Due to the
difficulty of testing our system in a real environment, we have
developed a simulator.
Among
other features, our simulator allows us to represent realistic
environments, simulate different types of events, define different
routes for vehicles (e.g.,
roads with different curve profiles) or extract real roads from
a roadmap,
set up speeds and traffic conditions, etc. To evaluate our approach, we
collect information regarding when the vehicles report the events to
the drivers, and when the vehicles meet the events.
Our simulator focuses on the evaluation of V2V systems. Our goal is
here to monitor the efficiency of the inter-vehicles communication
systems in terms of quality of data provided to the driver (e.g.,
relevance of events communicated to the drivers, time between the
moment when the warning about an event is received by a vehicle and the
moment when the vehicle really encounters this event, etc). Obviously,
the network aspects are very important. However, even if our simulator
observes parameters like the network load, it is not dedicated to
network monitoring contrary to many existing simulators (e.g., NS2,
GloMoSim, JiST-SWANS, etc.).
Besides, our simulator supports the potential modifications of the
itineraries of vehicles warned and the impact of such changes on the
routing/dissemination protocols used.
The
graphical
interface of the simulator has been developed to observe the behaviour
of the vehicles during the simulations. The graphical interface is
illustrated below for both the standard road and the parking lot
configurations. Obviously, the graphical interface can be
desactivated to improve the performances.
To
evaluate a prototype, it is plugged in the simulator. Then, the
simulator and the prototype interact through the wireless
communications manager. Our
simulations are usually run on Manfred, our Mac Pro
equipped with 2 xeon quad core 2,8 GHz and 32 Go RAM.
Standard
roads configuration
Different
colors
are used for their vehicles according to their status. The red
vehicle is responsible of the event generation. The green vehicles have
received a message describing the event and have estimated it relevant.
Finally, vehicles reprensented with the red dark color have received
the event but did not consider it relevant for their driver.
Our
simulator
enables the evaluation of different scenarios. Below, a first video
shows a simulation with a stationary direction dependent event (e.g.,
an accident) running on the highway between Valenciennes and Lille
(located in the north of France) whereas the second one runs in the
centre of Lille with a stationary non-direction dependent event.
Simulation on the highway: a vehicle disseminates a
message about an accident.
Simulation in centre of Lille
Parking
lot configuration
Since
our simulator enables the creation of roads, it is possible to define
parking lot configurations (roads and entries of parking lots are
usually not represented in existing roadmaps). This configuration
enables to study the exchange of "available parking space" events
(i.e., stationary non direction-dependent events).
Simulation of a small
"intelligent" parking lot: vehicles leaving their parking
space advertise it to the other drivers.
Here, for
the parking configuration, blue cars correspond to cars leaving a
parking space. Cars already parked are represented in black and the
colour of vehicles searching for a parking space ranges from orange to
red according to the duration of their search. The
direction of moving vehicles is also shown using a dashed line.
For
such a
configuration, the parameters observed during the simulations are the
average time needed by vehicles to find a parking space, the standard
deviation, the network load, etc.
The
parking lot
configuration enables the comparison of different solutions to allocate
the parking spaces (i.e., without any V2V communication features, using
VESPA's encounter probability, using the relevance function presented
in [1], etc.)
References
[1] B. Xu, A. M. Ouksel, and O. Wolfson.
Opportunistic resource exchange in
inter-vehicle ad-hoc networks. In 5th Int. Conf. on Mobile Data
Management, 2004.
