Page 2 Lecture Notes in Computer Science 2865 Edited by G. Goos ...

30 P. Narayan and V.R. Syrotiukrandom waypoint model [3] with a maximum velocity v, which varies from 0 m/sto 10 m/s. The mobility model is also characterized by a pause time.Our experiments use constant bit rate (CBR) data sources over UDP for datacommunication. The primary goal of our current experiments is to illustrate thebehaviour of routing protocols for simple cases. Therefore, we choose UDP asour transport layer protocol in order to eliminate the influence of congestion andflow control mechanisms on the performance of the routing protocol.3.2 Traffic and Mobility PatternsFor a typical ns-2 wireless simulation, a connection pattern file designates theconfiguration and behaviour of data connections in the network scenario to besimulated. It specifies the end points between which the data flow takes place,when the data flow over the connection should start and stop, and the typeof application data sent by the source. In our experiments, each connectionpattern file has one s-t pair and a CBR connection between them to eliminateany possibility of collisions or buffer overflow due to high network load.For each s-t pair, the data flow starts and ends at a random time. A script togenerate connection pattern files is provided with the CMU extensions to ns-2[3]. We modify the CMU script to impose the restriction that each connectionlast for at least 100 seconds.A mobility pattern file defines the motion of the nodes in the network andthe changes in the paths between the nodes over time. A program to generatethe mobility pattern is also provided [3]. By default, this program assumes afixed transmission range of 250 m to denote one hop. The initial positions of thenodes are designated at random and the move sequences are generated accordingto the random waypoint mobility model.Using an approach similar to Holland and Vaidya [9,10], we generate mobilitypatterns for the network of 20 nodes moving with a mean speed of 2 m/s for 1800seconds and a pause time of 0 s. We call these mobility pattern files base patterns.We generate 5 such base pattern files. These base pattern scripts are used togenerate mobility patterns for mean speeds of 4, 6, 8 and 10 m/s, respectively.3.3 Extraction of the Mobile GraphIn order to generate the mobile graph for a given scenario, we sample the currentposition of every mobile node at regular intervals and output the link transitionsthat occurred within each interval. The sampling interval is such that we obtain100 samples before the node travels through its transmission radius once.The initial connectivity matrix translates into the first graph sequence in themobile graph. By applying link additions and breakages recorded at differenttime intervals, we obtain the graphs in the graph sequence over the time scaleequal to the total simulation time.