2011 14th International IEEE Conference on

Intelligent Transportation Systems

Washington, DC, USA. October 5-7, 2011

A Survey of Vehicular-based
Cooperative Traffic Information Systems
Marcin Seredynski and Pascal Bouvry

Abstract— This paper summarises the research on cooper- cameras, infrared sensors, ramp meters and traffic signals.
ative traffic information systems based on vehicular ad hoc The information is processed by a traffic information agency
networks. Such systems are a promising concept for exchanging and then typically provided to the navigation systems of
traffic information among vehicles. Their application enables
reductions in fuel consumption, greenhouse gas emissions and vehicles using the Traffic Message Channel (TMC) technol-
travel time. Three classes of such systems are studied: in- ogy. The TMC messages are embedded into conventional
frastructureless solutions based on inter-vehicle communica- FM radio broadcasts using the Radio Data System (RDS)
tion, infrastructure-based solutions relying on the peer-to-peer communications protocol. However, there are four main
paradigm and infrastructure-based systems using client-server drawbacks to such an approach. Firstly, it is completely
architectures. Systems within each class are briefly introduced
and their strengths and weaknesses are analysed. centralised. Secondly, as it is based on a fixed and costly
Index Terms— Cooperative traffic information systems; ve- infrastructure, it is limited to the main roads (typically
hicular ad hoc networks; intelligent transportation systems. motorways). Thirdly, it lacks dynamicity, that is, information
updates are far from real-time as delay is typically in the
I. I NTRODUCTION range of 20–50 minutes [3]. Lastly, the RDS technology
The application of wireless technology to moving vehicles offers very restricted bandwidth, therefore, traffic informa-
enables the creation of vehicular ad hoc networks (VANETs). tion has to be limited in details. These drawbacks can be
Four main transmission types are present in VANETs [1]: in- overcome by cooperative traffic information systems (CTIS),
vehicle – communication between electronic units within a where traffic-related information is collected individually
vehicle, vehicle-to-vehicle (V2V) – communication among by vehicles and exchanged between themselves using wire-
nearby vehicles, vehicle-to-infrastructure (V2I) – commu- less networks. CTIS provide vehicles with real-time traf-
nication between vehicles and roadway infrastructure, and fic information allowing dynamic route guidance. Location
vehicle-to-backoffice (V2B) – communication between ve- awareness is provided to system users by Global Positioning
hicles and a central entity via standards, such as GSM or System (GPS) receivers. In general, CTIS can be classified
UMTS. Three basic groups of applications are envisioned as either infrastructureless or infrastructure-based [4]. The
in VANETs [1]: (i) safety-related, (ii) infotainment and ad- former are based on V2V communication. The latter use
vanced driver assistance services and (iii) resource efficiency infrastructure-based communications technologies and might
(traffic, environment). This article deals with the last group. rely on centralised traffic information agencies.
Efficient traffic management is becoming of great interest The purpose of this paper is to review and describe basic
today as traffic congestion becomes a more and more severe principles of the CTIS proposed in the literature. To the best
problem. Vehicles that are idling, or travelling at reduced of our knowledge, this is the first attempt to survey past
speeds due to congestion not only waste time of their work done in this field. The paper is structured as follows.
users but also emit more greenhouse gas and utilise more First, the main challenges of CTIS are presented. Next,
fuel. Therefore, congested flow conditions have a negative in Section III, systems belonging to the infrastructureless
impact on the economy, health, and environment. Moreover, class are presented. Their alternative, systems relying on
information about traffic conditions along the route tops the infrastructure, is described in Sect. IV. The final section
list of consumer interests in transportation information [2]. summarises and concludes the paper.
In addition, traffic efficiency applications do not require such II. C OOPERATIVE TRAFFIC INFORMATION SYSTEMS
a high market penetration level of VANETs as is the case for
safety-related applications. The infrastructureless CTIS typically apply data aggrega-
The improvement of traffic flow and congestion reduction tion techniques to limit bandwidth use and maintain scalabil-
can be achieved by means of traffic information systems ity. Usually, with increasing distance, observations regarding
(TIS). In general, their aim is to capture, evaluate and a given area become less precise. Thanks to store-and-
disseminate traffic-related information. The solutions that are forward techniques traffic information can be disseminated
currently used rely on conventional equipment like traffic in multiple partitions of VANETs. For more information
regarding communication patterns we suggest [5], [6]. As
M. Seredynski is with Interdisciplinary Centre for Security, Reliability the information is a subject of interest to many vehicles in a
and Trust, University of Luxembourg, 6, rue Coudenhove Kalergi, L-1359, given geographical area, the broadcast nature of V2V com-
Luxembourg, Luxembourg [email protected]
P. Bouvry is with Faculty of Sciences, Technology and Communication, munication fits very well the objectives of infrastructureless
University of Luxembourg [email protected] CTIS. However, such systems have two main drawbacks in

