout the simulation that generated traffic and the associated destination<br />
nodes for each source node also never changed. A different approach was<br />
implemented for traffic generation for the simulations discussed herein in an attempt<br />
to create stochastic behaviour.<br />
Each node is assigned a random Packet Generation Time (POT), normally distributed<br />
between 0 and a maximum packet generation time PGT. After POT seconds, the node<br />
selects a destination node randomly and generates session_length packets each of<br />
pkt_length bytes to be transmitted to the selected node. The packets are added to the<br />
appropriate buffer and the node assigns itself a new POT.<br />
3.7. Simulation Procedure<br />
The custom built simulator was used to compare AODV to the four DLAR schemes<br />
in terms of packet delivery ratio and delay. A total of 24 simulations were run for<br />
each of the four routing protocols for 3600s of simulation time. By varying the pause<br />
time from a low pause time to a high pause time, the network topology was varied<br />
from highly dynamic topologies with constant node movement to slow changing<br />
topologies with longer pause times. There were four different pause times used: 0,<br />
25s, 50s and 150s. For each pause time the load was varied for six different values of<br />
PGT: 0.25s, 0.5s, Is, 2s, 5s and 10s.<br />
3-15