vehicular communication ad hoc routing protocols: a survey
TRANSCRIPT
Author's Accepted Manuscript
Vehicular Communication Ad Hoc RoutingProtocols: A Survey
Baraa T. Sharef, Raed A. Alsaqour, MahamodIsmail
PII: S1084-8045(13)00196-3DOI: http://dx.doi.org/10.1016/j.jnca.2013.09.008Reference: YJNCA1124
To appear in: Journal of Network and Computer Applications
Cite this article as: Baraa T. Sharef, Raed A. Alsaqour, Mahamod Ismail,Vehicular Communication Ad Hoc Routing Protocols: A Survey, Journal ofNetwork and Computer Applications, http://dx.doi.org/10.1016/j.jnca.2013.09.008
This is a PDF file of an unedited manuscript that has been accepted forpublication. As a service to our customers we are providing this early version ofthe manuscript. The manuscript will undergo copyediting, typesetting, andreview of the resulting galley proof before it is published in its final citable form.Please note that during the production process errors may be discovered whichcould affect the content, and all legal disclaimers that apply to the journalpertain.
www.elsevier.com/locate/jnca
Vehicular Communication Ad Hoc Routing Protocols: A Survey
Baraa T. SharefaI, Raed A. Alsaqoura*II, Mahamod IsmailbI
aSchool of Computer Science, Faculty of Information Science and Technology, Universiti Kebangsaan Malaysia, Bangi, Selangor, 43600, Malaysia
bDepartment of Electrical, Electronics and System Engineering, Faculty of Engineering, Universiti Kebangsaan Malaysia, Bangi, Selangor, 43600, Malaysia
*Corresponding author. Tel.: +6012 6897 019.
Abstract
Vehicular communications are now the dominant mode of transferring information between
automobiles. One of the most promising applications of vehicular communications is the vehicular ad
hoc network (VANET), an approach to the intelligent transportation system (ITS). VANET is a subclass
of the mobile ad hoc network, which does not depend on fixed infrastructure, in which the nodes are
highly mobile. Therefore, the network topology changes rapidly. The design of routing protocols in
VANETs is crucial in supporting the ITS. As a prerequisite to communication, the VANET routing
protocols must establish an efficient route between network nodes. Furthermore, they should adjust
efficiently to the quickly varying topology of moving vehicles. In this paper, we discuss the main
characteristics and the research challenge of routing in VANETs, which may be considered in designing
various routing protocols. We also created taxonomy of the current routing protocols for VANETs, and
we surveyed and compared symbolized instances for all the classes of protocols. This organization and
description present the advantages and weaknesses of the current protocols in this field, and paves the
way for solutions to unaddressed problems.
Keywords
Mobile Ad hoc Network; Vehicular Ad hoc Network; Inter-vehicle Communication, Routing Protocols
1. Introduction
Alerting drivers about the conditions of roads, traffic, and related aspects is crucial to safety and to the regulation of vehicle
flow. To achieve this, timely and accurate information is essential. As shown in Figure 1, a vehicular ad hoc network
(VANET) typically addresses this problem [1, 2]. Emergencies can be avoided by taking advantage of the facilities supplied by
VANET technologies. In other words, all information related to traffic mobility on the road, such as traffic density, speed, and
directions of the vehicles as well as the weather, are gathered by using inter-vehicle and vehicle-to-roadside communication
technologies. This information helps organize road traffic and prevent accidents. This information is also useful in roadside
base station assistance to update motorists about the traffic situation and is also used to connect the inter-vehicle network to an
outside network in a manner that integrates heterogeneous emerging wireless technologies such as 3G cellular systems, long-
term evolution (LTE), LTE- Advance, IEEE 802.11, and IEEE 802.16e [3, 4].
VANET is a new technology that integrates the potentials of new-generation wireless networks into vehicles. VANET aims
to offer (i) continuous connectivity for mobile users while they are on the road, which enables them to link with other users
through the latter’s home- or office-based networks, and (ii) efficient wireless connection between vehicles without access to
any fixed infrastructure, which enables the ITS. Consequently, VANET is also known as inter-vehicle communication
(IVC).VANET devices such as on-board units are fixed in vehicles and function as the nodes to transmit and receive messages
through wireless networks. These devices provide drivers and passengers with the latest information on accidents, flooding,
rain, traffic jams, and any disturbances. By obtaining such information on time, drivers can make appropriate decisions and
avoid mishaps.
Figure 1 Vehicular ad hoc network
The features of VANET are typically similar to the operation technology of a mobile ad hoc network (MANET) in the
sense that the self-organization, self-management, low bandwidth, and shared radio transmission conditions remain the same.
However, the key operational impediment of VANET arises from the high speed and tentative mobility (in contrast to the
MANET) of the mobile nodes (vehicles) along the paths. This fact indicates that the competent design of routing protocol
requires improving the MANET architecture to efficiently accommodate the fast mobility of the VANET nodes. This issue has
introduced numerous research challenges to the design of a suitable routing protocol.
This paper focuses on a major issue in networking: the routing protocol for VANETs. The foremost objective of routing
protocols is to attain short communication time while using the least amount of network resources. Many routing protocols
have been designed for MANETs, of which few can be directly implemented to VANETs. Nevertheless, the outcomes of the
simulation show that the performance of VANETs is affected by the following factors: fast-moving vehicles, active
information transfer, and the associated high speed of mobile nodes, which differ from those of MANETs. Consequently,
identifying and administering routes is a demanding task for VANETs. This fact has guaranteed a variety of research
challenges to the design of a suitable routing protocol.
The main contribution of this study is a comprehensive survey of the topical research development of routing protocols in
VANET communication. In addition, this study categorizes routine protocols according to the architecture of VANETs and
presents a new taxonomy of VANET routing protocols. We discuss the routing strategies, strengths, and limitations of routing
protocols in each category. A qualitative comparison of the protocols revealed that position-based routing and geo-casting are
more capable than other routing protocols for VANETs because of environmental limitations. Moreover, the infrastructure-
based routing protocols are the most promising in VANET communication. At the end of this study, we identify the possible
directions for this evolving field.
The remainder of this paper is organized as follows. Section 2 reviews the architecture, applications, and unique challenges
faced by VANET technology and compares MANET and VANET in detail. An overview of VANET routing protocols is
presented in Section 3. Section 4 presents the most popular mobility models in VANET. In Section 5, we review and compare
the VANET routing protocols. Section 3 concludes this paper and provides a number of recommendations.