978-1-4577-2197-7/11/$26.00 ©2011 IEEE 163

long distance information dissemination. Firstly, they have a the information: more recent reports are assumed to be more
relatively high delay and secondly the information is limited accurate, and thus replace older ones. There are two types of
in its details (due to the distance-based data aggregation) [7], reports generated by vehicles: periodic and emergency. The
[8]. Another problem is that several overlapping aggregates aim of periodic reports is to inform the surrounding vehicles
for the same area may exist, making it difficult to compare about a vehicle’s position, speed, heading and to provide
them [4]. Therefore, the quality of V2V communication- its individual traffic analysis. These reports are generated
based approaches greatly depends on the quality of the periodically at time intervals based on the priority of the
aggregation techniques. For more information regarding ag- information. Upon reception of reports a vehicle updates its
gregation techniques for CTIS we suggest [9]. KB. The emergency reports are only used to send information
The drawback of infrastructure-based CTIS is that ser- about critical traffic incidents. Such reports have immediate
vice charges will most probably apply. However, they can access to the wireless channel. Each vehicle performs traffic
be deployed in the near future as in contrast to VANET analysis individually using its KB. An example of the analy-
technology, low-cost wireless Internet access will soon be sis provided in the paper is the average speed of all vehicles
very common [10]. Moreover, such systems require much in a given segment. Based on this information an estimation
lower market penetration compared to infrastructureless ap- of road condition in a given segment is made. The reports are
proaches. Infrastructure-based CTIS can rely on client-server exchanged between vehicles travelling in both directions. The
or peer-to-peer (P2P) models of data storage and communi- system was evaluated using the ns-2 network simulator [13]
cation [11]. In the client-server approach, traffic information and a simple vehicular traffic simulator based on cellular
is stored in a server which is typically accessed by vehicles automata.
over the Internet. The question arises of how to deal with In [3] the authors provide more information on the receive-
the huge number of updates and queries made to the system, analyse-send approach used in SOTIS. An adaptive broadcast
therefore scalability is an issue [4]. In addition, in such technique is introduced: broadcast intervals are adapted ac-
systems a single authority decides who may access data [8]. cording to provocation and mollification events. The former
However, compared to other solution types, client-server reduce the intervals, while the latter do the opposite. The
systems can provide much higher protection in terms of paper provides an extensive set of both types of events. In
information security for its users [12]. Moreover, route addition, data aggregation is introduced, as vehicles calculate
computation can be performed by the server. Finally, the the average speed per segment.
network layer problems that occur in V2V-based solutions
can be avoided [10]. In order to obtain CTIS independent B. TrafficView system
of central authorities and at the same time benefit from A system called TrafficView is presented in [14]. Its main
reliable infrastructure-based connectivity, a P2P architecture objective is to gather and disseminate information about the
can be used in place of the client-server model. In such a position and speed of vehicles. The information is restricted
case instead of using dedicated servers, data is distributed only to the vehicles positioned ahead of the current vehicle.
across the vehicles. The amount of traffic information is The approach for message exchange is very similar to that
positively correlated with the number of participating ve- used in SOTIS: vehicles periodically broadcast reports (con-
hicles, therefore P2P-based systems are scalable. However, tained in a single packet) about themselves and other vehicles
when compared with client-server approaches, the P2P-based they know about. Whenever a vehicle receives a report, it
solutions use a greater overall bandwidth [10]. The key updates its stored information, and sends the updated report
question in such systems is how to provide an information in the next broadcast period. Although the average size of
lookup service. A detailed comparison of P2P and V2V- the stored records is very small (on the order of 50 bytes)
based CTIS can be found in [4]. data aggregation is performed in order to fit all information
in a single packet. The authors propose aggregating data
III. I NFRASTRUCTURELESS APPROACHES
based on its semantics. For instance, the information about
A. SOTIS system two vehicles that are close to each other and are moving
A Self-Organizing Traffic Information System (SOTIS) with similar speeds will be stored in a single record. In
is proposed in [7]. It works within an approximate radius consequence, the record represents information about a set
of 50 to 100 km of an individual user, even if as few of vehicles. The selection of data for aggregation is based
as only 2% of all vehicles are using it. The precision of only on the relative distances of the vehicles concerned.
the information it provides decreases as the distance to the According to the authors, data compression techniques could
area of interest increases. The authors propose a distributed further minimise the size of the records. However, due to
receive-analyse-send algorithm. The information received the limited computation resources of portable devices, such
from other vehicles is first analysed, and only results of the techniques are not applied in this proposal. Collected data is
analysis are transmitted. Roads are divided into variable size processed into validated data sets using a validation module.
segments, and the information is exchanged on per segment The validation procedure resolves conflicts between the data
basis. Each vehicle stores information for every segment of records (e.g. multiple records containing information about
the road in its local database called Knowledge Base (KB). A the same vehicle), removes records containing information
GPS-based time-stamp is used to determine the accuracy of about vehicles behind its own vehicle, and updates esti-

