A Review of Information Dissemination Protocols for Vehicular Ad Hoc Networks

Abstract

With the fast development in ad hoc wireless communications and vehicular technology, it is foreseeable that, in the near future, traffic information will be collected and disseminated in real-time by mobile sensors instead of fixed sensors used in the current infrastructure-based traffic information systems. A distributed network of vehicles such as a vehicular ad hoc network (VANET) can easily turn into an infrastructure-less self-organizing traffic information system, where any vehicle can participate in collecting and reporting useful traffic information such as section travel time, flow rate, and density. Disseminating traffic information relies on broadcasting protocols. Recently, there have been a significant number of broadcasting protocols for VANETs reported in the literature. In this paper, we classify and provide an in-depth review of these protocols.

Full text

IEEE COMMUNICATIONS SURVEYS & TUTORIALS, ACCEPTED FOR PUBLICATION 1
A Review of Information Dissemination
Protocols for Vehicular Ad Hoc Networks
Sooksan Panichpapiboon, Member, IEEE, and Wasan Pattara-atikom, Member, IEEE
Abstract—With the fast development in ad hoc wireless com-
munications and vehicular technology, it is foreseeable that, in the
near future, traffic information will be collected and disseminated
in real-time by mobile sensors instead of fixed sensors used
in the current infrastructure-based traffic information systems.
A distributed network of vehicles such as a vehicular ad hoc
network(VANET)caneasilyturninto an infrastructure-less
self-organizing traffic information system, where any vehicle can
participate in collecting and reporting useful traffic information
such as section travel time, flow rate, and density. Disseminating
traffic information relies on broadcasting protocols. Recently,
there have been a significant number of broadcasting protocols
for VANETs reported in the literature. In this paper, we classify
and provide an in-depth review of these protocols.
Index Terms—Broadcasting protocols, vehicular ad hoc net-
works, traffic information systems.
I. INTRODUCTION
WITH the fast development in ad hoc wireless commu-
nications and vehicular technology, it is foreseeable
that, in the near future, there will be a paradigm shift in traffic
information systems. In particular, real-time traffic data will
be collected and disseminated by distributed mobile probes,
instead of fixed sensors used in the current infrastructure-
based systems. A distributed network of vehicles such as a
vehicular ad hoc network (VANET) can easily turn into an
infrastructure-less self-organizing traffic information system,
where any vehicle can become a mobile sensor, participating
in collecting and disseminating useful traffic information such
as section travel time, flow rate, and density.
Disseminating traffic information in a VANET is a unique
problem. In contrast to the unicast data typically transmitted in
a network such as the Internet, the traffic information generally
has a broadcast-oriented nature. In other words, the traffic
information is of public interest, and it usually benefits a
group of users rather than a specific individual. Consequently,
it is more appropriate to use a broadcasting scheme rather
than a unicast routing scheme in disseminating the traffic
information. The main advantage of a broadcasting scheme
is that a vehicle does not need to know a destination address
and a route to a specific destination. This eliminates the com-
plexity of route discovery, address resolution, and topology
Manuscript received 26 September 2010; revised 26 March 2011.
S. Panichpapiboon is with the Faculty of Information Technology, King
Mongkut’s Institute of Technology Ladkrabang, Bangkok 10520, Thailand
(e-mail: sooksan@it.kmitl.ac.th).
W. Pattara-atikom is with the Network Technology Laboratory, Na-
tional Electronics and Computer Technology Center (NECTEC), Pathumthani
12120, Thailand (e-mail: wasan@nectec.or.th).
Digital Object Identifier 10.1109/SURV.2011.070711.00131
management, which are difficulties in dynamic networks such
as VANETs. In this paper, we will mainly focus our attention
on broadcasting protocols for VANETs. For information on
unicast routing protocols, readers are referred to [1], [2]. In
addition, we will only concentrate on source broadcasting
(i.e., distributing packets in a one-to-all type of scenarios).
Other types of information dissemination methods such as
geocasting [3], [4], [5], multicasting [6], [7], [8], peer-to-peer
content distributing [9], [10], [11], and streaming [12], [13],
[14] will not be discussed in this paper.
Over the past few years, there have been a number of
broadcasting protocols for VANETs reported in the literature.
However, they can generally be divided into two main cate-
gories:
•Multi-hop Broadcasting
•Single-hop Broadcasting
A major contrast between these two types of protocols is in
the way that the information packets are spread in the network.
In multi-hop broadcasting, a packet propagates through the
network by way of flooding. In general, when a source vehicle
broadcasts an information packet, some of the vehicles within
the vicinity of the source will become the next relay vehicles
(nodes) and perform a relaying task by rebroadcasting the
packet further. Similarly, after a relay node rebroadcasts the
packet, some of the vehicles in its vicinity will become the
next relay nodes and forward the packet further. As a result,
the information packet will be able to propagate from the
source to the other distant vehicles.
On the contrary, in single-hop broadcasting, vehicles do not
flood the information packets. Instead, when a vehicle receives
a packet, it keeps the information in its on-board database.
Periodically, each vehicle selects some of the records in its
database to broadcast. Thus, with single-hop broadcasting,
each vehicle will carry the traffic information with itself as
it travels, and this information will be transferred to other
vehicles in its one-hop neighborhood in the next broadcast
cycles. Ultimately, a single-hop broadcasting protocol relies
heavily on vehicle mobility in spreading the information. As a
first glance, the broadcasting protocols discussed in this paper
are classified as shown in Fig. 1.
The rest of the paper is organized as follows. Multi-hop
broadcasting protocols and single-hop broadcasting protocols
are reviewed in Section II and in Section III, respectively. In
Section IV, we discuss the performance evaluation of these
protocols. Summary and discussion on open issues related to
broadcasting protocols in VANETs are provided in Section V.
A list of acronyms is given in the Appendix.
1553-877X/11/$25.00 c
2011 IEEE
This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

2IEEE COMMUNICATIONS SURVEYS & TUTORIALS, ACCEPTED FOR PUBLICATION
Fig. 1. A classification of broadcasting protocols for VANETs.
II. MULTI-HOP BROADCASTING
As mentioned earlier, in multi-hop broadcasting, an infor-
mation packet propagates through the network by way of
flooding. However, a pure flooding scheme, where every single
vehicle rebroadcasts the packet, is inefficient because of two
main reasons: (i) scalability and (ii) packet collision. As the
network becomes denser, the same information packet will
be rebroadcasted more redundantly. This wastes the limited
radio channel bandwidth; thus, it makes pure flooding not
scaled with the network density. In addition, in a dense
network, packet collision becomes a severe problem since a
large number of vehicles in the same vicinity may rebroadcast
the packet at the same time. This is usually referred to as a
broadcast storm problem [15]. A good multi-hop broadcasting
protocol must be able to resolve these issues.
A common solution employed by most researchers to solve
the scalability and collision problems is reducing the number
of redundant rebroadcast packets. This is typically done by
selecting only some of the vehicles to relay the packet as
opposed to letting every single vehicle rebroadcast it. In the
succeeding sections, we discuss existing approaches used in
reducing the number of packet transmissions in details.
A. Delay-Based Multi-hop Broadcasting
In a delay-based approach, a different waiting delay be-
fore rebroadcasting the packet is assigned to each receiving
vehicle [16], [17], [18], [19], [20], [21], [22], [23], [24],
[25], [26]. Basically, the vehicle with the shortest waiting
delay gets the highest priority in rebroadcasting the packet.
In addition, in order to avoid redundancy, the other vehicles
abort their waiting process once they know that the packet
has already been rebroadcasted. Typically, the delay assigned
to each vehicle is a function of the distance between the
vehicle and the transmitter. Generally, the farthest vehicle
is given the shortest delay and is implicitly selected as the
next rebroadcast node, since it maximizes the packet forward
progress. In the following, we review a variety of delay-based
multi-hop broadcasting protocols.
1) Urban Multi-hop Broadcast (UMB): The UMB proto-
col [16] is designed to solve the broadcast storm, the hidden
node, and the reliability problems in multi-hop broadcasting.
Basically, UMB divides a road inside the transmission range of
a transmitter into small segments, and it gives the rebroadcast
priority to the vehicles that belong in the farthest segment.
In UMB, two types of packet forwarding are defined: (i) di-
rectional broadcast and (ii) intersection broadcast. The direc-
tional broadcast works as follows. When a vehicle has a packet
to send, it first transmits a control packet called Request-to-
Broadcast (RTB), which includes its own position and the
direction of packet propagation. Once the vehicles in the
transmission range of the transmitter receives the RTB packet,
each of them starts transmitting a jamming signal called black-
burst for a specified period of time. The duration of the black-
burst is a function of the distance between the vehicle that
receives the RTB packet and the transmitter. More specifically,
each vehicle computes its own black-burst duration according
to the following function
L=d
R×Nmax×S(1)
where Lis the black-burst duration, dis the distance between
the vehicle and the transmitter, Ris the transmission range,
Nmax is the number of road segments inside the transmission
range, and Sis the time slot duration. With this assignment
function, a vehicle that is farther away from the transmitter
will have a longer black-burst duration.
This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