2. Overview of VANET technology
2.1 VANET architectures
As mentioned, VANETs have standards similar to those of MANET in the sense that neither depend on a permanent base
for communication and broadcasting information. VANET pertains to the extremely active milieu of road transportation.
Figure 2 illustrates the pure cellular/wireless local area network (WLAN), pure ad hoc, and hybrid architectures of VANETs.
In the pure cellular architecture (Figure 2a), VANETs might employ permanent cellular gateways and WLAN access points or
base stations at traffic junctions for Internet access, gathering traffic information, or routing. In these conditions, the network
architecture will be either cellular or WLAN. This type of VANET architecture is called vehicle-to-infrastructure (V2I)
communication and effectively integrates heterogeneous emerging wireless technologies such as 3G cellular systems, LTE,
LTE-Advance, IEEE 802.11, and IEEE 802.16e [3, 4].
Figure 2b illustrates the pure ad hoc architecture of VANET, which is also called vehicle-to-vehicle (V2V) communication.
In this architecture, the nodes might be forced to engage with each other, as economic constraints limit the construction of
cellular towers and wireless access points. The information gathered from the sensors fixed in vehicles will be extremely useful
in alerting other vehicles about accidents or other emergencies and will also assist the police in tracing criminals [5]. The
infrastructure-less network design is in the overall ad hoc group in which the nodes perform V2V communication.
Figure 2c illustrates the hybrid architecture (V2I and V2V) of VANET. In the hybrid architecture, the wireless networking
devices are fixed in roadside communication units such as cellular towers and access points and vehicles to facilitate
communication among them. A range of applications in metropolitan screening, security, driving support, and entertainment[6]
have employed infrastructure communication units to access active and affluent information beyond their network framework
and transmit this information through a peer-to-peer ad hoc, infrastructure-less communication. The hybrid architecture of
cellular/WLAN and ad hoc offers richer contents and superior flexibility in content sharing.
Figure 2 VANET network architectures: (a) pure cellular (V2I), (b) pure ad hoc (V2V), (c) hybrid (V2I & V2V)
2.2 VANET applications
VANET facilitates communications among nearby vehicles and between the vehicles and nearby fixed devices. All types of
vehicles benefit from VANET. Roadside devices are generally maintained by government agencies, but the operations are
privatized in some countries. The different types of VANET applications [2, 7-9] are categorized as follows:
• Safety applications. The safety applications enhance the protection of passengers by sending and receiving information
pertinent to vehicle safety. Generally, these alerts, such as co-operative collision warning, lane change warning,
emergency video streaming, and incident management, are directly sent to the drivers or are received by the automatic
active safety system.
• Comfort applications. Comfort applications are the VANET applications associated with the comfort level of the
traveler, such as IVC, electronic toll collection, parking lot payment, and traffic management. These applications are
expected to soon become extremely popular.
All VANET applications have a common set of requirements. The most common requirements are as follows: 10 m to
1,000 m coverage, a maximum vehicle speed of500 km/h, and latency in the range of 50 ms to 500 ms. Generally, safety
applications must not stop for more than 200 ms. Furthermore, the compactness of the network might be in the range of small
groups of 2 to 20 vehicles to traffic jams with up to thousands of vehicles per radio cell.
2.3 Unique VANET challenges
VANETs encompass vehicles with radio range, which function as portable nodes and routers for other nodes. Apart from
the resemblances of ad hoc networks, such as short radio transmission range, self-organization, self-management, and low
bandwidth, VANET has unique challenges that influence the design of the communication system and its routing protocols [9-
13]. These challenges include the following:
• Highly dynamic topology. The topologies of VANETs will not be constant, and they always vary based on vehicle
speed. If two vehicles each have a wireless transmission range of 250 m, they can be connected only as long as they are
250 m apart. Furthermore, the link can be sustained for only 10 s if the two cars travel in opposite directions at 25 m/s.
• Frequently disconnected network. VANETs will not have constant connectivity because of the high-speed movement
between vehicles. In low-density vehicles, the link is highly likely to be disconnected. In some applications that require
ubiquitous Internet access, the issue must be addressed. A probable solution is to pre-install a large number of relay nodes
or access points on the roadside to maintain connectivity.
• Mobility modeling and predication. Mobility modeling and predication play a vital part in the network protocol design
for VANETs based on their highly mobile node movement and dynamic topology [14]. Furthermore, the vehicular nodes
are generally limited by pre-built highways, roads, and streets; thus, the future position of the vehicle can be predicted
based on the speed and the street map.
• Various communication environments. VANETs are employed in two distinctive communication backgrounds
represented by(city environment and highway environment) [15]. Traffic conditions are very simple in highways but very
complex in cities and towns. The obstruction caused by buildings, trees and other obstacles in cities may occasionally
prevent a direct line of communication in the direction of projected data communication.
• Hard delay constraints. In a number of VANET applications, the network does not require high data rates but is limited
by hard delay. For instance, in an automatic highway system, when a vehicle brakes, a corresponding message must be
transmitted to other vehicles to avoid collision. In these applications, the maximum delay is more vital than the average
delay.
• Interaction with on-board sensors. Presumably, the nodes have on-board sensors that provide information to be used to
generate communication links and facilitate routing. Nowadays, global positioning system (GPS) devices are widely used
in vehicles, which offer information about location for routing purposes.
• Infrastructure access. Communication infrastructure along the road, such as roadside units (RSUs) and public hotspots,
allows access to network servers, typically in the Internet. Given that roadside units and public hotspots do not provide
full wireless coverage, it is expected that for security mechanisms, such as for the management and distribution of
cryptographic keys via a centralized architecture, infrastructure is not available all the time.
• High application requirements on data delivery. VANET applications are necessary in preventing road accidents and
ensuring safety. These applications have high requirements with respect to real-time functionality and reliability. An end-
to-end delay of seconds can render safety information meaningless. Loss of messages, e.g., due to security attacks, may
endanger human life. These applications are typically based on a broadcast distribution of data (geographically scoped
flooding) where the destination nodes are those located in a geographic area.
• No confidentiality of safety information. For safety applications, the information contained in a message is of interest to
all road users and is therefore not confidential.