164
mated position of vehicles using the stored speed. Three exchange is limited only to the maps with unusual speeds. In
different algorithms for selection of data for aggregation are consequence, each vehicle is able to build a map of expected
introduced: ratio-based, cost-based and information aging. speeds on roads that other vehicles have travelled on. Data
The aggregation procedure on the records in the validated aggregation is performed using clustering techniques, which
dataset is performed before each broadcast period. Perfor- combine related recordings of an unusual speed. Cluster
mance evaluation of TrafficView was carried out using the analysis is based on a modified k-means method. The authors
ns-2 network simulator and the CORSIM vehicular traffic suggest that unusual speed should not only refer to the posted
simulator [15]. The authors developed their own vehicular speed limit but should also consider predictable trends.
traffic scenario generator. In [16] the authors show that For instance, significant congestion on commuter routes at
the most efficient data dissemination in TrafficView can be certain hours can be treated as an expected event, which does
achieved if messages are relayed only by vehicles travelling not require reporting. The system was evaluated using the
in the opposite direction. authors own simulator of Manhattan’s grid of highways and a
random way point mobility model. The authors demonstrate,
C. TrafficRep system that extending the system with wireless base stations could
In [17] the authors examine the efficiency of a V2V- significantly improve the performance of the system.
based traffic information dissemination architecture. A small
fraction of the vehicles is assumed to use a device called E. IFTIS system
TrafficRep whose goal is to collect and disseminate traf- In [20] the authors claim that the previous systems pro-
fic information. The device is connected to a digital map posed in the literature (SOTIS and TrafficView) are not
database, a GPS receiver and a wireless communications well adapted to city environments. Therefore, in [20] they
device. Roads are divided into segments. A priori travel time introduce a new system called Infrastructure-Free Traffic
on a segment is set to the free-flow travel time. Each time a Information System (IFTIS) to tackle the urban environment.
vehicle reaches the end of a segment, its TrafficRep device Its main objective is to provide vehicles with an estimation
produces a travel log report (TLR). The report includes of traffic density in city roads on a segment-to-segment
travel time corresponding to the segment, the identifier of the basis. Based on the motivation that vehicles in geographical
segment, and a time-stamp. Using V2V communication, ve- proximity often share similar traffic information, the authors
hicles periodically exchange TLRs. The authors evaluate two introduce a location-based group concept, in which only a
dissemination approaches, naive scheme and smart scheme. group leader is responsible for information broadcast. The
In the naive scheme, vehicles broadcast TLRs correspond- main goal of the leader-based approach is to avoid scalability
ing to recently travelled segments at a pre-defined rate of issues. A segment of a street between two intersections is
dissemination. Only TLRs for which the difference between divided into fixed size cells. Each cell has its centre. The
the experienced and the expected travel time was greatest knowledge about cell’s location is a priori known by the
are transmitted. On the other hand, in the smart scheme, vehicles. This enables them to be arranged into dynamic
TLRs are sent when vehicles have significant information location-based groups defined by the position of the cells.
to share. As in the naive scheme, reports are also broadcast The vehicle that is closest to cell centre is considered to
at a pre-defined rate of dissemination. These broadcasts are be the group leader. Vehicles record information about the
however limited to TLRs for which the difference between speed and direction of neighbouring vehicles. This allows
the experienced and the expected travel time was greater than them to estimate the traffic density (defined as the number
a certain threshold. The proposed approach was evaluated of vehicles) of all cells in a segment in order to create a Cells
using a calibrated traffic simulation model of the Southern Density Packet (CDP). As soon as a group leader reaches the
New Jersey transportation network and the Paramics network intersection, it sends its CDP backwards to the beginning of
simulator [18]. It is shown that market penetration between the street section. Before reaching the beginning of the sec-
3 and 10% is enough to obtain the full benefits of V2V tion, the CDP is updated by other group leaders that relay the
communications technology for CTIS. Finally, the authors packet. When its final destination is reached, it is propagated
claim that if simple dissemination techniques like the smart to vehicles around the intersection. The IFTIS system was
scheme are used, then bandwidth is not an issue for VANET- evaluated using the Qualnet network simulator [21]. Mobility
based CTIS (contrary to the claim made in [14]). patterns were generated using VanetMobiSim [22].