PANICHPAPIBOON and PATTARA-ATIKOM: A REVIEW OF INFORMATION DISSEMINATION PROTOCOLS FOR VEHICULAR AD HOC NETWORKS 3
After a vehicle has transmitted the black-burst according to
the duration calculated in (1), it listens to the channel again.
If it detects that the channel is busy, then this means that the
other vehicles are still transmitting the black-burst. In this case,
the vehicle does nothing and relegates the rebroadcast duty to
those vehicles that are still transmitting the black-burst. On
the other hand, if the vehicle finds the channel idle after its
black-burst transmission, it transmits a control packet called
Clear-to-Broadcast (CTB) back to the vehicle that initiates the
RTB packet. The vehicle that successfully transmits the CTB
packet will be designated as the next rebroadcast node.
Once the vehicle that initiates the RTB packet receives the
CTB packet, it can start transmitting a DATA packet. When
the designated rebroadcast node receives the DATA packet,
it acknowledges by transmitting an ACK packet back to the
data transmitter. If the ACK packet is not received within a
certain amount of time, the whole process (RTB-CTB-DATA-
ACK) starts over again. However, UMB is not a collision-free
protocol. It is possible that there are more than one vehicle
on the same road segment, and these vehicles may transmit
the CTB packet at the same time, which will result in a
collision. It is proposed in [16] that the collision be resolved
among these vehicles by further splitting the segment into
Nmax subsegments and let the vehicles in conflict repeat the
black-burst transmission process again until the transmission
of the CTB packet is successful. The vehicle that wins the
CTB contention will be designated as the next rebroadcast
vehicle.
The second type of forwarding function defined in UMB
is intersection broadcast. This forwarding function is used for
broadcasting a packet at a road intersection. It is suggested that
a repeater be installed at an intersection in order to forward
a packet to other road directions. The UMB protocol is later
extended in [17] so that fixed repeaters at intersections can be
eliminated.
2) Smart Broadcast (SB): SB [18] is proposed to improve
the limitation of UMB. It is pointed out in [18] that UMB is
inefficient in a sense that the next rebroadcast vehicle has to
wait the longest before being able to transmit the CTB packet.
This is because the longest black-burst duration is assigned
to the next rebroadcast vehicle. SB solves this problem by
assigning the next rebroadcast vehicle the shortest waiting
delay.
A packet forwarding process in SB is done as follows. First,
when a source vehicle has a packet to send, it transmits a RTB
packet which contains its location and other information such
as packet propagation direction and contention window size.
Second, the vehicles in the range of the source that receive
the RTB packet determine the “sector” in which they belong
by comparing their locations with that of the source vehicle.
Next, each vehicle that receives the RTB packet chooses a
contention delay based on the sector that it resides. According
to [18], given that there are Nssectors, the waiting delay
Wrfor vehicles in sector ris randomly obtained from the
following set
Wr={(r−1)cw, (r−1)cw +1,...,rcw−1}(2)
where r=1,2,...,N
sis the sector number (r=1refers
to the outermost sector), and cw is the duration of contention
window. With this delay function, the vehicles in the outermost
sector (i.e., farthest away from the source) will be given the
shortest waiting time. In addition, since vehicles in the same
sector randomly pick the waiting times, this further reduces
packet collision.
When its waiting time expires, a vehicle transmits a CTB
packet back to the source. Once the source receives the CTB
packet successfully, it then transmits the data packet. The data
packet also specifies the identification (ID) of the vehicle that
is chosen as the next rebroadcast vehicle. This protocol is
better than UMB in terms of latency. As mentioned earlier,
a designated rebroadcast vehicle in the UMB protocol will
have the highest waiting delay before rebroadcasting, but in
the SB protocol a designated rebroadcast vehicle will have the
shortest delay.
In [18], the performance of SB is compared with that of
two other protocols, which are the UMB protocol and the Ge-
ographic Random Forwarding (GeRaF) protocol [27]. GeRaF
is a position-based routing protocol, where a packet will
be routed toward the geographical location of a destination.
Instead of using a destination address, a source specifies the
geographical area of the destination node in a packet. A packet
will be cooperatively forwarded by intermediate nodes that
receive it. Basically, when a node receives a packet, it decides
whether it should act as a relay based on its relative position
to the destination. It is shown that SB outperforms UMB and
GeRaF in message propagation speed. The main reason is that
UMB and GeRaF have a higher number of collisions when
the vehicle density increases; therefore, they waste more time
in resolving the collisions. It is also shown that the message
propagation speed in SB is constant even when the vehicle
density increases. Thus, SB is also robust to density changes.
In addition, it is also later reported in [28] that a timer-based
protocol such as SB performs better than a black-burst-based
protocol such as UMB.
3) Efficient Directional Broadcast (EDB): EDB [19] is a
delay-based multi-hop broadcasting protocol that works quite
similar to UMB and SB protocols. However, the RTB and CTB
control packets are not used in this protocol. In addition, EDB
also exploits the use of directional antennas. In particular, it is
proposed that each vehicle be equipped with two directional
antennas, each with 30-degree beamwidth.
Similar to UMB, there are two types of packet forwarding
in EDB, namely directional broadcast on the road segment
and directional broadcast at the intersection. In directional
broadcast on the road segment, a source vehicle broadcasts a
packet and the downstream vehicles will rebroadcast it further.
To reduce the number of redundant rebroadcast packets, EDB
assigns a different waiting time before rebroadcasting to each
vehicle within the range of the transmitter. The waiting time
is a function of the distance between the vehicle and the
transmitter. In fact, when a vehicle receives a packet, it
computes its own waiting time according to the following
function
W=1−d
RmaxW T (3)
where Ris the transmission range, dis the distance between
the vehicle and the transmitter, and maxW T is the maximum
This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

4IEEE COMMUNICATIONS SURVEYS & TUTORIALS, ACCEPTED FOR PUBLICATION
waiting time. With this waiting time assignment function, a
vehicle that is farther away from the transmitter will be given
a higher rebroadcast priority. After its waiting time expires, the
vehicle immediately transmits an ACK packet. This is done
to inform the other neighboring vehicles that they need not
perform the rebroadcasting task. After transmitting the ACK
packet, the vehicle can start rebroadcasting the data packet.
In addition, in order to increase reliability, the vehicle will
periodically keep rebroadcasting the packet if no other vehicle
forwards the packet within the interval maxW T . Similar to
the UMB protocol, it is proposed in [19] that a repeater be
installed at an intersection to broadcast the packets further to
other road directions.
The performance of EDB is compared, by simulation,
with two other variations called Random Directional Broad-
cast (RDB) and Simple Distance-based Directional Broadcast
(SDDB). These variations differ by how a waiting delay is
assigned. In RDB, each receiver simply waits for a random
time before rebroadcasting the packet, and in SDDB each
receiver waits for the duration computed in (3) but no ACK
packet is used. It is shown that the EDB protocol performs
better than the other two variations in terms of packet delivery
ratio and average forward nodes ratio (a ratio between the
number of vehicles that rebroadcast the packet and the total
number of vehicles in the network). The main reason why
EDB has a higher packet delivery ratio is that it can reduce
packet collision through the use of ACK packets. In addition,
since the neighbors of a rebroadcast node are informed through
a quick transmission of an ACK packet, they can refrain
from rebroadcasting the duplicates. This improves the average
forward nodes ratio.
4) Slotted 1-Persistence Broadcasting: Packet forwarding
in the Slotted 1-Persistence Broadcasting protocol [20] is
similar to those of the other delay-based multi-hop broad-
casting protocols, where the rebroadcast priority is given to
the vehicles that are farther away from the transmitter. In this
protocol, when a vehicle receives a packet, it rebroadcasts
the packet according to an assigned time slot, where the time
slot is a function of the distance between the vehicle and the
transmitter. In particular, each vehicle computes the time slot
in which it will rebroadcast the packet based on the following
function
TSij =Sij ×τ(4)
where τis the estimated one-hop propagation and medium
access delay, and Sij is the assigned slot number. The assigned
slot number is computed from
Sij =Ns1−min(Dij ,R)
R (5)
where Dij is the distance between transmitter iand vehicle j,
Ris the transmission range, and Nsis the pre-determined
number of slots. It is stated in [20] that Nsshould be chosen
carefully based on the traffic density. In general, the value
of Nsshould increase as the traffic becomes denser. Next,
a vehicle rebroadcasts the packet in the time slot computed
in (4) if it hears the packet for the first time and no one has
transmitted the packet before its waiting time expires. It is
possible that more than one vehicle will transmit in the same
time slot, resulting in packet collision. Thus, the number of
slots will have an impact on the protocol performance.
A similar slotted-based broadcasting protocol called
Vehicle-density-based Emergency Broadcasting (VDEB),
where a waiting time slot is assigned based on the distance
between a transmitter and a receiver, is presented in [24].
As an improvement over the Slotted 1-persistence protocol,
VDEB explicitly takes the vehicle density into consideration
when determining an appropriate number of slots to use. The
density is estimated from the number of neighbors around
the transmitter. Another variation of the Slotted 1-persistence
protocol can also be found in [25].
5) Reliable Method for Disseminating Safety Information
(RMDSI): The RMDSI protocol [21] aims at solving the reli-
ability problem when the network becomes disconnected. This
protocol also uses delay to differentiate the rebroadcast priority
of each vehicle. Similar to EDB, when a vehicle receives a
packet, it computes a waiting time before rebroadcasting the
packet according to the function given in (3). After the waiting
time expires, the vehicle rebroadcasts the packet. Vehicles that
hear the duplicate rebroadcast before their waiting times expire
cancel their retransmissions.
An additional feature of this protocol is a mechanism
for solving the network fragmentation problem. Basically,
a relay vehicle will keep a copy of the packet that it has
just rebroadcasted until it hears a duplicate transmission by
the next relay node or until the packet lifetime expires. Not
hearing a duplicate rebroadcast by the next relay vehicle is
an indication that the network may be fragmented. In this
case, the relay vehicle keeps rebroadcasting a small control
packet until it finds the next relay vehicle to which it can
forward the packet. In the case that the packet lifetime expires
before the vehicle is able to find the next relay vehicle, it stops
rebroadcasting that packet.
The performance of this protocol, in terms of the percentage
of vehicles that receive the broadcast packets, is compared
with that of the UMB protocol. It is shown, by simulation, that
when the network is heavily fragmented, RMDSI performs
better than UMB which does not take the network fragmen-
tation problem into consideration.
6) Multi-hop Vehicular Broadcast (MHVB): MHVB [22]
is also a delay-based multi-hop broadcasting protocol. Like
other protocols in this category, when a vehicle receives a
packet, it computes the waiting time before rebroadcasting the
packet based on the distance between itself and the transmitter.
A shorter waiting time will be assigned to a vehicle that
is farther away from the transmitter. However, the waiting
time assignment function is not explicitly given in [22]. After
the waiting time expires, the vehicle rebroadcasts the packet.
Vehicles that hear a duplicate rebroadcast from a vehicle that is
farther away from themselves cancel their packet rebroadcast
processes.
In addition, MHVB also has a traffic congestion detection
feature. Intuitively, when the traffic is congested, the network
will be dense. As a result, the interval in which each vehicle
broadcasts its own information should be extended. The traffic
congestion detection mechanism in MHVB is done as follows.
Basically, each vehicle uses the number of its neighbors and
its speed as an indication of congestion. For example, if
This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