2.4 Comparison among MANET and VANET technologies
MANETs and VANETs are extremely similar in various technical aspects. Their differences can also be observed based on
the parameters presented in Table 1.
Table 1 Comparison between MANET and VANET
Parameters MANET VANET
Cost of production Cheap Expensive
Change in network topology Slow Frequent and very fast
Frequently topology change Low High
Node density Sparse Dense and frequently variable
Bandwidth Hundred kpbs Thousand kpbs
Node lifetime Depends on power resource Depends on lifetime of vehicle
Multi-hop routing Available Weakly available
Reliability Medium High
Moving pattern of nodes Random Regular
Addressing scheme Attribute-based Location-based
3. Overview of VANET routing protocols
Given diverse architectures, applications, and challenges, researchers have proposed a wide range of routing protocols for
VANETs. These protocols all primarily aim to maximize throughput while minimizing packet loss and controlling overhead.
One of the major technological challenges faced by VANETs is developing an efficient routing protocol for a highly
changeable topology. VANET require new types of routing protocols. As opposed to wired infrastructure, no dedicated router
nodes are used, and routing protocols have to be performed by the user nodes (vehicles), which may be mobile and unreliable.
com
(V2
3.1
div
3.1
the
As shown i
mmunication
2I) routing pro
1 Vehicle-to-v
V2V protoco
vided into five
(1) Topolog
(2) Position
(3) Cluster-
(4) Geo-cas
(5) Multica
(6) Broadca
1.1 Topolog
This class of
e protocols are
(1) Proactiv
(2) Reactiv
(3) Hybrid
VANET Routing Protocols
n Figure 3,
into two main
otocols.
vehicle-based
ols perform v
e groups:
gy-based (ad h
n-based routin
-based routing
st-based routin
ast-based routi
ast-based rout
gy-based (ad
f routing proto
e categorized a
ve (table-drive
ve (on-demand
routing.
Vehicle‐to‐vehicle(V2V)
Vehicle‐to‐infrastructure
(V2I)
we classified
n categories: v
Figu
(V2V) routin
vehicle-to-veh
hoc) routing p
ng protocols,
g protocols,
ng protocols
ing protocols,
ting protocols
hoc) routing p
ocols employ
as:
en) routing,
d) routing, and
topology‐based
Position‐based
Cluster‐based
Geo cast‐based
Multicast ‐based
Broadcast‐based
Static RSUs
Mobile infrastructure
d the current
vehicle-to-veh
ure 3 Taxonom
ng protocols
hicle commun
protocols,
, and
.
protocols
s the link info
d
Proactive
Reactive
Hybrid
Non‐Delay Tolerant
Be
Delay Tolerant
Hybrid
Tree‐based
Mesh‐based
t VANET rou
hicle-based (V
my of VANET
nication but d
ormation that
Beacon GPS
eaconless
uting protoco
V2V) routing p
T routing proto
do not focus o
exists in the
DSDV,
TORA, A
SR, GPSR+AGF, GSR, SALOUVRE,
CBR, CBDRP, LOR
IVG, CGR, AG
MAO
BROADCOMM, UMB,
SADV, RAR
MIB
ols based on
protocols and v
ocols
on fixed infra
network to ex
GSRP, FSR, OLSR, WR
AODV, PRAODV, DSR, A
ZRP, HARP
AR, A‐STAR, STAR, MU DIR, ROMSGP,JARR, E
CB
SKVR, VADD, GeOpp
GeoDTN+Nav
RA‐CBF, COIN, TIBCRPH
R, ROVER, Mobicast
DV, ADMR, MAV‐AOD
ODMRP, D‐ODMR
DV‐CAST, EAEP, HyDi,
R, VAGP, IAGR, MOVE
BR, MGRP, PBR
the architec
vehicle-to-inf
astructure on
xecute packet
RP,TBRPF
AODV+PGB
RU, PDGR, GPCR, GPGEBGR, B‐MFR, AMAR,
BF
ps
H
DV, MOLSR
RO
, DECA
ture of VAN
frastructure-ba
roads. It can
t forwarding,
GR, PBR‐DV, CAR, GyTATO‐GO.
NET
ased
n be
and
AR,
3.1.1.1 Proactive (table-driven) routing protocols
This type mainly depends on the shortest path algorithms[16]. The information of all associated nodes is stored in the form
of tables, as these protocols are table-based. These tables are also distributed with their neighbors, and the nodes renew their
routing tables when the network topology changes.
The proactive protocols do not have initial route discovery delay but consumes significant bandwidth for periodic topology
updates. The strategies implemented in proactive algorithms are distance-vector routing such as DSDV [17] and link-state
routing such as OLSR [18].
Proactive routing protocols may be unsuitable for high mobility nodes because distance vector routing requires
considerable bandwidth to share routing information with neighbours. Furthermore, the table is large in the case of large
networks, and considerable memory and processing may be required in case of link state routing.
As in VANET, nodes (vehicles) have high mobility and move at a high speed. The proactive routing protocol is unsuitable
for this. This routing protocol category may fail in VANET because it requires more bandwidth and large table information.
The most popular proactive routing protocols are DSDV, GSRP, FSR, OLSR, WRP, and TBRPF [17] [19] [20] [18] [21] [22].
1. Destination-Sequenced Distance Vector (DSDV) [17]. Perkins and Bhagwat in 1994 proposed the DSDV routing protocol,
which depends on the Bellman–Ford algorithm [23]. This protocol removes the looping, increases the convergence speed,
and minimizes overhead of the control message. In the DSDV, all the nodes sustain a next-hop information table and are
exchanged table’s information with their neighbors. The DSDV provides a loop-free single path to the destination and
sends two types of packets: full dump and incremental. In full dump packets, all the routing information is sent, whereas in
the incremental type, only updates are sent. This function decreases bandwidth utilization by sending only updates instead
of complete routing information. The incremental packet still increases the overhead in the network because the packets are
so frequent and are therefore unsuitable for large-scale networks.
2. Global State Routing protocol (GSRP) [19]. GSRP is similar to DSDV. It appliesthe concept of link state routing but
enhances it by preventing flooding of routing messages. In this protocol, all the nodes have a neighbor list, a topology table,
a next-hop table, and a distance table. The neighbor list of a node holds the records of its neighbors (nodes that can be
listened to by a node are supposed to be its neighbors). In case of every single destination node, the topology table encloses
the link state information, as documented by the location and the timestamp of the information. In case of every single
destination, the subsequent hop table includes the next hop where the packets for this destination have to be delivered. The
distance table includes the shortest distance to each destination node. The routing messages are produced on an exchanged
link as in link state protocols. Upon receiving a routing message, the node upgrades its topology table if the sequence
number of the message is newer than the sequence number saved in the table. Later, the node rebuilds its routing table, and
broadcasts the information to its neighbors.