D. StreetSmart Traffic system F. Trafficinfo system

The StreetSmart Traffic system introduced in [19] proposes The Trafficinfo system introduced in [23] deals with band-
to aggregate traffic information using distributed clustering width issues by using rank-based store-and-forward tech-
algorithms with an epidemic diffusion model. It is designed niques. Similarly to the approach presented in [17], roads
to perform well even if only a small fraction of vehicles are divided into segments with unique identifiers. As soon
participates in the system. Moreover, the system does not as a vehicle reaches the end of a road segment it generates a
require constant connectivity. Each vehicle records its speed travel time report for that segment (similar to that proposed
and on this basis builds a local traffic map. Vehicles that are in the TrafficRep system). Each vehicle broadcasts to its
close to each other exchange their speed maps. However, the neighbours only the most relevant (highest rank) reports

165
from its database. The rank of the reports is based on their has not received a similar report within a certain period. The
supply and demand characteristics individually evaluated by CoCar system was evaluated using the OMNeT++ network
each vehicle. Supply indicates the number of vehicles that traffic simulator [26] coupled with the SUMO road traffic
already have the report. Each vehicle can estimate the supply simulator. A large motorway interchange in the Frankfurt
of report R by analysing the number of times that R has area was used as a test case. The authors also analyse the
been received from other vehicles (using a machine learning- environmental impact of the CoCar system using the model
based algorithm). On the other hand, demand of R depends proposed in [27].
on two spatial-temporal factors of the vehicle performing
the evaluation: its distance to the road segment reported by C. TraffCon system
R and age of the report. Broadcasts are triggered by the The aim of the TraffCon system presented in [28] is to
following events: (i) the vehicle receives a new travel time provide vehicles with driving directions (instead of keeping
report from another system user, (ii) the vehicle reaches drivers better informed about traffic conditions as is the case
the end of a road segment and produces a new travel time for other systems). Its goal is to globally optimise the usage
report and (iii) the time since the last broadcast made by of the road network and avoid flash crowd driving patterns.
the vehicle exceeds a broadcast time threshold (BTT). The A vehicle submits (using a cellular network) the information
authors propose two ways of setting the value of BTT: static about its current position and desired destination to a TIC.
and dynamic. The former is a fixed threshold, while the latter Route computation is performed by the TIC, which calculates
is based on vehicle’s speed and transmission range. Dynamic the quality of k shortest routes. The quality is defined by
BTT guarantees that two vehicles travelling on the same road a fitness function. The vehicle receives the route with the
segment but in the opposite directions will initiate broadcasts best fitness value. The procedure is repeated periodically.
at least once. The system was evaluated using the STRAW Therefore, if traffic conditions change, or the vehicle does
system [24], which integrates vehicular traffic and network not obey the instructions, a new route is produced by the
simulators. A scenario corresponding to downtown Chicago TIC. The quality of a route takes into account four objectives:
