1017 
The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part Bl. Beijing 2008 
which data dissemination protocols are suitable for vehicular ad 
hoc networks. 
To the best of our knowledge, (FiiBler et al., 2002) analyzed the 
quantitative behavior of ad hoc routing algorithms for data 
dissemination between vehicles, and evaluated the performance 
of a reactive ad hoc routing protocol (DSR (Johnson and Maltz, 
1996)) and of a position-based approach (GPSR/RLS (Karp and 
Kung, 2000)). Simulation results suggested that position-based 
ad hoc routing protocol has significant advantages over reactive 
non-position-based approach. In contrast to highway scenarios, 
(Lochert et al., 2003) first evaluated ad hoc routing protocols 
over a realistic vehicle mobility pattern for a city scenario, and 
presented a simulation study that compares a position-based 
routing (GSR) approach with classical ad hoc routing methods 
(AODV(Perkins and Royer, 1999) and DSR(Johnson and Maltz, 
1996)). Simulation results also demonstrated that position-based 
routing outperforms topology-based approaches with respect to 
delivery rate and latency. 
(Jaap et al., 2005) also evaluated the performance of routing 
protocols (AODV(Perkins and Royer, 1999), DSR(Johnson and 
Maltz, 1996), FSR, and TORA(Park and Corson, 1997)) in city 
traffic scenarios, and found out that TORA is completely 
unsuitable for vehicular environment, whereas AODV and FSR 
showed promising results, DSR suffered from very high end-to- 
end delay. 
(Naumov et al., 2006) studied the behavior of routing protocols 
(AODV(Perkins and Royer, 1999) and GPSR(Karp and Kung, 
2000)) in an inner city environment and on a highway segment 
by using realistic mobility traces obtained from a microscopic 
vehicular traffic simulation on real road maps of Switzerland. 
Both exhibit serious performance problems in the investigated 
VANET scenarios. 
3. SIMULATION MODELS 
3.1 Highway mobility model 
Vehicular mobility models can be classified as macroscopic and 
microscopic^Haerri et al., 2006).When following a macroscopic 
approach, motion constraints (e.g., roads, crossings and traffic 
lights) are considered and generation of vehicular traffic (e.g., 
traffic density, traffic flows, and vehicle distributions) are also 
defined. In contrast, with a microscopic approach the movement 
of each individual vehicle and the vehicle behavior with respect 
to others are determined. It is obvious that the combination of 
micro-macro approach is more suitable for vehicular mobility 
model. We developed a mobility scenario for highway traffic in 
China, which is modeled as 2 X 2 scenario, viz. bidirectional, 
two lanes each scenario. Vehicles can move along roadways 
with high speed towards the two opposite directions, which are 
separated by the median zone. The two lanes in each direction 
can be further divided into normal-speed (right-hand side) and 
overtaking (left-hand side). We depict our highway scenario in 
Figure 1, without loss of generality. The width of each lane and 
median is taken as 3.75m and 2.00m, respectively. 
In our highway scenarios, we assume that all vehicles follow 
the directional mobility model, in which each vehicle randomly 
selects a waypoint ahead in the same direction and then moves 
from its current position to the selected waypoint. The running 
speed is determined by the intelligent driver model (IDM) 
(Treiber et al., 2000), which belongs to the class of car 
following model. As shown in figure 1, the instantaneous accel 
eration of vehicle i is denoted as follows: 
Axi.,(t) Ax ft) 
Figure 1. A segment of highway scenarios 
V' 1DM (v i ,s i ,Av i ) = a ] 
- „(0 
max 
1- 
( 
, ( o 
V y des 
/(v,.,Av.) 
(1) 
where, v is the current instantaneous speed 
v des is the desired velocity 
Av is the velocity difference (approaching velocity) to 
the preceding vehicle 
a max is the maximum acceleration 
S is the acceleration exponent 
s and s are the desired distance and actual distance gap 
between adjacent vehicles on the same lane, respectively; 
subscript i or superscript (z) represents corresponding 
parameters of the vehicle 
The first two items interpret the vehicle’s running acceleration 
on freeways, and the third describes the desired deceleration in 
case of vehicle i approaches to the vehicle in front. The desired 
distance gap of vehicle i is denoted as follows: 
r (v, Av) = sj 0 + s[ n —- + T (i) v + 
v-Av 
(0 
v des 
(2) 
where, both s 0 and sj are the jam distance 
T is the safe time headway 
a is the maximum acceleration 
b is the desired deceleration 
3.2 Network evaluation model 
The network evaluation model assumes that all the vehicles are 
equipped with wireless transceivers, by which can dynamically 
construct ad hoc networks on the fly. The nodes in the network 
act as either a host (sender/receiver) or a router to perform data 
disseminations. We model the network as a communication 
graph G={V(G), E(G)}, where the vertex set V(G)={v l5 v 2 , ..., 
v n } represents all the participating vehicles in the network, and 
the edge set E(G)={(Vj, Vj) | distance^, vj) < R and i^j} 
represents all the communication links between corresponding 
nodes if and only if both are within the transmission range R of 
each other; thereinto, distance^, Vj) is defined as the Euclidean