PANICHPAPIBOON and PATTARA-ATIKOM: A REVIEW OF INFORMATION DISSEMINATION PROTOCOLS FOR VEHICULAR AD HOC NETWORKS 5
vehicle Adetects that the number of its neighbors is more than
a threshold Xand its own speed is less than a threshold Vmax,
then this might indicate that the traffic is congested. In this
case, a source vehicle increases its periodic broadcast interval.
The performance of this protocol is evaluated via simulation.
However, its performance is not compared with other existing
protocols.
7) Reliable Broadcasting of Life Safety Messages (RBLSM):
In RBLSM [23], after receiving a packet from the source,
each node in the transmission range of the source determines
its waiting time before rebroadcasting the packet. In contrast
to the conventional strategy where the rebroadcast priority
is given to the farthest vehicle, in this protocol the priority
is given to the vehicle nearest to the transmitter. The main
reason is that the nearer vehicle is considered more reliable
than the vehicles that are farther away from the transmitter.
For example, a nearer vehicle supposedly has a better received
signal strength. This protocol also employs the use of the RTB
and CTB control packets. The performance of the protocol is
evaluated via simulation; however, only a single hop latency
is provided.
A similar protocol which assigns the waiting delay based
on a link quality, called Link-based Distributed Multi-hop
Broadcast (LDMB), is also proposed in [26]. In computing
the waiting delay, LDMB not only takes the distance between
a transmitter and a receiver into consideration but it also
takes other factors such as traffic density, transmission range,
and packet transmission rate into account as well. However,
LDMB does not perform significantly better, in terms of
packet delivery ratio, than a simple distance-based protocol
where the broadcast priority is given to the farther vehicle.
B. Probabilistic-based Multi-hop Broadcasting
While a different delay is assigned to each vehicle in
a delay-based broadcasting protocol, a different rebroadcast
probability is assigned to each vehicle in a probabilistic-based
protocol. In probabilistic-based broadcasting, each vehicle
rebroadcasts a packet according to its assigned rebroadcast
probability [29], [30], [31], [32]. Since not all the vehicles
will rebroadcast the packet, the number of redundant packets
as well as the number of collisions are reduced. One of the
main challenges in this type of protocols is in determining
an optimal probability assignment function. There are many
ways to assign the rebroadcast probability (also referred to
as forwarding probability). While the simplest protocol uses
a pre-defined fixed value for the forwarding probability, more
sophisticated protocols let each vehicle adjust its forwarding
probability dynamically based on factors such as vehicle
location and network density. Some of the probabilistic-based
multi-hop broadcasting protocols proposed in the literature are
discussed in the following.
1) Weighted p-Persistence: In the Weighted p-Persistence
protocol [20], a vehicle that receives a packet for the first
time computes its own rebroadcast probability based on the
distance between itself and the transmitter. The distance can
be obtained by comparing its current position with the posi-
tion of the transmitter specified in the packet. In particular,
the rebroadcast probability is computed from the following
function [20]
pij =Dij
R(6)
where Dij is the distance between transmitter iand vehicle j,
and Ris the transmission range. Based on this function, the
vehicles that are farther away from the transmitter will be
given higher rebroadcast probabilities. However, this probabil-
ity assignment function does not take the vehicle density into
account; therefore, the number of rebroadcast packets can still
be large if the network is dense. A similar probabilistic-based
approach, where the rebroadcast probability is proportional to
the distance between the transmitter and the receiver, is also
considered in [32].
2) Optimized Adaptive Probabilistic Broadcast (OAPB): In
OAPB [29], the forwarding probability is computed from the
local vehicle density (in terms of the number of neighbors).
To select an appropriate forwarding probability, each vehicle
uses the local density information, which is obtained from
exchanging periodic HELLO packets. In particular, when a
vehicle receives a packet, it computes its own forwarding
probability based on the following function
φ=P0+P1+P2
3(7)
where P0,P1,andP2are functions of the number of one-hop
neighbors, the number of two-hop neighbors, and a set of two-
hop neighbors that can only be reached through a particular
one-hop neighbor [29].
In addition, in order to further reduce the number of
rebroadcast vehicles, each rebroadcast vehicle that has the
same forwarding probability φwill be assigned a different
delay, which is computed from
Δ(t)=Δ(t)max ×(1 −φ)+δ(8)
where Δ(t)max is a maximum delay time and δis a random
variable which takes values on the order of milli-seconds.
The performance of OAPB is compared with that of the
Deterministic Broadcast (DB) protocol, where each vehicle
rebroadcasts with a fixed probability. It is shown that OAPB
outperforms DB in terms of broadcast overheads and broadcast
delivery ratio. This is clearly due to the fact that nodes in
OAPB are allowed to adjust the forwarding probability based
on the network characteristics.
3) AutoCast: In AutoCast [30], the rebroadcast probability
is calculated from the number of neighbors around the vehicle.
In particular, the probability is obtained from the following
function
p=2
Nh×0.4(9)
where Nhis the number of one-hop neighbors. With this
probability assignment function, the rebroadcast probability
decreases as the number of neighbors increases. Obviously,
this function only works when the number of neighbors, Nh,is
greater than or equal to 5. However, it is not clearly specified
in [30] how the probability is assigned in the cases where
Nh<5.
Due to the nature of probabilistic flooding, a packet may
not always be able to reach the distant vehicles because some
This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

6IEEE COMMUNICATIONS SURVEYS & TUTORIALS, ACCEPTED FOR PUBLICATION
vehicles may decide not to forward it. In order to increase
coverage and reachability, in this protocol, a packet is also
rebroadcasted periodically. The rebroadcast interval is also
adjusted dynamically based on the following function
t=Nh/α (10)
where αis a constant specifying the desired number of
broadcast packets per second. The performance of AutoCast
is also compared with pure flooding and the other protocols
called MILE and MILE-on-demand [33]. MILE is simply a
periodic broadcast protocol where a node broadcasts the re-
ceived data periodically. On the other hand, MILE-on-demand
is an improved version of MILE where each node periodically
broadcasts small metadata units instead of a complete data unit
in each broadcast cycle. In addition, a node can also request
any missing data on demand. This reduces the amount of data
that need to be transmitted and updated in each cycle at the
cost of additional delay. It is shown that AutoCast outperforms
these protocols in terms of delivery ratio and dissemination
speed. The reason is that AutoCast also takes density (i.e.,
in terms of number of neighbors) into consideration when
assigning the rebroadcast probability. This helps in reducing
packet collision and thus increases the delivery ratio.
4) Irresponsible Forwarding (IF): The IF protocol [31],
[34] assigns the forwarding probability based on the distance
between the vehicle and the transmitter as well as the vehicle
density. In IF, when a vehicle receives a packet from a trans-
mitter, it computes its own forwarding probability according
to the following function
p=e−ρs(z−d)
c(11)
where ρsis the vehicle density, zis the transmission range,
dis the distance between the vehicle and the transmitter,
and c≥1is a coefficient which can be selected to shape
the rebroadcast probability. Basically, the higher the value
of cis, the higher will be the rebroadcast probability. Note
that the forwarding probability given in (11) is different from
the conventional forwarding probability assignment function,
where the probability is normally a linear function of the
distance. Based on this probability assignment function, the
rebroadcast probability increases as the distance between the
vehicle and the transmitter increases. In addition, the rebroad-
cast probability decreases as the network becomes denser,
which is a desirable property.
It is shown in [31] that the number of rebroadcast packets
can be controlled by adjusting the shaping parameter c.In
addition, the IF protocol is able to keep the expected number of
rebroadcast packets at a constant level even when the vehicle
density increases. Thus, it scales with the network density.
C. Network Coding-Based Multi-hop Broadcasting
Recently, network coding has caught attentions of many
researchers in the field of ad hoc wireless communications.
Network coding is a new way of information dissemination
which is expected to yield a much higher throughput than
the traditional way of transmission [35]. A good overview of
network coding concept can be found in [36]. The idea of
Fig. 2. (a) Example of traditional transmission (b) Example of network
coding-based transmission.
network coding and its distinction from the traditional trans-
mission approach can best be described through the following
classical example [37]. Consider a simple wireless network
shown in Fig. 2.a, where node Cis an intermediate node
between node Aand node B. In addition, in this considered
scenario, node Aand node Bare not directly connected.
Suppose that Ahas a packet destined to B,andBalso has
a packet destined to A. In a traditional packet transmission
approach, Awill need to send its packet to Cand then let
Cforward the packet further to B. Similarly, Bwill need
to send its packet to Avia C. Note that the total number
of transmissions required to complete the packet exchanges
in this case is equal to four, as illustrated in Fig. 2.a. In
contrast, using network coding, the packet exchanges between
node Aand node Bcan be accomplished in a fewer number
of transmissions as illustrated in Fig. 2.b. First, Atransmits its
packet to C. Second, Btransmits its packet to C. Next, node C
creates an encoded packet by XORing the packets it received
from Aand Btogether, and then it broadcasts the XORed
packet to both Aand B. Finally, node Aand node Beach can
decode the packet sent to them by XORing the received packet
with their own packet. Note that this process requires only
three transmissions, which is fewer than that of the traditional
approach.
As illustrated in the above example, disseminating informa-
tion with network coding potentially requires a fewer number
of packet transmissions. This helps utilize the bandwidth
more efficiently. Although there are a number of work which
investigate how to apply the concept of network coding to
broadcasting in mobile ad hoc wireless networks, there are not
many network coding-based broadcasting protocols designed
specifically for VANETs. However, since the concept may be
adapted for VANETs, we briefly mention a few of network
coding-based protocols here.
1) COPE: A network coding-based protocol, called COPE,
is introduced in [37]. Although COPE is a unicast routing
protocol, it is a foundation which many protocols build upon,
and thus it is worth mentioning here. The objective of COPE
is to extend and realize the benefit of network coding beyond
This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