3. Fisheye State Routing (FSR) [20]. FSR is an enhanced version of GSRP. In GSRP, a substantial amount of network
bandwidth is wasted by the large size of updated messages (flooding). In FSR, the updated messages do not hold
information about all nodes. Rather, FSR swaps information about closer nodes more frequently as against the farther
nodes, thereby minimizing the size of the updated message. Therefore, every single node receives appropriate information
about neighbors, and the details and precision of the information are reduced as the range from the node increases.
Figure 4 illustrates the scope of the fisheye for the center node C. The scope is outlined concerning the nodes that can be
arrived at in a particular number of hops. The center node has the most precise information about the nodes in the white
circle and all others. Although a node does not have appropriate information about distant nodes, the packets are properly
routed because the route information becomes increasingly appropriate as the packet moves closer to the destination. FSR
effectively scales huge networks as the overhead is controlled in this scheme.
Figure 4 FSR scope
4. Optimized Link State Routing (OLSR) [18]. OLSR consists of well-known unicast routing protocols for MANETs, which
have been successfully adapted to VANETs. In OLSR, all the nodes in the network choose a pair of neighbor nodes known
as multi-point relays (MPR), which retransmit its packets. The neighbor nodes not found in the MPR set can read and
process the packet. This process minimizes the volume of retransmissions in a broadcasting method. The main advantage of
this process is that each node always has a route to every other node in the network. This advantage is a result of a large
message overhead for maintaining the routes.
5. Wireless Routing Protocol (WRP) [21]. WRP is a table driven-based distance-vector routing protocol intended to manage
the routing information through all the nodes in the network. Every node in the network is responsible for managing the
distance table, routing table, link-cost table, and message retransmission table.
We assume that x, y, and z are nodes in the network. The contents of node x in the distance table are: (1) the distance of
each destination node y through each neighbor z of x, and (2) the downstream neighbors of node z through which this path
is investigated. The node x contents in the routing table are: (1) the distance of each destination node y from node x; (2) the
node x predecessor and successor on this path, which it used to determine loops and avoid counting to infinite problems;
and (3) a tag to identify whether the entry is a simple path, loop path, or invalid path. The link-cost table contains: (1) the
cost of link to each neighbor of the node and the number of timeouts since an error-free message was received from that
neighbor. The message retransmission table contains information that tells a node if its neighbor has not acknowledged its
updated message and to resend an update to that neighbor.
In WRP, nodes provide information on the availability of their neighbors from the acknowledgments received from the
neighbor nodes. If a node is not sending packets, it must send a Hello beacon within a specified period to ensure that
connectivity information is properly reflected. Otherwise, the lack of messages from the node can indicate the failure of that
wireless link and may cause a false alarm. When a mobile receives a Hello beacon from a new node, that new node
information is added to the mobile’s routing table, and the mobile sends the new node a copy of its routing table
information.
C
A unique feature of the WRP protocol is that it checks the consistency of all its neighbors every time it detects a change
in the link of any of its neighbors. A consistency check in this manner helps eliminate looping situations in a better way and
also has fast convergence.
6. Topology Dissemination Based on Reverse-Path Forwarding (TBRPF) Routing [22]. TBRPF is a link-state routing
protocol intended for ad hoc networks. All the nodes assemble a source tree based on partial topology information, which
comprises paths to all the accessible nodes. It employs a modification of Dijkstra’s algorithm and the topology information
store in its topology table. The nodes are regularly reorganized with only the distinctions between the preceding and present
network condition by employing Hello beacons. Consequently, the routing messages are smaller, which means they can be
sent to neighbors more frequently.
3.1.1.2 Reactive protocols (on-demand)
Reactive protocols are known as on-demand routing protocols because they regularly renew the routing table. However,
these protocols use a flooding method for route discovery that initiates more routing overhead and also suffer from the initial
route discovery process. Thus, they become unsuitable for security applications in VANET. These types of routing protocols
continuously update their routing information and the carried knowledge of each neighboring node. Therefore, this type of
reactive routing can be adopted in highly mobile ad hoc networks such as VANET. The different kinds of reactive routing
protocols are TORA, AODV, PROAODV, DSR and AODV+PGB [24] [25] [5] [26] [27].
1. Temporally Ordered Routing Algorithm (TORA) [24]. Based on the concept of link reversal routing algorithms, Park and
Corson in 1997 proposed an adaptive and scalable routing protocol called TORA. In highly dynamic roving, this routing
protocol finds multiple routes from the source to the destination. The basic functions of this routing protocol are route
creation, route maintenance, and route erasure. Based on the height of the tree rooted at the source, the nodes use a “height”
metric to build a directed cyclic graph (DAG) during the route creation and route maintenance stages. The DAG directs the
packet flow and ensures reachability to all nodes. The link, based on the proportional height metric of the neighboring
node, can be either upstream or downstream.
In the TORA protocol, all the nodes construct a DAG by sending a query packet. Upon receiving a query packet, and if
the node has a downstream link to the destination, it sends a reply packet or else the pocket will be dropped. Furthermore,
upon receiving a reply packet, a node will revise its height as long as the height of the packet is less than that of other reply
packets.
Each node in the TORA has the following associated metrics:
• The logical time of a link failure,
• The unique ID of a node that represents the new reference level,
• A reflection indicator bit,
• A propagation ordering parameter, and
• The unique ID of the node.
The first three metrics jointly define the reference level. A new reference level is represented each time a node loses its
last downstream link because of a link failure. The last two metrics represent a delta with regard to the reference level. The
flow diagram in Figure 5 explains all possible route maintenance paths in the TORA.
The main advantage of this protocol is that it offers a route to all the nodes in the network, but preserving all these
routes in VANET is difficult because of the dynamic VANET topology.
Figure 5 Flow diagram of route maintenance in TORA [24]
2. Ad hoc On-Demand Distance Vector (AODV) [25]. AODV is the refinement of the DSDV protocol. AODV is different
from DSDV in the sense that it reduces the number of broadcast massages by creating routes on demand. However, DSDV
maintains all the listed routes in the routing table.