was used. travel time, effect on congestion, fuel consumption and gas
emissions. As there is no other exchange of traffic infor-
IV. I NFRASTRUCTURE - BASED APPROACHES mation in the system, its performance strongly depends on
A. SOCRATES system the penetration rate of participating vehicles. TraffCon was
One of the first client-server CTIS was SOCRATES (Sys- evaluated using the SWANS ad hoc network simulator [29].
tem Of Cellular RAdio for Traffic Efficiency and Safety) A small part of Boston (MA, USA) road network was used
introduced in [25]. However, only a generic architecture as the test scenario.
for the system can be found in this work. It assumes two-
D. PeerTIS system
way cellular-based communication between vehicles and a
traffic information centre (TIC) using 1/2G cellular tech- The idea of establishing a P2P overlay over the Internet
nologies (Mobitex and GSM). The vehicles periodically send for CTIS was first presented in [30]. Its aim is to reduce
anonymous traffic reports (containing their position, status the information propagation delays present with store-and-
and travel time) to the TIC. The reports are stored and forward techniques and at the same time to avoid typi-
processed by the centre. On the basis on these reports the cal drawbacks of a client-server approach. A full-fledged
TIC calculates current conditions and also predicts traffic system called PeerTIS is introduced in [8]. Direct V2V
flows. The information is sent back to the vehicles. The communication between vehicles is replaced with a cellular
authors claim that only between 1 and 2% of vehicles need network: using the IP-based communication channel vehicles
to be equipped with the system in order to allow the TIC create a P2P overlay over the Internet. Traffic information
to calculate accurate traffic flows. However, no evaluation of storage, lookup and exchange are based on the structured
the system is presented. P2P paradigm. Subsequently, the observed traffic conditions
are deterministically published to a particular node, based
B. CoCar system on the lookup mechanism. Roads are divided into segments
A client-server system called CoCar is proposed in [12]. with unique identifiers. Vehicles generate travel time reports
The system is used to investigate the application of UMTS similar to that proposed in the TrafficRep system. The reports
technology to CTIS. As in the SOCRATES system, the are distributed and stored by the vehicles, while the lookups
heart of CoCar is a TIC. Vehicles equipped with the CoCar are performed using distributed hash tables (DHTs). Each
system send traffic reports to the centre using Internet access entry of a DHT is a tuple (key, traffic information). The
via UMTS. The reports are aggregated and integrated with key identifies the road segment. In contrast to most P2P
information obtained from other sources. The centre sends applications, the usage pattern in the CTIS context has
processed traffic information to all vehicles that belong to the locality correlations: vehicles typically request or modify the
cell from which the reports were originated. Communication reports about segments that are geographically close together.
between the TIC and vehicles is made using Fast Traffic Alert In order to keep these correlations the authors propose to
Protocol (FTAP). In order to limit the number of reports sent remove hashing of the key from DHTs. This enables the
to the centre, a vehicle reports an incident to the TIC only if it topology of the geographical area to be maintained within