PANICHPAPIBOON and PATTARA-ATIKOM: A REVIEW OF INFORMATION DISSEMINATION PROTOCOLS FOR VEHICULAR AD HOC NETWORKS 7
the simple duplex flows example discussed in Fig. 2. The
operation of COPE is based on three key techniques: (i) oppor-
tunistic listening, (ii) opportunistic coding, and (iii) neighbor
state learning. Opportunistic listening simply lets nodes take
advantages of the wireless broadcast medium by snooping
all the packets that they overhear. Each node will store the
overheard packets in its buffer for a limited period. These
packets will later be used for coding when the opportunity
presents. The second technique, opportunistic coding, defines
some ground rules for a node to encode and transmit a packet.
Basically, a node should ensure that its next hop neighbor has
enough information to decode the transmitted encoded packet.
Generally, a node will be able to correctly decode a packet pi
from an encoded packet created from packets p1,p
2,...,p
n
if it has n−1of these packets. Thus, learning what packets
its neighbors have is crucial, and this is achieved through a
periodic broadcast of reception reports. Basically, each node
periodically announces to its neighbors which packets are
stored in its reception buffer.
The concepts of network coding is extended to apply more
specifically for VANETs in [38]. Particularly, a new protocol
called Local-directed Network Coding (LDNC) is introduced.
The main idea is similar to that of COPE; however, LDNC
exploits the directions of packet propagation in VANETs. In
LDNC, all packets incoming to each node are classified and
placed into two separate queues based on their propagation
directions: (i) forward propagation and (ii) backward propa-
gation. From each node’s perspective, packets in the forward
propagation direction are those to be relayed to the vehicles in
front whereas packets in the backward propagation direction
are those to be relayed to the vehicles behind it. In addition,
each node looks for an opportunity to encode the packets
by XORing the packets from these two queues. This creates
a setting which is similar to the example given in Fig. 2.
However, LDNC is still a unicast routing protocol.
2) CODEB: CODEB is a network coding-based broad-
casting protocol introduced in [39]. It extends the concepts
and techniques used in COPE to cover broadcasting scenarios
in ad hoc wireless networks. Similar to COPE, it relies on
opportunistic listening, where each node has to snoop all pack-
ets that it overhears. Moreover, each node also periodically
broadcasts a list of its one-hop neighbors. This allows a node
to build a graph of its two-hop neighbors, which will further
be used to construct a broadcasting backbone. In addition,
CODEB also relies on opportunistic coding, which determines
if a node can exploit coding opportunities to transmit coded
packets to its neighbors. It is pointed out that opportunistic
coding for broadcast is quite different from coding for unicast.
In unicast routing, only the intended next-hop node needs
to receive a given packet whereas in broadcasting all the
neighbors of the node must receive the packet. This adds
another level of complexity as a node must ensure that all
of its neighbors are able to decode the packet.
Unlike a probabilistic-based broadcasting protocol where
the next neighbors to forward the packet are selected ran-
domly, CODEB selects a subset of neighbors to forward
the packet deterministically. Particularly, each node creates a
forwarder list using a Partial Dominant Pruning (PDP) algo-
rithm [40]. A forwarder list of a node contains the minimum
number of broadcast nodes such that all nodes in its two-
hop neighborhood are covered. Only nodes in the forwarder
list are allowed to forward the packet. However, a forwarder
may decide not to transmit the packet if it determines that
all of its neighbors have already received the given packet.
The performance of CODEB is evaluated by simulation. It
is shown that CODEB outperforms a scheme that only uses
PDP without network coding in terms of packet delivery ratio
and the number of transmissions required to deliver a packet
to all nodes in the network. This is due to the fact that
network coding can save the number of transmissions required
to broadcast a packet at the expense of coding and decoding
operations.
3) Efficient Broadcasting Using Network Coding and Di-
rectional Antennas (EBCD): EBCD is a network coding-based
broadcasting protocol that combines the advantage of network
coding with that of directional antennas [41]. Similar to
CODEB [39], EBCD determines a subset of neighboring nodes
which will be performing a forwarding task deterministically.
However, EBCD uses a different algorithm called Dynamic
Directional Connected Dominating Set (DDCDS). Ultimately,
the DDCDS algorithm constructs a directional virtual network
backbone where each node determines both its forwarding
status as well as the outgoing edges (antenna sectors) in
which the packets will be transmitted. Another main difference
between EBCD and CODEB is that, in EBCD, network coding
is applied in each sector of the directional antennas around
the node instead of omnidirectional. It is shown in [41] that
using both directional antennas and network coding provides a
significant improvement, in terms of number of transmissions,
over a scheme that uses only network coding and a scheme
that uses none of these two techniques.
4) DifCode: The goal of DifCode is to reduce the number
of transmissions required to flood packets in an ad hoc wireless
network [42]. Similar to CODEB, DifCode selects the next
forwarding nodes deterministically. However, the selection
algorithm used in DifCode is based on multi-point relay
(MPR) [43]. An MPR of a node is defined as a set of its
one-hop neighbor that covers its two-hop neighborhood. In
DifCode, each node in the network encodes and broadcasts
only the packets that are received from nodes that select it as
their MPR.
A main distinction between DifCode and CODEB is in the
opportunistic coding techniques. In CODEB, all the neighbors
of a transmitter are required to be able to decode the re-
ceived packets immediately. This limits coding opportunities.
In DifCode, this constraint is relaxed by allowing nodes to
buffer packets that are not immediately decodable. Particularly,
each node will maintain buffers for keeping three types of
packets: (i) successfully decoded packets, (ii) not immediately
decodable packets, and (iii) packets that need to be encoded
and broadcasted further. Classifying packets into these three
types allows DifCode to exploit background decoding,which
is a process where a node XORs an incoming encoded packet
with those that are stored in the type-ii buffer in order to
decode it. If decoding results in a native (uncoded) packet, then
it is moved to the type-i buffer. This improves an opportunity
of being able to decode the packets. It is shown in [42]
that DifCode is able to achieve a lower redundancy rate
This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

8IEEE COMMUNICATIONS SURVEYS & TUTORIALS, ACCEPTED FOR PUBLICATION
than a probabilistic network coding protocol, where the next
forwarding nodes are chosen randomly.
III. SINGLE-HOP BROADCASTING
As opposed to multi-hop broadcasting, in single-hop broad-
casting a packet broadcasted by a vehicle will not be flooded
in the network. When a vehicle receives a packet, it does not
rebroadcast the information immediately. Instead, a vehicle
updates its information database according to what it learns
from the received packet, provided that the information in
the packet is considered newer than that in its database.
Periodically, a vehicle disseminates some of the information
in its database to the other vehicles in the network. The design
choices that need to be considered in this type of protocols
are: (i) broadcast interval and (ii) information that needs to
be broadcasted. To reduce redundancy and keep the most
up-to-date information, the broadcast interval should be set
appropriately (i.e., not too long and not too short). In addition,
only the important and relevant information should be selected
to broadcast. We divide the single-hop broadcasting protocols
into two categories, which are the fixed broadcast interval
protocols and the adaptive broadcast interval protocols. While
the main focus of the fixed broadcast interval protocols is only
on the selection and aggregation of information, an adjustment
of broadcast intervals is also taken into consideration in the
adaptive broadcast interval protocols.
A. Fixed Broadcast Interval
1) TrafficInfo: In TrafficInfo [44], each vehicle in the
network periodically broadcasts the traffic information stored
in its database. A particular type of traffic information reported
is the travel times on the road segments. The broadcasting
process is done as follows. It is assumed that each vehicle
has a digital map of the road network, and each road segment
has a unique ID number. In addition, each vehicle is equipped
with a global positioning system (GPS) receiver so it knows its
own position on the digital map at any time. When a vehicle
travels through a road segment, its travel time is recorded and
kept in its on-board database. In addition to the travel time
on its current road segment, a vehicle also learns about the
travel time on the other road segments from other vehicles
when they report the travel times on their road segments.
As in any single-hop broadcasting scheme, it is inefficient to
broadcast all the records in the vehicle’s database. TrafficInfo
has a way to select the most relevant information to broadcast.
In order to use the bandwidth efficiently, only the top kmost
relevant information in the database will be broadcasted by the
vehicle. The relevance of the information is determined by a
ranking algorithm, which is based on the current location of
the vehicle and the current time. For example, if the vehicle
is on segment A, the information about the road segment B
which is adjacent to segment Ais more relevant than the
information about the road segment Cwhere Cis ten blocks
away from segment A. Basically, the relevance of information
decreases with distance and time. A formal way of ranking by
quantifying the relevance of information in terms of demand
and supply is also presented in [44].
2) TrafficView: TrafficView is a single-hop broadcasting
scheme [45] designed for enabling an exchange of traffic infor-
mation among vehicles. The types of information exchanged
among the vehicles are speed and position. In this scheme,
when a vehicle receives a broadcast packet, it stores the infor-
mation in its database. The information is then rebroadcasted
in the next broadcast cycle. However, instead of broadcasting
every record in its database, the vehicle aggregates the speed
and positions of many vehicles into a single record and
then broadcast this aggregated information. Two aggregation
algorithms, namely the ratio-based algorithm and the cost-
based algorithm, are presented. In the ratio-based algorithm,
a road will be divided into small regions, and an aggregation
ratio will be assigned to each region. An aggregation ratio in
each region is assigned based on the importance of the region
and the level of information accuracy required for that region.
A region which is assigned a small value of aggregation ratio
will have information with less accuracy. On the other hand,
the cost-based algorithm also takes the cost of aggregating
the records into consideration. The aggregation cost can be
regarded as the loss of accuracy incurred from combining the
records. The performance of these algorithms are evaluated
and compared by simulation. It is shown that although the
cost-based algorithm yields better accuracy, the ratio-based
algorithm gives more flexibility. The main reason why the
cost-based algorithm yields better accuracy than the ratio-
based algorithm is that it selects records with the lowest loss
of accuracy (i.e., minimum cost) to aggregate whereas the
ratio-based algorithm simply aggregates records blindly.
B. Adaptive Broadcast Interval
1) Collision Ratio Control Protocol (CRCP): In
CRCP [46], each vehicle disseminates the traffic information
periodically. The traffic information in this case are the
location, speed, and road ID. It is assumed that these data can
be measured at every second. In this protocol, a mechanism
for dynamically changing a broadcast interval based on
the number of packet collisions is proposed. Basically, the
protocol aims at keeping the collision ratio at a targeted level
regardless of the vehicle density. Intuitively, the number of
packet collisions increases as the network density increases.
Thus, in order to keep the number of packet collisions at the
desired level, vehicles need to adaptively adjust their broadcast
intervals. In particular, in this protocol, the broadcast interval
will be doubled if the estimated collision ratio observed by
a vehicle and the estimated bandwidth efficiency are greater
than the pre-defined thresholds. Otherwise, the broadcast
interval is shorten by one second.
In addition to the broadcast interval adjustment mechanism,
three methods for selecting the data to be disseminated are also
proposed: (i) Random Selection (RS), (ii) Vicinity Priority Se-
lection (VPS), and (iii) Vicinity Priority Selection with Queries
(VPSQ). In the RS scheme, a vehicle randomly chooses the
information in its database to disseminate, whereas in the
VPS and VPSQ schemes the information of the nearby road
segments (e.g., within a distance of xkm from the vehicle)
will be given the priority in being selected for dissemination.
In other words, a certain amount of traffic data in the nearby
This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