In AODV, the source node initiates the routing protocol by using Hello beacons to detect its neighbors. To find a path to
the destination, the source node broadcasts a route request packet (RREQ), and then its neighbors broadcast the RREQ to
their neighbors until it reaches an intermediate node with a route to the destination or until it reaches the destination (Figure
6a). The RREQ packet carries the IP address of the source node, current sequence number, IP address of the destination
node, and the last known sequence number.
After receiving an RREQ packet, the nodes register the address of the node that sends the query in their routing table.
The method of registering its preceding hop is known as backward learning. After reaching the destination node, a route
reply packet (RREP) is transmitted through the total path acquired from backward learning to the source (Figure 6b).
When the source node moves, the route to the destination is established. If one of the intermediate nodes moves, then
the moved node’s neighbor realizes the link failure and sends a link failure notification to its upstream neighbors and so on
until it reaches the source. The source can then reinitiate route discovery if necessary.
This protocol tends to exhaust extra bandwidth because of periodic beaconing. Moreover, heavy control overhead
occurs if a single RREQ packet has multiple RREP packets.
Figure 6 AODV routing protocol: (a) propagation of RREQ and (b) RREP’s path to the source
3. Prediction-based AODV (PRAODV) [5]. PRAODV is a prediction-based routing protocol proposed by Namboodiri et al.
in 2004. It retains most of the features of the AODV protocol. The main modification is in the RREP sent from the
destination or intermediate nodes to the source. The information related to the velocity and location of the packet is
included whenever a node sends an RREP. Every single consecutive node that receives this RREP en route to the source of
the request makes a link lifetime prediction based on its own location and velocity as well as on the values inside the reply
packet from the node that sent it. It contributes its predicted link value to the RREP by swapping the old predicted value if
its approximation of the life span of the link is lower than any previous predictions of any link of the route. It also changes
the location and velocity information of the earlier node with its individual values prior to sending it to the source. The idea
is to have an approximated value, which is the lowest of all links along the route. This value is the predicted lifetime of the
route. A new request is merely delivered prior to the expiry of this predicted lifetime to create a new route to the
destination. However, the concept is intended to prevent waiting until the packets are decreased due to route failure and to
attempt to build a new alternative route where probable. This is the only difference between AODV and PRAODV, which
both use the minimum hop count as the measure to choose between multiple paths to the same destination.
4. Dynamic Source Routing (DSR) [26]. DSR was proposed by Johnson et al. in 1996 and is one of the most well-known
routing protocols. It is a source-routed protocol, which is an on-demand algorithm. Source routing means that the header of
each packet carries the complete sequence list of nodes through which the packet should transmit. This routing protocol
consists of two basic phases: route discovery and route maintenance.
(1) Route discovery is the mechanism by which a source node intends to transmit a packet to a destination node and
produces a source route to the destination. Route discovery is used only when the source tries to transmit a packet to the
destination and does not have a prior known route to the destination.
(2) Route maintenance is the mechanism by which a source node can detect the broken route between the source and the
destination. When route maintenance indicates that a source route is broken, the source can attempt to use any other
known route to the destination, or it can invoke route discovery again to find a new route. Route maintenance is used
only when the source is actually sending packets to the destination.
DSR employs source routing rather than relying on transitional node routing table. Therefore, the routing overhead
always relies on the extent of the path. However, this protocol is limited by the inability of the route maintenance process to
mend a broken link locally. The increasing mobility adversely affects the performance of this protocol.
5. Preferred Group Broadcasting (AODV+PGB) [27]. This is a broadcasting mechanism intended to decrease the broadcast
overhead connected with AODV’s route detection and to present the route stability, which is significant for VANETs,
where rapidly traveling vehicles are utilized as wireless hosts. Depending on the signal received from the broadcast, the
receivers can determine whether the nodes are in the ideal collection and which one in the group should be broadcast. Only
one node is endorsed for broadcast and the ideal group is not necessarily the one that generates significant progress toward
the destination. Therefore, the path detection may be more time consuming than before. Another disadvantage is that the
broadcast may cease if the group becomes empty. Packet replication can ensue as the two nodes in the ideal group can be
simultaneously transmitted. According to [27], broadcast replication can be addressed by including the antecedents in the
packet, which generates similar overhead in the packet as DSR.
3.1.1.3 Hybrid protocols
Hybrid routing protocols are a combination of the reactive and proactive routing protocols to make routing more scalable
and efficient. Most hybrid routing protocols are zone-based, which means that the number of nodes is divided into different
zones to make route discovery and maintenance more reliable.
The overall characteristic of hybrid protocols is that they reduce the network overhead caused by proactive and reactive
routing, handles the network delay caused by reactive routing protocols, and performs route discovery more efficiently.
The drawback of these protocols is that they are not designed for environments characterized by highly dynamic nodes
behavior and rapidly changing topology, such as VANET. In other words, these routing protocols are specifically designed for
networks where nodes are not highly mobile, and the network size depends on a limited number of nodes. The best known
hybrid routing protocols are ZRP and HARP [28, 29].
1. Zone Routing Protocol (ZRP) [28]. In ZRP, the network is divided into intersecting zones. The zone is referred to as a
group of nodes, which are in the radius of the zone. The dimension of a zone is determined by a radius of length α, where α
is the volume of hops to the boundary of the zone. In ZRP, an intra-zone proactive routing protocol (IARP) is employed in
intra-zone interaction and an inner-zone reactive routing protocol (IERP) is employed in intra-zone interaction. The source
sends data straight to the destination if both are in a similar routing zone, or else the IERP responsively instigates path
detection.
2. Hybrid Ad Hoc Routing Protocol (HARP) [29]. HARP breaks down the complete network into non-overlapping zones and
aims to establish a steady route from a source to a destination to improve the delay. Moreover, it implements route
discovery among the zones to confine the run over in the network and selects the best route depending on the constancy
features. In HARP, the routing is carried out on two levels, namely, intra-zone and inter-zone, based on the location of the
destination. It correspondingly employs proactive and reactive protocols in intra-zone and inter-zone routing.
