The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Voi XXXVII. Part Bl. Beijing 2008 
distance of node v* and Vj. At each simulated timestamp, the 
network model tries to construct the graph G by searching for a 
path that connects each communication pair. The existence of 
such paths indicate the possibility of routing data packets along 
the path. 
Supposing that the width of a road is far less than the range of 
radio frequency (RF), the I VC networks constructed in highway 
scenarios can be regarded as 1-dimensional (Franz et al., 2005), 
and the destination is located either in front of the sender or 
somewhere behind. To make it easier to understand, we propose 
the following assumptions: 
• Not considering the multipath effects, which is the term 
given to the phenomenon where a radio signal arrives at the 
receiving antenna after being reflected off a surface, so if 
two arbitrary nodes located within the transmission range of 
each other, we deem that only one wireless communication 
link exists. 
• Symmetrical link path, which mainly appears that the band 
width and the radio transmission range of all the nodes in 
the network are symmetrical. If node A can communicate 
with node B, then node B can also communicate with node 
A. 
• Considering border effects, which we can disable the nodes 
iff they are located out of the observation scenario at that 
moment, thus they cannot participate in the network or for 
ward the message. 
4. SIMULATION RESULTS 
To further study how the features of each protocol affect their 
performance of data disseminations, we did an extensive 
performance comparison using the implementations of these 
protocols in ns-2 with CMU wireless extensions under Linux. 
The common parameters we used in the simulations are listed in 
Table 1. 
We obtained the original mobility scenario from the FleetNet 
(FiiBler et al., 2006) project, and the dataset comprises each 
vehicle’s position (X), lane, speed, and acceleration values at 
each timestamp. We preprocessed the data in accordance with 
the highway scenarios in China. Due to the lane information 
only in the dataset, the action of lane-changing is usually done 
between the two adjacent simulation timesteps. We mapped 
each vehicle’s lane into its position (Y) based on the system of 
coordinates demonstrated in Figure 1, that is: 
PositionY = (Lane+ 0.5) x Lane Width ± 0.5 x MedianWidth, Lane = ±1,±2.0) 
Parameter 
Simulated Value 
Channel type 
Wireless channel 
Antenna model 
Omnidirectional antenna 
Radio propagation model 
Two ray ground 
Network interface type 
Lucent WaveLAN (915MHz) 
Nominal radio range 
250m 
MAC type 
IEEE 802.11 DCF 
Interface queue type 
Priority queue (50 packets max.) 
Number of nodes 
108/240/374/500 
Simulation time 
60s 
X dimension 
12km 
Table 1. Simulation parameters 
The traffic sources are constant bit rate (CBR), generated with 
the help of cbrgen.tcl script, and only 512-byte data packets are 
used. The randomly chosen source-destination pairs are spread 
in the network. The percentage of the communication pairs is 
varied from 10% to 30% for evaluating the load of the network. 
Three important performance metrics are evaluated: 
• Packet delivery fraction - the ratio of the data packets 
received to the data packets sent; 
• Normalized routing load - the ratio of the routing packets 
delivered to the data packets received; 
• Average end-to-end delay - this includes all possible delays 
caused by buffering during route discovery latency, queuing 
at the interface queue, retransmission delays at the MAC, 
and propagation time. 
Figure 2 shows the packet delivery fraction of different ad hoc 
routing protocols vs. number of nodes in the network; thereinto, 
Figure 2(a) is for 10 percent of the communication pairs, 2(b) is 
for 20 percent of the communication pairs, and 2(c) is for 30 
percent of the communication pairs. We understand that AODV 
and DSR have similar performance, the former outperforms the 
latter for larger network size measured in the number of nodes, 
whereas the latter outperforms the former for smaller network 
size. The DSDV is totally unsuitable for our mobility scenarios, 
especially in the case of higher percent of communication pairs 
in networks, where no more than 35% of data packets delivered 
successfully. 
<x 
500 
<x 
¿r 
Routing Protocol: 
♦ AODV 
■ DSDV 
♦ DSR 
100 
200 
300 
400 
500 
œ 
2r 
Routing Protocol: 
♦ AODV 
■ DSDV 
♦ DSR ♦ 
100 
200 
300 
400 
500 
Number of Nodes in the network 
Number of Nodes in the network 
Number of Nodes in the network 
2(a) 10% communication pairs 
2(b) 20% communication pairs 
2(c) 30% communication pairs 
Figure 2. Packet delivery ratio vs. number of nodes