PANICHPAPIBOON and PATTARA-ATIKOM: A REVIEW OF INFORMATION DISSEMINATION PROTOCOLS FOR VEHICULAR AD HOC NETWORKS 9
area will be selected for transmission first. If, however, there
is still room, the data in the distant area will then be selected.
The main difference between VPS and VPSQ is that, in VPSQ,
a node also has an option of querying the traffic information in
the area of interest rather than only passively collecting them.
The performance of the CRCP protocols that use each of these
three data selection schemes are evaluated via simulation. It
is shown that the CRCP protocol which uses VPSQ as its
data selection scheme generally performs better than the cases
where it uses the other two data selection schemes. This is due
to the ability of VPSQ to query information in the relevant
areas.
2) Abiding Geocast: The Abiding Geocast protocol [47] is
designed for disseminating safety warnings within an effective
area where these warnings are still relevant and applicable. In
this scheme, when an emergency situation occurs, the first
vehicle that detects it starts broadcasting a warning packet.
In the packet, an effective area where the warning is still
relevant and should be kept alive is also specified. When
another vehicle receives the warning packet, it will become an
active relay node and keep broadcasting the warning packet
as long as it is still traveling in the effective area. The vehicle
stops broadcasting when it goes outside of the effective region.
In order to keep the number of redundant warning packets
at minimum, each vehicle adjusts its rebroadcast interval
dynamically. The rebroadcast interval is determined from the
transmission range, speed, and the relative distance between
the vehicle and the emergency site. Basically, the rebroadcast
interval increases as the distance between the active relay node
and the emergency site increases. In addition, the rebroadcast
interval also increases as the vehicle speed decreases. The
performance of the protocol is evaluated in terms of the
number of overheads via simulation. However, a comparison
with other protocols is not given.
3) Segment-oriented Data Abstraction and Dissemination
(SODAD): In this protocol [48], roads are divided into seg-
ments of known length. Each vehicle collects the data on
its current segment either by sensing the information itself
or observing what the other vehicles report. In order to
reduce the number of redundant rebroadcast packets, each
vehicle adaptively adjusts its broadcast interval. Particularly,
a vehicle adjusts its transmission interval based on the in-
formation it receives from the other vehicles. In fact, the
information received will be characterized as one of these two
events: (i) provocation and (ii) mollification. A provocation
event is defined as an event that will reduce the time until
the next broadcast, whereas a mollification event is defined as
an event that will increase the time until the next broadcast.
When a vehicle receives a packet, it will determine whether
a provocation or a mollification event has occurred. This is
done by assigning a weight to the received packet. A weight
is computed from the discrepancy between the received data
and those in the vehicle’s knowledge database. If the newly
received information is considered newer than that in the
database, then the assigned weight will be high. Based on the
weight of the packet, a node determines whether a provocation
or mollification event has occurred by comparing it to a
threshold. The time until the next rebroadcast is increased or
decreased based on the weight. In this study, the performance
of the interval adaptation scheme is compared with the static
scheme in both simulation and in a prototype system. The
results confirm that using the adaptive scheme can reduce
the number of packet collisions caused by the static periodic
broadcast scheme.
IV. PERFORMANCE EVA L U AT I O N
While it is possible to discuss the performance of the
broadcasting protocols qualitatively as shown in Table I, it is
quite challenging to quantitatively compare their performance
based on the results reported in the literature. One of the
main reasons is that these protocols are often evaluated by
different performance metrics. Moreover, most of the time
they are judiciously evaluated because only the metrics which
are in favor of their performance are presented while the
others are ignored. Thus, based on the results reported in their
original papers alone, it is not possible to make a quantitative
comparison among them effectively. Instead of attempting to
judge each protocol based on biased information, we will
discuss the metrics commonly used in the current literature
and then suggest a new unified metric that is capable of giving
a protocol a fair overall evaluation.
A. Existing Metrics
Essentially, what most researchers are interested to know
about the broadcasting protocols in VANETs are: (i) how
frequently the information packets are duplicated, (ii) how far
the information packets can propagate, and (iii) how fast the
information can be spread. The existing metrics commonly
used in the literature are listed in Table II. The first column
of Table II indicates the domain where each metric belongs.
In our perspective, these metrics can be classified into four
domains, namely frequency,space,time,andmixed.The
metrics in the frequency domain are those related to frequency
counting (e.g., counting of packets or the number of vehicles).
The metrics in the space domain involve the measurements
of distance whereas those in the time domain involve the
measurements of time. The metrics in the mixed domain are
those created from a combination of metrics in more than one
domain. The second column indicates the ID that we assign to
each metric, which will later be used for referencing purpose.
The third column of Table II lists the metric names. It can be
observed that some of the metrics are very similar although
they are called differently. The fourth column describes how
each metric is computed while the fifth column describes what
each metric is designed to measure. The sixth column specifies
the unit of each metric. Finally, the last column suggests the
favorable value for each metric (i.e., indicating whether a low
value or a high value of the considered metric is desirable).
In the frequency domain, the following three attributes of
broadcasting protocols are usually of interest, and the metrics
in this domain are designed to quantify them.
•Redundancy—Redundancy is a key performance indica-
tion of a broadcasting protocol. A good protocol should
be able to disseminate information with the least amount
of redundancy or overheads. As listed in Table II, the
metrics used for quantifying the redundancy are redun-
dancy rate,load generated per broadcast packet,forward
This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

10 IEEE COMMUNICATIONS SURVEYS & TUTORIALS, ACCEPTED FOR PUBLICATION
TAB L E I
QUALITATIVE COMPARISON OF BROADCASTING PROTOCOLS FOR VANETS.
Broadcasting
Typ e
General Characteristics Advantages Disadvantages
Multi-hop
•Packets are disseminated by ways
of smart flooding
•Reduce redundancy by varying
broadcast probability or delay
•Packets can be disseminated quickly
•Good for safety alerts and emergency
warning applications
•No need for large storage space to
keep unbroadcast packets
•Need an algorithm to deal with
the broadcast storm problem
•No packet persistency
Single-hop
•Packets are disseminated by ways
of periodic broadcast
•No packet flooding
•Reduce redundancy by varying
the broadcast period of each node
•Rely on node mobility to carry
and spread the information
•Good for applications that need
packet persistency
•No broadcast storm
•Packet dissemination speed may
be slow
•Not suitable for delay-sensitive
applications
•May require large storage space
to keep unbroadcast information
TAB L E I I
EXISTING PERFORMANCE METRICS FOR AN EVALUATION OF BROADCASTING PROTOCOLS.
This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