3.1.2 Position-based routing protocols
In position-based routing protocols, all nodes recognize their own locations and their neighbor node geographic locations
through position-pointing devices such as GPS. It does not manage any routing table or exchange any information related to
the link state with the neighbor nodes. The information from the GPS device is used in making routing decisions. This type of
routing performs better as creating and maintaining a global route from the source node to the destination node are not
necessary. The position-based routing protocols can be classified as non-delay tolerant network (non-DTN) routing protocols,
delay tolerant network (DTN) routing protocols, and hybrid routing protocols.
3.1.2.1 Non-delay Tolerant Networks (non-DTNs) routing protocols
The non-DTN position routing protocols do not utilize alternating connectivity and are only realistic in efficiently
populated VANETs. These protocols aim to transmit data packets to the destination as soon as possible. The basic outlook in
the greedy approach of non-DTN routing protocols is that a node advances its packet to its neighbor, which is close to the
destination. However, the forwarding strategy might be unsuccessful if the neighbors are not nearer to the destination than the
node. Therefore, we can claim that the packet has attained the local maximum at the node, as it has achieved the utmost local
growth at the present node. The routing protocols in this group have their individual recovery approach to tackle such failures.
The non-DTN routing protocols can be classified as beacon, beaconless, and hybrid protocols.
1. Beacon protocols
Beacon protocols refer to a cyclical broadcast of short Hello beacons [30]. Beacon pinpoints the existence and location of
a node. The access will be isolated from the receiving node’s neighbor table if it did not obtain a beacon after a specific period
from the equivalent node. The following is an overview of all beacon routing protocols in VANET.
1.1 Greedy Perimeter Stateless Routing (GPSR) [31]. GPSR is a position-based routing protocol aimed at handling mobile
environments. GPSR is most suitable for highways, where the nodes are uniformly distributed. The routing approach of the
protocol depends on two modes[14]:
• Greedy mode where the node forwards a packet to a direct neighbor that is closer to the destination node by using
information about the position of immediate neighbors in the network topology as shown in Figure 7a, where node x
wants to send a data packet destined for node D; x sends the data packet to node y, which is listed in x’s list of
neighbors as shown in Table 2 and is closer to D than any of x’s other neighbors. This greedy forwarding process is
repeated by nodes y, k, z, and w until the data packet reaches the destination node D.
• Perimeter mode is a protocol recovery mode used when a packet reaches a local maximum. Thus, a recovery mode is
used to forward a packet to a node that is nearer to the destination than the node where the packet faces the local
maximum. A simple example of such a topology is shown in Figure 7b where node x is closer to the destination node
D than its neighbors w and y. Although two paths to D, x-y-z-D and x-w-v-D exist, x will not choose to forward the
data packet to w or y by using greedy forwarding. In this case, the GPSR protocol declares x as the local maximum to
D and the shaded region without nodes as a void region. To route the data packet around the void region, the
perimeter forwarding strategy constructs a planarized graph b for the neighbors of node x and routes the data packet
around the void region by using the right-hand rule. The right-hand rule states that when arriving at node x from y, the
next traversed edge is the one that is sequentially counterclockwise about x from edge (x, y). By applying the right-
hand rule shown in Figure 7b, node x forwards the data packet to hop w.
Table 2 Node x neighbors’ list
Node-id Neighbor (x, y coordinates)
A A(x1, y1)
B B(x2, y2)
C C(x3, y3)
Y Y(x4, y4)
F F(x5, y5)
Figure 7 GPSR protocol: (a) Greedy forwarding strategy and (b) Perimeter forwarding strategy [14]
GPSR does not suit the urban environment because of two reasons. First, the greedy forwarding fails if impediments are
present because of the lack of direct communication between nodes. Second, the GPSR toggles to face routing (recovery
mode) if the greedy forwarding strategy fails to find a neighbor closer to the destination than itself. The face routing uses an
extended path to reach the destination, which results in more delays.
1.2 Advanced Greedy Forwarding (GPSR+AGF) [27]. Naumov et al. in 2006 observed two issues with GPSR in VANETs.
The first issue, due to the mobility of VANETs, is that a neighbor table of a node frequently comprises obsolete information
about the position of neighbors. The issue will be resolved by escalating the incidence of the beacons’ frequently, but such a
solution only increases overcrowding and results in potential collisions. The second issue is that the position of the destination
in the routing packet is never updated because of the mobility of the destination while the packet is routed from the
source.When a packet is sent from a source node, the location of the destination node is written into the packet header. This
information is never updated by intermediate nodes while the packet travels toward the destination. In a highly mobile
environment such as a highway, the destination node may travel a substantial distance in a short period.
To resolve these two issues, Naumov et al. proposed the Advanced Greedy Forwarding (AGF) protocol, which integrates
the direction and the velocity of a node in the beacon packet and the whole travel time, along with the time to process the
packet, which is equal to the present forwarding node within the data packet. Based on the velocity vector, speed, and
direction, all nodes can filter the out-of-date nodes in its neighbor table. With the entire travel time, each forwarding node can
better establish the divergence of the destination’s original position and approximate its current location. The results
demonstrated a minimum of triple improvement in the packet delivery ratio to GPSR.
1.3 Geographic Source Routing (GSR) [32]. The GSR protocol merges the position-based routing with topological
information and is aimed at routing in urban surroundings. To obtain the information of the destination node’s location, the
protocol employs the reactive location service (RLS) [33], a direct translation of the route discovery procedure used in reactive
non-position-based ad hoc routing protocols to the position discovery of position-based routing. Essentially, the querying node
floods the network with a “position request” for a specific node identifier. When the node that corresponds to the requested
identifier receives the position request, it sends a “position reply” back to the querying node.
By employing digital maps, the source S understands the locality of all the junctions from the source to destination D. It
uses the Dijkstra shortest path algorithm [34] to compute the shortest path from source to destination. Junctions known as GSR
anchors are included in the shortest path through which the packet has to pass to reach the destination. Moreover, all data
packets in GSR are tagged with the location of the S, D, and GSR anchors; as a result, the GSR does not recognize the existence
of adequate transports on the roads to provide connectivity. Consequently, the GSR does not reflect traffic intensity on the
lanes before choosing the path from source to destination.
The packet delivery ratio of the GSR protocol’s simulation is better than that of the AODV and DSR protocols in [32]. The
drawback of this routing protocol is that it ignores situations such as a sparse network where the number of nodes for
forwarding packets is insufficient. This routing protocol also suffers from a high routing overhead because it uses Hello
beacons frequently as control messages.