166
the overlay network. Each key is specified by geographical V. D ISCUSSION AND CONCLUSION
coordinates of a road segment. The key space is divided
into zones (one zone might contain several keys), which are The future TIS will rely on the cooperation of vehicles,
then assigned to vehicles. Each vehicle stores the information their drivers, infrastructure and traffic information agencies.
assigned to all keys that belong to its zone. In addition, it is Recently, several approaches introducing cooperative data
responsible for monitoring the situation concerning the zone. collection and dissemination by vehicles have been proposed.
Participants in the system send their traffic reports to the The primary differences between them lie in the underlying
vehicles responsible for the location covered by the report. communications technologies. A comparison of CTIS is
When a new vehicle joins, it takes over half of the zone of given in Table I. The main issue of the infrastructureless
one of the system participants. When the vehicle leaves, its systems – how to provide a scalable solution – is answered
zone is merged with other zones. The system was evaluated in several ways: by aggregating traffic information (SOTIS,
using the VISSIM road traffic simulator [31]. TrafficView, StreetSmart), by sharing only the highest rank
PeerTIS is further improved in [11]. The street network reports (Trafficinfo) or by limiting communication to selected
is represented by a graph, which is partitioned into sub- nodes (IFTIS). These systems cannot be deployed in the near
graphs when a new vehicle joins the system. The handling future, as one has to wait until the necessary market penetra-
of the updates is more efficient due to the introduction of tion of V2V communications technologies has been reached.
a publish/subscribe scheme. Vehicles are informed about However, the infrastructureless systems do not require such
relevant changes regarding the segments they are subscribed a high penetration rate of these technologies as one might
to. The improved PeerTIS was evaluated using the SUMO expect: the authors of TrafficRep demonstrate that a market
road traffic simulator [32] coupled with the OverSim P2P penetration between 3 and 10% is enough to obtain the
simulator [33]. A realistic map of Düsseldorf was taken from full benefits of V2V communication for CTIS. On the other
the OpenStreetMap project [34]. hand, due to the availability of low-cost cellular Internet,
the infrastructure-based CTIS can be deployed in the near
E. Hybrid system future. A client-server architecture allows such systems to
The idea of using a P2P overlay over an infrastructure- globally optimise traffic flows (e.g. TraffCon), and include
based wireless connection is also exploited in [35]. However, information from non-vehicular sources (e.g. CoCar). As
there are two main differences between the system proposed infrastructure-based CTIS typically imply a centralised au-
in this work (hereafter referred to as a Hybrid system) and thority running the system, they have higher potential interest
PeerTIS. Firstly, in addition to a P2P overlay, it is also based from commercial service providers and traffic authorities.
on V2V communication. Therefore, a two-tier architecture is However, they introduce a single point of failure. This prob-
applied: in the lower tier vehicles communicate using direct lem is avoided if a P2P-based approach is used instead, that
V2V links, while in the upper tier a P2P overlay network is, vehicles form a self-organising overlay network hosted
is created. Secondly, unstructured P2P is used (PeerTIS by the Internet Protocol. However, the question that arises
has structured P2P). This means that system users have no in such a case is on how to efficiently perform information
knowledge about the mapping between traffic reports and the lookups. For instance, in PeerTIS this is addressed by using
node that stores them. Vehicles that are close to each other DHT without hashing. The Hybrid system proposes another
are grouped into clusters using the Max-Min heuristic. The approach by combining unstructured P2P and clustering
heuristic is also applied to select cluster heads. The vehicles techniques (leaving the lookup duty to the clusterheads).
selected as cluster heads are called supernodes, while the In order to identify the areas of interest, a common
remaining system users are referred to as regular nodes. solution is to use fixed road segmentation known a pri-
Due to the mobility of nodes the selection of supernodes ori to system users (SOTIS, TrafficRep, IFTIS, Trafficinfo,
is performed periodically. Members of a cluster exchange PeerTis). The significant question that each system should
traffic information (defined as driving speed) using V2V address is when to send the reports. In SOTIS the broadcast
communication. The supernode aggregates the reports into intervals are adapted according to predefined provocation and
a single value. A two-step procedure is used. Firstly, the mollification events. Participants in TrafficRep send reports
average driving speed in the cluster is calculated. Secondly, when reaching the end of a road segment only if they have
this average speed is mapped to one of the four traffic significant reports to share. In StreetSmart only group leaders
Level of Service (LOS) classes. For instance, if the average send the reports. The systems also differ in the type of
recorded speed on the highway is between 0 and 40 km/h, information that is exchanged. It is typically average speed
then it is mapped to value D. When a supernode needs traffic (SOTIS, TrafficView, StreetSmart), travel time per segment
information about other clusters, it will contact other cluster (TrafficRep, PeerTis) or traffic density per segment (IFTIS).
heads using scoped flooding. A regular node requests traffic On the other hand, in the TraffCon system there is no
information by contacting the supernode of its cluster (using exchange of traffic information between vehicles. System
V2V communication). In order to limit the overhead of the participants submit their destinations to a centralised server,
lookup procedure carried out by supernodes, the authors which returns driving directions. Therefore, cooperation be-
propose adapting a location-based routing protocol. The tween vehicles is achieved indirectly, using a centralised
system was evaluated using the SUMO road traffic simulator. entity.