PANICHPAPIBOON and PATTARA-ATIKOM: A REVIEW OF INFORMATION DISSEMINATION PROTOCOLS FOR VEHICULAR AD HOC NETWORKS 11
TABLE III
BROADCASTING PROTOCOLS IN VANETS AND THE METRIC DOMAINS USED F OR EVALUATION.
Broadcasting
Typ e
Protocols Evaluation
Models
Simulation Platforms Metric Domains
Frequency Space Time Mixed
Multi-hop UMB [16] Simulation MATLAB [49], CSIM [50] 2,616
SB [18] Analysis &
Simulation
MATLAB 11 15 16
EDB [19] Simulation Proprietary 3,6
Slotted 1-Persistence [20] Simulation OPNET [51] 4,7,912 14
Weighted p-Persistence [20] Simulation OPNET 4,7,912 14
RMDSI [21] Simulation NS-2 [52] 614
MHVB [22] Simulation NS-2 6,9
RBLSM [23] Simulation MATLAB 15
VDEB [24] Simulation NS-2 114
LDMB [26] Simulation Unspecified 614
OAPB [29] Simulation NS-2 5,614
AutoCast [30] Simulation NS-2 616
IF [31], [34] Analysis &
Simulation
MATLAB, NS-2 1,7,814
CODEB [39] Simulation NS-2 6
EBCD [41] Simulation NS-2 1,6
DifCode [42] Simulation OPNET 1
Single-hop TrafficInfo [44] Simulation STRAW/SWANS [53] 6
TrafficView [45] Simulation NS-2 10
CRCP [46] Simulation NETSTREAM [54] 9
Abiding Geocast [47] Simulation OMNeT++ [55] 5
SODAD [48] Simulation NS-2 914
node ratio,link load,andbroadcast overhead. Generally,
these metrics measure the number of duplicate packets or
the number of duplicate bits used in disseminating one
information packet.
•Reachability—An information should be disseminated
in such a way that it reaches all the reachable nodes
in the target area. Keeping the redundancy low while
maintaining high reachability is one of the main chal-
lenges in designing a broadcasting protocol. The metrics
used for quantifying the reachability are delivery ratio
and reception rate. As observed from Table II, these
two metrics are defined a bit differently. Basically, the
delivery ratio is derived from the total number of vehicles
in the network whereas the reception rate is derived
from the number of reachable nodes. In other words,
the reception rate measures the proportion of “connected
nodes” that receives the broadcast packet.
•Failure rate—If the rebroadcast mechanism of a broad-
casting protocol is not designed carefully, there could
be a lot of packet collisions since many vehicles in the
same vicinity may rebroadcast the packets at the same
time. This is usually referred to as the broadcast storm
problem [15]. Obviously, the failure rate should be kept
at minimal. The metric commonly used in quantifying
the failure rate is collision ratio or packet loss ratio.
The metrics in the space domain typically measure how
far a packet can propagate. Propagation distance measures a
distance that a packet can propagate from the point where it
is originated in unit of meters whereas the number of hops
propagated measures how far a packet can traverse in terms
of the number of hops. Sustainable number of hops,onthe
other hand, measures the number of hops that a packet can
traverse while maintaining a desired quality of service (QoS),
for example, in terms of bit error rate [56]. In addition to
the total propagation distance, a progress made at each hop is
also an important quantity, and this is typically measured by
forward progress. Basically, the forward progress measures the
distance gained beyond the current transmitter if a particular
vehicle was selected as a next rebroadcast node. Normally, a
vehicle with the largest forward progress will be selected.
In the time domain, propagation time measures a time it
takes a packet to traverse from a source to the other point
in the network. This includes the “air time,” which is the
delay incurred from passing a packet from one vehicle to
the other via the wireless communication channel, and the
“ground time,” which is the delay incurred while a vehicle
is carrying the packet before rebroadcasting it to the other
vehicles. Rebroadcast latency measures a time until the packet
is received successfully by the next vehicle.
Finally, the metrics in the mixed domain are those created
from a combination of the metrics in the frequency, space,
and time domains. Based on our survey, the only metric in
the mixed domain currently defined in the literature is prop-
agation speed or dissemination speed. Basically, it measures
the distance at which a packet can traverse the network per
unit time.
B. Suggested Metric
As previously discussed, one of the main problems in
comparing the performance of broadcasting protocols is that
researchers often use different metrics in assessing the perfor-
mance of their protocols. Moreover, most of the times only the
metrics which are advantageous to the proposed protocols are
selectively presented while other metrics are ignored. In order
to illustrate this point more clearly, we list the broadcasting
protocols and the metric domains used in evaluating their
performance in Table III. In each row of the table, the metrics
used in evaluating each broadcasting protocol are indicated by
numbers. These numbers, appeared in the last four columns
of Table III, correspond to the IDs of the metrics listed in
This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

12 IEEE COMMUNICATIONS SURVEYS & TUTORIALS, ACCEPTED FOR PUBLICATION
Table II. For instance, 1 would correspond to redundancy
rate, and 16 would correspond to propagation speed. It can
be observed that most of the protocols are evaluated by the
metrics in only one or two domains. In other words, most
protocols are not given a complete evaluation from all the
angles.
In order to evaluate the performance of each protocol fairly,
the metrics from the three independent domains (i.e., fre-
quency, space, and time) should be considered. In this regard,
we introduce a new metric called Dissemination Efficiency
(DE), which is defined as
DE =Propagation Distance ×Success Ratio
Propagation Time ×Redundancy Rate.(12)
This is a simple, yet intuitive, performance metric. Dis-
semination Efficiency combines the metrics in all the three
independent domains; hence, it can truly reflect the overall
performance of a broadcasting protocol. The propagation dis-
tance is measured in meters, the propagation time is measured
in seconds, the redundancy rate and the success ratio are unit-
less; therefore, DE has a unit of m/s. Intuitively, it measures
how far an information packet can propagate through the net-
work per unit time and per the amount of overheads generated.
In addition, it is also weighted by the success ratio, which
measures the proportion of nodes that successfully receives
the broadcast packet. As a result, effects of packet collision
or failures are also well captured by DE. Ultimately, the DE
value increases if the information can be distributed farther,
faster, with high success rate, and with less redundancy.
In order to demonstrate how DE can be used to compare the
performance of the broadcasting protocols for VANETs, we
perform a simple simulation of a few multi-hop broadcasting
protocols. Particularly, the Weighted p-persistence protocol,
the IF protocol, and the Slotted 1-Persistence protocol are
investigated. The Weighted p-persistence protocol and the IF
protocol are selected to illustrate the effects of rebroadcast
probability assignment functions on the protocol performance.
In addition, the Slotted 1-persistence protocol, a delay-based
broadcasting scheme, is also selected for a comparison with
its counterpart protocol (i.e., the Weighted p-persistence proto-
col). The simulation is implemented in MATLAB by following
the approach described in [31]. In each simulation trial, N
vehicles are placed on a straight line of length Laccording
to a Poisson point distribution with density ρs=N/L.We
assume that L=10km in all the simulation scenarios. The
transmission range of each vehicle is assumed to be 200 m.
The first vehicle is designated as a source, and transmits
one packet. After the source transmits a packet, each vehicle
within the source’s transmission range decides to rebroadcast
the packet according to the protocol in consideration. In the
Weighted p-persistent protocol, each vehicle rebroadcasts the
packet with the probability given in (6). In IF, each vehicle
rebroadcasts the packet according to the rebroadcast proba-
bility given in (11) with c=5. In the Slotted 1-persistence
protocol, each vehicle rebroadcasts the packet according to
its designated time slot, where the slot number is calculated
from (4) and (5). The number of time slots, Ns,isfixed at 10
slots, and the length of each time slot, τ, is 1 ms. In all the
schemes, it is assumed that a node will not rebroadcast if it
0.1 0.2 0.3 0.4 0.5
0
0.5
1
1.5
2
2.5
3
3.5
4
ρs [veh/m]
DE
[m/s]
Irresponsible Forwarding
Weighted p−Persistence
Slotted 1−Persistence
Ideal Best
Fig. 3. Dissemination efficiency as a function of vehicle density.
hears a duplicate broadcast from the nodes downstream. The
rebroadcasting process continues until the packet dies (i.e.,
no vehicle rebroadcasts the packet further) or until the packet
reaches the last vehicle in the network, whichever occurs first.
It is assumed that the packet transmission time is 1 second.
The DE value is collected at the end of each simulation trial.
The simulation is repeated for 10,000 trials and the average
DE value is calculated.
In Fig. 3, the average DE value is shown as a function
of the vehicle density. In addition to the three multi-hop
broadcasting protocols mentioned above, we also investigate
the performance of an ideal best scheme where the packet is
always rebroadcasted by the farthest vehicle in the range of
the transmitter in each hop and the only delay incurred from
rebroadcasting in each hop is from the packet transmission
time. It can be observed from Fig. 3 that the DE value of
the Slotted 1-persistence protocol decreases as the vehicle
density increases. This is due to the fact that when the network
becomes denser, the number of vehicles that will select the
same time slot for transmission will increase. As a result,
the number of packet collisions will increase. This certainly
reduces the DE value. In contrast, it can be observed that in
the cases of IF and Weighted p-persistence protocols, the DE
values generally increase as the vehicle density increases. The
main reason is that these two protocols are probabilistic-based.
Consequently, they are able to further reduce the number
of duplicate packets as well as the number of collisions.
Moreover, it can be observed that the IF protocol is able to
achieve a higher DE value than the Weighted p-persistence
protocol. This is expected because the rebroadcast probability
assignment function in the IF protocol takes the vehicle
density into consideration whereas the probability assignment
function in the Weighted p-persistence protocol does not. This
makes the IF protocol more efficient than the Weighted p-
persistence protocol.
V. S UMMARY AND OPEN ISSUES
In this paper, we have reviewed a variety of broadcasting
techniques for information dissemination in the self-organizing
vehicular ad hoc networks. A classification of these protocols
This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