1.4 Spatially Aware Packet Routing (SAR) [35]. Tian et al. in 2003 proposed the SAR protocol to address the disadvantages of
the GPSR and GSR with a recovery procedure to avoid a local maximum. As mentioned, the greedy forwarding function in the
GPSR fails to overcome impediments due to the lack of direct communication between nodes. In such a case, GSR ignores
situations such as the sparse network with insufficient nodes for forwarding packets. Conversely, SAR suggests finding an
alternative path from the current location where the local maximum occurs and then replaces the original route with the new
one.
The concept of SAR is illustrated in Figure 8a. Source node S intends to transmit a packet to destination node D. By using
location forwarding, node S forwards the packet to its neighbor A, which is closer to the destination than B. This strategy seems
to be an optimal local decision if the spatial environment is not considered. However, as Figure 8b shows, the node distribution
is strictly bounded to the underlying road structure. Since node A is located on the left road segment, the packet will be
greedily forwarded for probably a large number of hops (so long a neighbor is closer to the destination) before a greedy failure
is recognized and eventually recovered. If the only path to the destination is on the road segment to the right, the packet has to
be forwarded back and goes through node B. Given that a greedy failure will not be memorized in stateless routing such as
GPSR, the forwarding of each subsequent packet may fail in the same way and may have to be recovered each time. However,
with the spatial awareness model, source node S can avoid the forwarding failure in this situation by forwarding packets to the
more suitable neighboring node B instead of A.
Therefore, in SAR, the source node can map itself and the destination node into the spatial model, which is internally
represented as a graph G(E, V) which consists of a set of vertices that refer to significant places together with a set E of edges
that denote the interconnections between places. Thus, nodes that move from one place to another can be considered as moving
from one vertex to another along edges in the graph model, and then the source node calculates the shortest path to the
destination with a shortest path algorithm such as the Dijkstra algorithm. The source then sets the GSR to the shortest path,
which consists of a list of intermediate vertices. The GSR will be embedded into the header of all data packets sent by the
source node. After a vertex in the GSR is reached (i.e., the forwarding node finds the vertex to be located within its radio
range), this vertex is removed from the GSR and the packet is forwarded to the next vertex of the GSR. With this approach, a
packet moves successively closer to the destination along the GSR from one vertex to the next.
Figure 8 Concepts of SAR [35]: (a) flat forwarding and (b) road forwarding
1.5 Anchor-based Street and Traffic Aware Routing (A-STAR) [36]. A-STAR is a position-based routing protocol
exclusively designed for IVC in a city environment. A street map is used by A-STAR to calculate the series of junctions
(anchors) through which a packet has to pass to arrive at its destination. In contrast to GSR, A-STAR calculates the anchor
paths with traffic awareness. A-STAR varies from GSR and GPSR in two major elements. Initially, it integrates traffic
awareness by employing statistically rated maps (keeping track of the number of city bus routes on each street to determine the
anchor paths of optimum connectivity) or dynamically rated maps (dynamically scrutinizing the state of the latest traffic to
recognize the best anchor paths) to determine an anchor path with a high connectivity for packet delivery. Next, A-STAR
utilizes a novel local recovery strategy for packets routed to a local minimum, which is more appropriate for a metropolitan
environment, as against the greedy approach of GSR and the recovery mode of GPSR.
A-STAR overcomes the shortcomings of GSR and provides a new recovery strategy which computes a new anchor path
from the local maximum through which the packet is routed. With traffic awareness and the novel recovery strategy, A-STAR
surpasses GPSR and GSR [36]. However, as the routing path might not be optimal because of its existence along with the
anchor path, the disadvantage of A-STAR is that it eventually becomes time consuming.
1.6 Spatial and Traffic Aware Routing (STAR) [37]. STAR differs from other position-based routing algorithms. It considers
both the spatial information of distribution nodes in the space and the traffic-aware street topology information obtained from
geographic information systems. Unlike the previous routing protocols that deal with high node mobility in one of these
considerations, STAR is intended to recover the drawbacks of the SAR algorithm. Although SAR has the benefit of its primary
spatial awareness model which allows it to send the packets along lanes, SAR has no knowledge of whether any vehicles are
currently positioned along the streets it selects, which will lead to the local maximum problem. STAR overcomes this problem
by only forwarding packets along streets occupied by vehicles.
STAR is capable of equally utilizing street topology information acquired from the geographic information systems and the
information regarding traffic to make precise routing decisions. In STAR, the traffic conditions are scrutinized and
disseminated by swapping network-level beacons and moving the observations of node neighborhoods. The observations are
conserved in data structures administered by the component of scrutinizing traffic. A node administers the locality of its
neighbors in the neighbor’s table. The node neighborhood is exposed through the beacons. Occasionally, every node
broadcasts a beacon to its neighbors that constitutes the identifier information of the sender, and the sender coordinates the
traffic conditions.
The Dijkstra algorithm is implemented on the acquired graph to identify the shortest route. The packet header comprises
the following elements: the destination identifier, the destination position, and a limited number of anchor points (APs), which
are nodes near a junction or road curve and are recorded in the path discovery packet. However, when the routing fails, the
recovery process accepted by STAR computes a new route from the present node and utilizes the updated traffic information
[37].
Figure 9 illustrates the packet routing in STAR and how STAR surpasses the local maximum issue of the SAR, which
does not have any information regarding the whereabouts of vehicles along the selected streets. A source node S has a packet
addressed to D, as shown in Figure 9. From current knowledge about vehicular traffic and Dijkstra’s algorithm, the node S
computes AP1 and AP2, and sends the packet to the closest neighbor to AP1 (shown as dashed arrows). At a certain point, a
node that passed AP1 takes charge of the packet forwarding toward AP2, but it fails because it has no neighbors in the
appropriate direction; that is, the packet is in a local maximum. Exchanging traffic information allows more accurate routing
decisions but does not make routing failure impossible. The recovery procedure adopted by STAR involves computing new
APs (shown as empty squares) from the current node and exploiting updated traffic information. The packet is forwarded along
the new route (shown as black arrows).