167
 TABLE I
A COMPARISON OF INFRASTRUCTURELESS (V2V COMMUNICATION - BASED ) AND INFRASTRUCTURE - BASED CTIS: COMMUNICATION SCHEMES ,
SCALABILITY TECHNIQUES , TYPES OF EXCHANGED EVENTS AND SIMULATION TOOLS USED TO EVALUATE THE SYSTEM .

communication schemes scalability event types simulation tools

SOTIS V2V information aggregation avr. speed, warnings ns-2
V2V-based

TrafficView V2V information aggregation average speed ns-2, CORSIM

TrafficRep V2V only significant information is exchanged travel time Paramics
StreetSmart V2V information aggregation, only unusual information is exchanged average speed self-developped
IFTIS V2V broadcasts only by group leaders traffic density Qualnet, VanetMobiSim
Trafficinfo V2V exchange limited to top-rank reports travel time STRAW
infrastructure

SOCRATES V2B - travel time -

CoCar V2B information aggregation warning messages OMNeT++, SUMO
TraffCon V2B - destination, directions SWANS
PeerTIS P2P via Internet P2P paradigm travel time SUMO, OverSim
Hybrid P2P via Internet, V2V P2P, lookups by supernodes, information aggregation average speed class SUMO

The choice between infrastructure-based and infrastruc- [16] T. Nadeem, P. Shankar, and L. Iftode, “A comparative study of data
tureless solutions, as outlined in Section II, is multi-objective dissemination models for VANETs,” in Proc. Third Annual Interna-
tional Conference on Mobile and Ubiquitous Systems: Networking and
in its nature. It particular, it depends on the penetration ratio Services (MobiQuitous 2006), 2006, pp. 1–10.
of a given communications technology, the level of security [17] S. Goel, T. Imielinski, and K. Ozbay, “Ascertaining viability of WiFi
desired, and involvement of traffic authorities. based vehicle-to-vehicle network for traffic information dissemina-
tion,” in Proc. The 7th International IEEE Conference on Intelligent
Transportation Systems (ITSC 2004), 2004, pp. 1086 – 1091.
R EFERENCES [18] Quadstone Paramics v4.0, [online] http://www.paramics-online.com/.
[19] S. Dornbush and A. Joshi, “StreetSmart Traffic: Discovering and
disseminating automobile congestion using VANETs,” in Proc. IEEE
[1] R. P.-Zeletin, I. Radusch, and M. A. Rigani, Vehicular-2-X Commu- 65th Vehicular Technology Conference (VTC2007-Spring), 2007, pp.
nication: State-of-the-Art and Research in Mobile Vehicular Ad hoc 11–15.
Networks. Springer, 2010. [20] M. Jerbi, S.-M. Senouci, T. Rasheed, and Y. Ghamri-Doudane, “An
[2] “Executive summary of the ITS strategic research plan, 2010–2014,” infrastructure-free traffic information system for vehicular networks,”
U.S. Departament of Transportation, Tech. Rep., 2009. in Proc. IEEE 66th Vehicular Technology Conference (VTC-2007 Fall),
[3] L. Wischof, L. Ebner, and H. Rohling, “Information dissemination in 2007, pp. 2086 – 2090.
self-organizing intervehicle networks,” IEEE Transactions on Intelli- [21] Qualnet simulator, [online] http://www.scalable-
gent Transportation Systems, vol. 6, no. 1, pp. 90–101, 2005. networks.com/products/qualnet/.
[4] C. Lochert, J. Rybicki, B. Scheuermann, and M. Mauve, “Scalable data [22] VanetMobiSim mobility generator, [online] http://vanet.eurecom.fr/.
dissemination for inter-vehicle-communication: Aggregation versus [23] T. Zhong, B. Xu, P. Szczurek, and O. Wolfson, “Trafficinfo: An
peer-to-peer,” it – Information Technology, vol. 50, no. 4, pp. 237– algorithm for vanet dissemination of real-time traffic information,” in
242, 2008. Proc. 15th World Congress on Intelligent Transport Systems and ITS
[5] T. L. Willke, P. Tientrakool, and N. F. Maxemchuk, “A survey of America’s 2008 Annual Meeting, 2008.
inter-vehicle communication protocols and their applications,” IEEE [24] D. R. Choffnes and F. E. Bustamante, “An integrated mobility and