PANICHPAPIBOON and PATTARA-ATIKOM: A REVIEW OF INFORMATION DISSEMINATION PROTOCOLS FOR VEHICULAR AD HOC NETWORKS 13
is illustrated in Fig. 1. The common focus in designing
these protocols is in suppressing the excessive rebroadcast
packets. In multi-hop broadcasting protocols, the reduction
of redundant rebroadcast packets is typically done through
the delay and probability assignment functions, which adjust
the waiting delay and the rebroadcast probability based on
the vehicle location and the physical characteristics of the
network such as the vehicle density. The number of packet
transmissions can also be reduced by using a network coding
approach. In single-hop broadcasting protocols, where each
vehicle rebroadcasts the packet periodically, the suppression
of excessive rebroadcast packets is usually done by letting
each vehicle adjust its rebroadcast interval dynamically.
Based on the information reported in the literature alone, it
is quite difficult to quantitatively compare the performance of
the existing broadcasting protocols. One of the main reasons
is that these protocols are often evaluated by different metrics,
and there has not been a unified metric that can capture
the overall performance of each protocol. In this regard, we
introduce a new metric called Dissemination Efficiency, which
is a combination of the metrics in all the independent domains,
and hence it is able to reflect the overall performance of a
broadcasting protocol. Basically, the value of Dissemination
Efficiency increases if a packet can be broadcasted farther,
faster, with high success rate, and with less redundancy.
Although there are already many broadcasting protocols
proposed for VANETs, there are still open issues that need
to be considered:
•Theoretical fundamental performance limits—It is
evident from Table III, that there are only a few theoret-
ical analyses of the broadcasting protocols. Most of the
protocols are evaluated only by simulation. This makes it
difficult to determine the fundamental performance limits
of the broadcasting protocols for VANETs. A general
theoretical framework for analyzing the performance
of these protocols is worth developing. The theoretical
analytical framework should be able to model the two
main parts that significantly affect the protocol perfor-
mance. First, it should be able to model the vehicle
movement which will in turn affect the topology and
network connectivity. The vehicle movement has been
thoroughly studied in the field of transportation science.
Examples of the widely adopted mobility models are
the car-following model [57] and the cellular automata
model [58]. These mobility models should be parts of
the theoretical analytical framework.
Second, the dynamics of packet forwarding should be
modeled in the theoretical framework. How a packet is
forwarded or rebroadcasted depends on the broadcasting
protocol. In a probabilistic-based multi-hop broadcast-
ing protocol, the dynamics of packet forwarding (e.g.,
how far it can traverse, how long it takes, etc.) could
be derived from the rebroadcast probability assignment
function. Similarly, the packet forwarding behavior in
a delay-based multi-hop broadcasting protocol could be
determined from the delay assignment function. In a
single-hop broadcasting scheme, this could be derived
from the broadcast interval adjustment function.
•Assessment of the protocols in realistic scenarios—
Based on our survey, the broadcasting protocols reported
in the literature are mostly tested on a simple straight
road section. However, a situation where a dissemination
of traffic information will likely be the most useful is in
the urban scenario in which the road structures are fairly
complex. As a result, the broadcasting protocols still need
a thorough test under more complex interconnected road
structures, where the network characteristics (e.g., vehicle
density) on each section are likely to be interdependent.
A realistic evaluation could be done either by a field
experiment or a thorough simulation. There are tradeoffs
between these two methods.
Although the best way to determine which protocol
really works in practice is to conduct a field experiment,
it is still quite a challenge to carry out an experiment on
a grand scale (i.e., involving a large number of vehicles).
An alternative approach is to use an integrated simulator
that is capable of realistically simulating the road network
environment, the vehicle mobility, and the communica-
tion among vehicles. An example of such an integrated
simulator is GrooveNet [59]. In this type of simulators,
a map of a real road network can be imported as an
input, creating a realistic road topology that vehicles will
travel. In addition, GrooveNet also has a feature which
enables a communication between a real GPS-equipped
vehicle and a simulated vehicle, making it possible and
scalable to test a protocol in a field experiment. However,
GrooveNet is still not perfect and it still needs further
development.
APPENDIX
List of Acronyms
CRCP Collision Ratio Control Protocol
CTB Clear-to-Broadcast
DB Deterministic Broadcast
DDCDS Dynamic Directional Connected Dominating Set
DE Dissemination Efficiency
EBCD Efficient Broadcasting Using Network Coding and
Directional Antennas
EDB Efficient Directional Broadcast
GeRaF Geographic Random Forwarding
IF Irresponsible Forwarding
LDMB Link-based Distributed Multi-hop Broadcast
LDNC Local-directed Network Coding
MHVB Multi-hop Vehicular Broadcast
MPR Multi-point Relay
OAPB Optimized Adaptive Probabilistic Broadcast
PDP Partial Dominant Pruning
RBLSM Reliable Broadcasting of Life Safety Messages
RDB Random Directional Broadcast
RMDSI Reliable Method for Disseminating Safety Infor-
mation
RS Random Selection
RTB Request-to-Broadcast
SB Smart Broadcast
SDDB Simple Distance-based Directional Broadcast
SODAD Segment-oriented Data Abstraction and Dissemi-
nation
UMB Urban Multi-hop Broadcast
VANET Vehicular Ad Hoc Network
VDEB Vehicle-density-based Emergency Broadcasting
VPS Vicinity Priority Selection
VPSQ Vicinity Priority Selection with Queries
This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

14 IEEE COMMUNICATIONS SURVEYS & TUTORIALS, ACCEPTED FOR PUBLICATION
REFERENCES
[1] F. Li and Y. Wang, “Routing in vehicular ad hoc networks: A survey,”
IEEE Veh. Technol. Mag., vol. 2, no. 2, pp. 12–22, Jun. 2007.
[2] J. Bernsen and D. Manivannan, “Unicast routing protocols for vehicular
ad hoc networks: A critical comparison and classification,” Perv as i ve
and Mobile Computing, vol. 5, no. 1, pp. 1–18, Feb. 2009.
[3] C. Maihofer, “A survey of geocast routing protocols,” IEEE Commun.
Surveys Tutorials, vol. 6, no. 2, pp. 32–42, Second Quarter 2004.
[4] A. Festag, P. Papadimitratos, and T. Tielert, “Design and performance of
secure geocast for vehicular communication,” IEEE Trans. Veh. Technol.,
vol. 5, no. 59, pp. 2456–2471, Jun. 2010.
[5] R. J. Hall, “An improved geocast for mobile ad hoc networks,” IEEE
Trans. Mobile Comput., vol. 10, no. 2, pp. 254–266, Feb. 2011.
[6] L. Junhai, Y. Danxia, X. Liu, and F. Mingyu, “A survey of multicast
routing protocols for mobile ad-hoc networks,” IEEE Commun. Surveys
Tutor ia ls , vol. 11, no. 1, pp. 78–91, First Quarter 2009.
[7] O. Badarneh and M. Kadoch, “Multicast routing protocols in mobile
ad hoc networks: A comparative survey and taxonomy,” EURASIP J.
Wireless Comm. and Networking, vol. 2009, pp. 1–42, Jan. 2009.
[8] A. Sebastian, M. Tang, Y. Feng, and M. Looi, “A multicast routing
scheme for efficient safety message dissemination in VANET,” in
Proc. IEEE Wireless Comm. and Networking Conf. (WCNC), Sydney,
Australia, Apr. 2010, pp. 1–6.
[9] E. K. Lua, J. Crowcroft, M. Pias, R. Sharma, and S. Lim, “A survey and
comparison of peer-to-peer overlay network schemes,” IEEE Commun.
Surveys Tutorials, vol. 7, no. 2, pp. 72–93, Second Quarter 2005.
[10] L. Zhou, Y. Zhang, K. Song, W. Jing, and A. V. Vasilakos, “Distributed
media services in P2P-based vehicular networks,” IEEE Trans. Veh.
Technol., vol. 60, no. 2, pp. 692–703, Feb. 2011.
[11] U. Shevade, Y. Chen, L. Qiu, Y. Zhang, V. Chandar, M. K. Han, H. H.
Song, and Y. Seung, “Enabling high-bandwidth vehicular content distri-
bution,” in Proc. ACM Int’l Conf. on Emerging Networking Experiments
and Technologies (CoNEXT), Philadelphia, PA, Nov. 2010, pp. 1–12.
[12] M. Guo, M. Ammar, and E. W. Zegura, “V3: A vehicle-to-vehicle live
video streaming architecture,” Pervasive and Mobile Computing,vol.1,
no. 4, pp. 404–424, Dec. 2005.
[13] F. Soldo, C. Casetti, C. Chiasserini, and P. Chaparro, “Streaming
media distribution in VANETs,” in Proc. IEEE Global Telecomm. Conf.
(GLOBECOM), New Orleans, LA, Nov. 2008, pp. 1–6.
[14] Y. Chu and N. Huang, “Delivering of live video streaming for vehicular
communication using peer-to-peer approach,” in Proc. IEEE Workshop
on Mobile Networking for Vehicular Environments, Anchorage, AK,
May 2007, pp. 1–6.
[15] S. Ni, Y. Tseng, Y. Chen, and J. Sheu, “The broadcast storm problem in a
mobile ad hoc network,” in Proc. ACM Int’l Conf. on Mobile Computing
and Networking (MOBICOM), Seattle, WA, 1999, pp. 151–162.
[16] G. Korkmaz, E. Ekici, F. ¨
Ozg¨uner, and ¨
U. ¨
Ozg¨uner, “Urban multi-
hop broadcast protocol for inter-vehicle communication systems,” in
Proc. ACM Int’l Workshop on Vehicular Ad Hoc Networks. (VANET),
Philadelphia, PA, Sep. 2004, pp. 76–85.
[17] G. Korkmaz, E. Ekici, and F. ¨
Ozg¨uner, “An efficient fully ad-hoc multi-
hop broadcast protocol for inter-vehicular communication systems,” in
Proc. IEEE Int’l Conf. on Comm. (ICC),Istanbul, Turkey, Jun. 2006,
pp. 423–428.
[18] E. Fasolo, A. Zanella, and M. Zorzi, “An effective broadcast scheme for
alert message propagation in vehicular ad hoc networks,” in Proc. IEEE
Int’l Conf. on Comm. (ICC), Istanbul, Turkey, Jun. 2006, pp. 3960–
3965.
[19] D. Li, H. Huang, X. Li, M. Li, and F. Tang, “A distance-based directional
broadcast protocol for urban vehicular ad hoc network,” in Proc. IEEE
Int’l Conf. on Wireless Comm., Networking and Mobile Computing
(WiCom), Shanghai, China, Sep. 2007, pp. 1520–1523.
[20] N. Wisitpongphan, O. K. Tonguz, J. S. Parikh, P. Mudalige, F. Bai, and
V. Sadekar, “Broadcast storm mitigation techniques in vehicular ad hoc
networks,” IEEE Wireless Commun., vol. 14, no. 6, pp. 84–94, Dec.
2007.
[21] S. Khakbaz and M. Fathy, “A reliable method for disseminating safety
information in vehicular ad hoc networks considering fragmentation
problem,” in Proc. IEEE Int’l Conf. on Wireless and Mobile Communi-
cations (ICWMC), Athens, Greece, Jul. 2008, pp. 25–30.
[22] T. Osafune, L. Lin, and M. Lenardi, “Multi-hop vehicular broadcast
(MHVB),” in Proc. IEEE Int’l Conf. on ITS Telecomm. (ITST), Chengdu,
China, Jun. 2006, pp. 757–760.
[23] M. Taha and Y. Hasan, “VANET-DSRC protocol for reliable broad-
casting of life safety messages,” in Proc. IEEE Int’l Symp. on Signal
Processing and Information Technology, Dec. 2007, pp. 104–109.
[24] Y. Tseng, R. Jan, C. Chen, C. Wang, and H. Li, “A vehicle-density-based
forwarding scheme for emergency message broadcasts in VANETs,” in
Proc. IEEE Int’l Conf. on Mobile Ad Hoc and Sensor Sys. (MASS),
San Francisco, CA, Nov. 2010, pp. 703–708.
[25] R. S. Schwartz, R. Barbosa, N. Meratnia, G. Heijenk, and H.Scholten, “A
simple and robust dissemination protocol for VANETs,” in Proc. IEEE
European Wireless Conf., Lucca, Italy, Apr. 2010, pp. 214–222.
[26] Q. Yang and L. Shen, “A multi-hop broadcast scheme for propagation of
emergency messages in VANET,” in Proc. IEEE Int’l Conf. on Comm.
Technology (ICCT), Nanjing, China, Nov. 2010, pp. 1072–1075.
[27] M. Zorzi and R. Rao, “Geographic random forwarding (GeRaF) for ad
hoc and sensor networks: Energy and latency performance,” IEEE Trans.
Mobile Comput., vol. 2, no. 4, pp. 349–365, Oct.-Dec. 2003.
[28] G. Ciccarese, M. De Blasi, P. Marra, C. Palazzo, and L. Patrono, “On
the use of control packets for intelligent flooding in VANETs,” in
Proc. IEEE Wireless Comm. and Networking Conf. (WCNC), Budapest,
Hungary, Apr. 2009, pp. 1–6.
[29] H. Alshaer and E. Horlait, “An optimized adaptive broadcast scheme
for inter-vehicle communication,” in Proc. IEEE Vehicular Technology
Conf. (VTC), Stockholm, Sweden, May 2005, pp. 2840–2844.
[30] A. Wegener, H. Hellbr¨uck, S. Fischer, C. Schmidt, and S. Fekete, “Au-
toCast: An adaptive data dissemination protocol for traffic information
systems,” in Proc. IEEE Vehicular Technology Conf. (VTC), Baltimore,
MD, Sep. 2007, pp. 1947–1951.
[31] S. Panichpapiboon and G. Ferrari, “Irresponsible forwarding,” in
Proc. IEEE Int’l Conf. on ITS Telecomm. (ITST), Phuket, Thailand, Oct.
2008, pp. 311–316.
[32] M. Slavik and I. Mahgoub, “Stochastic broadcast for VANET,” in Proc.
IEEE Consumers Comm. and Networking Conf. (CCNC),LasVegas,
NV, Jan. 2010, pp. 1–5.
[33] H. Hellbr¨uck and S. Fischer, “MINE and MILE: Improving connectivity
in mobile ad-hoc networks,” ACM Mobile Computing and Comm.
Review, vol. 8, no. 4, pp. 19–36, Oct. 2004.
[34] S. Busanelli, G. Ferrari, and S. Panichpapiboon, “Efficient broadcast-
ing in IEEE 802.11 networks through irresponsible forwarding,” in
Proc. IEEE Global Telecomm. Conf. (GLOBECOM), Honolulu, HI,
Nov. 2009, pp. 1–6.
[35] R. Ahlswede, N. Cai, S.-Y. R. Li, and R. W. Yeung, “Network infor-
mation flow,” IEEE Trans. Inf. Theory, vol. 46, no. 4, pp. 1204–1216,
Jul. 2000.
[36] R. W. Yeung, “Network coding: A historical perspective,” Proc. IEEE,
vol. 99, no. 3, pp. 366–371, Mar. 2011.
[37] S. Katti, H. Rahul, W. Hu, D. Katabi, M. Medrad, and J. Crowcroft,
“XORs in the air: Practical wireless network coding,” IEEE/ACM Trans.
Netw., vol. 16, no. 3, pp. 497–510, Jun. 2008.
[38] L. Haojie, Z. Dongdong, and Y. Chen, “Local-directed network coding
in vehicular ad-hoc networks,” in Proc. IEEE Int’l Symp. on Intelligent
Signal Processing and Comm. Sys. (ISPACS), Chengdu, China, Dec.
2010, pp. 1–4.
[39] L. Li, R. Ramjee, M. Buddhikot, and S. Miller, “Network coding-based
broadcast in mobile ad hoc networks,” in Proc. IEEE Conf. on Computer
Comm. (INFOCOM), Anchorage, AK, May 2007, pp. 1739–1747.
[40] W. Lou and J. Wu, “On reducing broadcast redundancy in ad hoc
wireless networks,” IEEE Trans. Mobile Comput., vol. 1, no. 2, pp.
111–122, Apr.-Jun. 2002.
[41] S. Yang and J. Wu, “Efficient broadcasting using network coding and
directional antennas in MANETs,” IEEE Trans. Parallel Distributed
Sys., vol. 21, no. 2, pp. 148–161, Feb. 2010.
[42] N. Kadi and K. Agha, “MPR-based flooding with distributed fountain
network coding,” in Proc. IFIP Annual Mediterranean Ad Hoc Network-
ing Workshop (Med-Hoc-Net), Juan-les-pins, France, Jun. 2010, pp. 1–5.
[43] A. Qayyum, L. Viennot, and A. Laouiti, “Multipoint relaying for flood-
ing broadcast messages in mobile wireless networks,” in Proc. IEEE
Hawaii Int’l Conf. on Sys. Sciences (HICSS), Big Island, HI, Jan. 2002,
pp. 3866–3875.
[44] T. Zhong, B. Xu, and O. Wolfson, “Disseminating real-time traffic
information in vehicular ad-hoc networks,” in Proc. IEEE Intelligent
Vehicles Symp. (IV), Eindhoven, The Netherlands, Jun. 2008, pp. 1056–
1061.
[45] T. Nadeem, S. Dashtinezhad, C. Liao, and L. Iftode, “TrafficView: A
scalable traffic monitoring system,” in Proc. IEEE Int’l Conf. on Mobile
Data Management (MDM), 2004, pp. 1–14.
[46] T. Fujiki, M. Kirimura, T. Umedu, and T. Higashino, “Efficient acquisi-
tion of local traffic information using inter-vehicle communication with
queries,” in Proc. IEEE Intelligent Transportation Sys. Conf. (ITSC),
Seattle, WA, Sep. 2007, pp. 241–246.
This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

