Performance Evaluation of Vehicle-to-Vehicle Communications for a Collective Perception Application in Vehicular Ad hoc Networks
1. Patrick SONDI a, Lucas RIVOIRARD b, Martine WAHL b
a Univ. Littoral Côte d‘Opale, LISIC - EA 4491, F-62228 Calais, France
b Univ Lille Nord de France, F-59000 Lille, IFSTTAR, COSYS, LEOST, F-59650 Villeneuve d’Ascq
Vehicle to vehicule - V2V DECENTRALIZED
http://www.digitaltrends.com/
http:// motorvehicleregs.com/
• Lowcostdeployment, integrated inthevehicle
• Cooperative ad hoc networks
• Distributed coordination of network
functions
• Expensive deployment for road managers
• What happens if a terminal fails?
MULTIPOINT
RELAY
Features:
• Routing protocol for VANETs
• One-hop Clusters
• Proactive distributed
algorithm
Leaf node : an
ordinary node that
links to the nearest
branch node.
Branch node : a relay
node responsible of a
group of leaf nodes.
Elected by other nodes
to forward application
messages and build a
chain.
A chain: a virtual
backbone consisting of
a set of relay nodes.
ROAD
NETWORK
CLUSTERING
CBL Components
Objective: to offer, via V2V communication, a routing service enabling
communications between close and distant vehicles.
Objective: performance of the CBL scheme with perception application
Road Context:
• 10-km A27 highway section (2x2 lanes)
• 4-km D90 by-way (2x1 lane)
patrick.sondi@univ-littoral.fr
lucas.rivoirard@ifsttar.fr
martine.wahl@ifsttar.fr
Future work
• Evaluation of CBL on other ITS applciations in VANETs
• A new CBL version which increases the connection time
between the leaf nodes and their elected branch node
Conclusion - performance
The results show how to fix the frequency of the exchanges in the
collective perception application in order to achieve both predefined
packet delivery ratio and end-to-end delay at WLAN and IP levels.
Vehicle to infrastructure - V2 CENTRALIZED
CBL (Chain Branch Leaf)
Technology Context:
• IEEE 802.11p
• OPNET Riverbed Modeler
Vehicular communication technologies
A27
RD90
Performance Evaluation of Vehicle-to-Vehicle
Communications for a Collective Perception Application in
Vehicular Ad hoc Networks
#1570473915 29th Annual International Symposium on Personal, Indoor and Mobile Radio Communications (PIMRC)
• Traffic density
recorded on the
A27 on April 6th,
2017, 12:35 p.m.
Comparative evaluation
CBL results
• With its own sensors (radar,
lidar, camera) the perception of
the autonomous car is limited to
200 meters
• Exchange information allows to
build a Local Dynamic Map with
1km-perception
Collective percetion application
Description
for static and
mobile object
Ssize
(octect)
Static
Mobile
Position 4 ✓ ✓
Time 8 ✓ ✓
Heading 2 ✗ ✓
Speed 2 ✗ ✓
Acceleration 2 ✗ ✓
Size 6 ✓ ✓
Packet delivery ratio
End-to-end WLAN delay
End-to-end IP delay
• PDR increases when the
waiting interval between two
packet increases.
• PDR is still above 88%.
• PDR is 91% when the waiting
interval is greater than 350 ms.
• The end-to-end delay
observed at the WLAN level
suggests that real-time
transmission of one PEM
every 50 ms may be
achieved with a very low
delay of 0.5 ms.
• IP delays are always less
than 600 ms and they
increase when the time
interval is decreasing.
• Sending EPM message
every 175 ms insure that
these message are received
before sending a new packet
Hendrik-jorn Gunther, Oliver Trauer, and Lars Wolf. The potential of collective perception in vehicular ad-hoc networks,
ITS Telecommunications (ITST), 2015*
Source*