Communications Surveys & Tutorials, vol. 11, no. 2, pp. 3–20, 2009. traffic model for vehicular wireless networks,” in Proc. 2nd ACM
[6] A. Kesting, M. Treiber, and D. Helbing, “Connectivity statistics of international workshop on Vehicular ad hoc networks (VANET 2005),
store-and-forward inter-vehicle communication,” IEEE Transactions 2005, pp. 69–78.
on Intelligent Transportation Systems, vol. 11, no. 1, pp. 172–181, [25] I. Catling and F. O. de Beek, “SOCRATES: System of cellular radio
2010. for traffic efficiency and safety,” in Proc. Vehicle Navigation and
[7] L. Wischof, A. Ebner, H. Rohling, M. Lott, and R. Halfmann, Information Systems Conference, vol. 2. IEEE, 1991, pp. 147–150.
“SOTIS - a self-organizing traffic information system,” in Proc. The [26] Omnet++ traffic simulator, [online] http://www.omnetpp.org/.
57th IEEE Semiannual Vehicular Technology Conference (VTC2003- [27] A. Cappiello, I. Chabini, E. Nam, A. Lue, and M. A. Zeid, “A
Spring), 2003, pp. 2442–2446. statistical model of vehicle emissions and fuel consumption,” in
[8] J. Rybicki, B. Scheuermann, M. Koegel, and M. Mauve, “PeerTIS: Proc. 5th IEEE International Conference on Intelligent Transportation
a peer-to-peer traffic information system,” in Proc. 6th ACM interna- Systems (ITSC 2002), 2002, pp. 801–809.
tional workshop on VehiculAr InterNETworking, 2009, pp. 23–32. [28] K. Collins and G.-M. Muntean, “Route-based vehicular traffic man-
[9] C. Lochert, B. Scheuermann, and M. Mauve, “A probabilistic method agement for wireless access in vehicular environments,” in Proc. IEEE
for cooperative hierarchical aggregation of data in VANETs,” Ad Hoc 68th Vehicular Technology Conference (VTC2008-Fall), 2008, pp. 1–5.
Networks, vol. 8, no. 5, pp. 518–530, 2010. [29] SWANS network simulator, [online] http://jist.ece.cornell.edu/.
[10] M. Mauve and B. Scheuermann, “VANET convenience and efficiency [30] J. Rybicki, B. Scheuermann, W. Kiess, C. Lochert, P. Fallahi, and
applications,” in VANET Vehicular Applications and Inter-Networking M. Mauve, “Challenge: peers on wheels - a road to new traffic infor-
Technologies. Wiley, 2010, ch. 4, pp. 81–106. mation systems,” in Proc. 13th annual ACM international conference
[11] J. Rybicki, B. Pesch, M. Mauve, and B. Scheuermann, “Supporting on Mobile computing and networking (MobiCom 2007), 2007, pp.
cooperative traffic information systems through street-graph-based 215–221.
peer-to-peer networks,” in Proc. 17th GI/ITG Conference on Com- [31] VISSIM traffic simulator, [online] http://www.vissim.de/.
munication in Distributed Systems (KiVS 2011), 2011. [32] SUMO traffic simulator, [online] http://sourceforge.net/apps/ medi-
[12] C. Sommer, A. Schmidt, Y. Chen, R. German, W. Koch, and awiki/sumo/.
F. Dressler, “On the feasibility of UMTS-based traffic information [33] I. Baumgart, B. Heep, and S. Krause, “Oversim: A flexible overlay
systems,” Ad Hoc Networks, vol. 8, no. 5, pp. 506–517, 2010. network simulation framework,” in Proc. 10th IEEE Global Internet
[13] ns-2 netwtwork simulator, [online] http://nsnam.isi.edu/nsnam/. Symposium, 2007, pp. 79–84.
[14] T. Nadeem, S. Dashtinezhad, C. Liao, and L. Iftode, “TrafficView: [34] OpenStreetMap project, [online] http://www.openstreetmap.org/.
Traffic data dissemination using car-to-car communication,” ACM [35] Y.-C. Yang, C.-M. Cheng, P.-Y. Lin, and S.-L. Tsao, “A real-time road
SIGMOBILE Mobile Computing and Communications Review, vol. 8, traffic information system based on a peer-to-peer approach,” in Proc.
no. 3, pp. 6–9, 2004. IEEE Symposium on Computers and Communications (ISCC 2008),
2008, pp. 513 – 518.
[15] CORSIM simulator, [online] http://mctrans.ce.ufl.edu/featured/tsis/ ver-
sion5/corsim.htm.

168