PANICHPAPIBOON and PATTARA-ATIKOM: A REVIEW OF INFORMATION DISSEMINATION PROTOCOLS FOR VEHICULAR AD HOC NETWORKS 15
[47] Q. Yu and G. Heijenk, “Abiding geocast for warning message dissem-
ination in vehicular ad hoc networks,” in Proc. IEEE Int’l Conf. on
Comm. (ICC), Beijing, China, May 2008, pp. 400–404.
[48] L. Wischhof, A. Ebner, and H. Rohling, “Information dissemination in
self-organizing intervehicle networks,” IEEE Trans. Intell. Transp. Syst.,
vol. 6, no. 1, pp. 90–101, Mar. 2005.
[49] The MathWorks, Inc., “MATLAB website,” [Online], 2011, available:
http://www.mathworks.com/.
[50] H. Schwetman, “CSIM user’s guide rev. 1,” Microelectronics and
Computer Technology Corporation, Tech. Rep. ACT-126-90, 1991.
[51] OPNET Technologies, Inc., “OPNET website,” [Online], 2011, avail-
able: http://www.opnet.com/.
[52] “The Network Simulator (NS-2) website,” [Online], 2011, available:
http://nsnam.isi.edu/nsnam/index.php/.
[53] D. R. Choffnes and F. E. Bustamante, “An integrated mobility and traffic
model for vehicular wireless networks,” in Proc. ACM Int’l Workshop on
Vehicular Ad Hoc Networks. (VANET), Cologne, Germany, Sep. 2005,
pp. 69–78.
[54] E. Teramoto, M. Baba, H. Mori, H. Kitaoka, I. Tanahashi, and
Y. Nishimura, “Prediction of traffic conditions for the Nagano olympic
winter games using traffic simulator: NETSTREAM,” in Proc. World
Congress on ITS, 1998, pp. 1801–1806.
[55] “OMNeT++ network simulator framework,” [Online], 2011, available:
http://www.omnetpp.org/.
[56] S. Panichpapiboon and W. Pattara-atikom, “Connectivity requirements
for self-organizing traffic information systems,” IEEE Trans. Veh. Tech-
nol., vol. 57, no. 6, pp. 3333–3340, Nov. 2008.
[57] M. Brackstone and M. McDonald, “Car-following: A historical review,”
Transportation Research Part F, vol. 2, no. 4, pp. 181–196, Dec. 1999.
[58] S. Maerivoet and B. De Moor, “Cellular automata models of road
traffic,” Physics Reports, vol. 419, no. 1, pp. 1–64, Nov. 2005.
[59] R. Mangharam, D. S. Weller, R. Rajkumar, P. Mudalige, and F. Bai,
“GrooveNet: A hybrid simulator for vehicle-to-vehicle networks,” in
Proc. IEEE Int’l Conf. on Mobile and Ubiquitous Sys., San Jose, CA,
Jul. 2006, pp. 1–8.
Sooksan Panichpapiboon (S’05-M’07) received the
B.S., M.S., and Ph.D. degrees in electrical and com-
puter engineering from Carnegie Mellon University,
Pittsburgh, PA, in 2000, 2002, and 2006, respec-
tively. In April 2008, he was a Visiting Researcher
with the Department of Information Engineering,
University of Parma, Italy. He is currently a faculty
member in the Faculty of Information Technology,
King Mongkut’s Institute of Technology Ladkra-
bang, Bangkok, Thailand. He has served as a tech-
nical program committee member for many inter-
national conferences. His current research interests include ad hoc wireless
networks, intelligent transportation systems, radio frequency identification
(RFID) systems, and performance modeling. Dr. Panichpapiboon was the
recipient of the ASEM DUO-Thailand Fellowship in 2007 and the Doctoral
Dissertation Award from the National Research Council of Thailand in 2011.
Wasan Pattara-atikom (M’04) received the B.E.
degree in computer engineering from Khon Kaen
University, Khon Kaen, Thailand. He received the
M.S. degree in telecommunications, the M.B.A.
degree in business administration, and the Ph.D.
degree in information science from the University
of Pittsburgh, Pittsburgh, PA.
He is currently a Senior Researcher with the
Network Technology Laboratory, National Electron-
ics and Computer Technology Center, Pathumthani,
Thailand where he leads a project on vehicular
traffic prediction. His current research interests are vehicular ad hoc network,
intelligent transport system, and machine learning.
This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.