Figure 9 Routing procedure of STAR
1.7 Multi‐hop Routing Protocols for Urban VANET (MURU) [38]. The MURU protocol estimates the quality factors of a
route based on vehicle position, speed, and trajectories. Based on the quality factors, MURU introduces a new metric called the
expected disconnection degree (EDD). Consequently, the MURU nodes require recognizing its individual location and having
an external street map, which includes the existence of an effective location service. This novel metric value is regarded as low
as EDD and is an evaluation of probability, which establishes the breakability of the route at any given period. Depending on
the destination location and street map, the source node computes the shortest trajectory to the destination to find the route.
The shortest trajectory detail is stored in the packet and is used as a directional guideline for the RREQ message. The EDD of
the link between two subsequent nodes is calculated by the node that receives the RREQ message.
The scalability of the RREQ message is enhanced by MURU by employing the pruning method, where the node that
receives the RREQ message will wait for back-off delay that is directly proportional to the EDD between the previous
forwarder of the RREQ and the current one. The node determines whether to decrease the RREQ message or rebroadcast it
during the back-off interval. Meanwhile, the pruning method makes the broadcasting area smaller to receive the RREQ
broadcast. Ultimately, when the destination obtains the RREQ message from various routes, it selects the route with a
minimum EDD. This more compact broadcasting area becomes more challenging if the next hop node is positioned outside the
broadcasting range. Nevertheless, with low cost and delay, MURU offers a superior route with a higher throughput
percentage.
1.8 Predictive Directional Greedy Routing (PDGR) [39]. Gong et al. in 2007proposed the PDGR protocol in which the
weighted score is calculated from two strategies: position first forwarding and direction first forwarding. The next-hop
selection is done based on prediction and is not reliable in all situations. It does not guarantee the delivery of packets to the
node on the edge of the transmission range of the forwarding node, which is considered the most suitable next hop because
ofhigh dynamics of vehicles. This situation leads to low packet delivery ratio, high end-to-end delay, and increased routing
overhead.
1.9 Greedy Perimeter Coordinator Routing (GPCR) [40]. This protocol is independent of the digital map. The digital map is a
fundamental constituent in A-STAR and GSR. GPCR depends on the constant transmission of data packets to a node on the
junction instead of transmitting them across the junction. As shown in Figure 10, node A forwards packets to node B, which is
located on the junction even though the radio range on node A can cover node C, which results in the local maximum problem.
Figure 10 GPCR routing along junctions
A node positioned near the junction is called a coordinator. For packet forwarding, the restricted greedy approach is used.
The coordinator is constantly favoured over a non-coordinator. After the occurrence of the local maximum, it employs a
recovery mode [40].
When GPCR is in recovery mode, packets are backtracked in a greedy fashion (i.e., bringing maximum progress) to a
junction node to find an alternate solution to return to the greedy mode. At the junction node, the right-hand rule is used to find
the next road segment to forward the packets.
1.10 GpsrJ+ [41]. Lee et al. in 2007 proposed GpsrJ+ as an intuitive predictive scheme that clears the impediments at an
intersection while maintaining the plans of topographical maps. It employs two-hop neighbor beaconing to visualize the road
segment that would be occupied by the neighboring junction node. If the forecast states that its neighboring junction will send
the packet against a road with a dissimilar direction, it sends to the junction node; otherwise, it diverges from the junction and
sends the packet to its outermost neighboring node. In other words, each node will send a beacon message about its coordinates
and the road segments on which its neighbors are located.
As shown in Figure 11, GpsrJ+ can avoid the junction area and forward the packet from the source node to node E directly.
The source node in the figure has Nodes B and E as its neighbors. Node B sends a beacon message with its location as well as
the road segmentsRoad1, Road2, andRoad3 to the Source node. In the same way, Node E sends a beacon message by its
location as well as the road segment Road2 to the Source node. Based on the road segments it receives from its neighbors,
especially its junction neighbors, the Source node pre-computes the forwarding segment where its next hop will be from the
perspective of Node B according to the right-hand rule. If the pre-computed forwarding segment is the same as the road
segment that the source node’s furthest neighbor is on, the source node will forward to its furthest neighbor E; otherwise, it
will forward to its junction neighbor B.
Figure 11 Illustration of the prediction of GpsrJ+
The main advantage of this protocol is that it does not require an expensive planarization strategy. However, its realistic
roads follow a highly complex trajectory even it used a simple line trajectory.
1.11 Grid-based Predictive Geographical Routing (GPGR) [42]. The GPGR protocol applies the road sections depending on a
routing method with street awareness and utilizes the information of the road topology provided by an inert street map. As a
result, practically all data packets are routed among the vehicles, subsequent to the road topology and the road segments.
GPGR is intended to enhance the routing protocol for IVC depending on the information of the movements of the vehicles
including position, direction, and velocity, along with the information on road topology. To achieve this, GPGR assumes that
all the vehicles know their locations by using GPS that has the most correlated geographic routing protocols and possesses a
digital street map that provides the road information.
In the GPGR protocol, the geographic area of VANET is split into a two-dimensional logical grid (Figure 12). The grids are
tagged subsequent to accustomed x and y coordinates. The size of each grid is d × d. Based on the physical location, having a
map from the location to its grid coordinates is necessary where each vehicle has a radio range of r, and the size of the grid d is
determined by, √
, with a maximum value of d, so that a vehicle, which is positioned at a place in a grid, can broadcast
data to any vehicle in the eight neighboring grids, as shown in Figure 12.
Figure 12 GPGR two-dimensional grid geographical area [42]
GPGR reduces the possibility of link fail and a local maximum by choosing relay nodes depending on the information
related to mobility and road topology. The outcomes of the replication proved that the GPGR generates an extremely small
prospect of a local maximum, prospects of low link fail, and the maximum packet delivery rate for VANETs in contrast to the
GPSR and GPCR protocols.
1.12 Position-based Routing with Distance Vector (PBR-DV) Recovery Protocol [43]. PBR-DV combines the position-based
greedy routing with AODV-style recovery. This protocol follows the position-based greedy routing approach, which is used in
GPSR. Thus, every vehicle periodically transmits beacon messages with their position and vehicle id. Location services are
used to acquire the position of the destination if it is unknown.
While greedy routing is the default behavior, the PBR-DV protocol employs an AODV-style recovery because the packets
fit into a local maximum. A request packet is transmitted by the node at the local maximum, which comprises the location of
the destination. After acquiring the request packet, initially the node will ensure its proximity with the destination than the
node at the local maximum. However, in the opposite way, it considers the node from which it acquires the request packet, and
the request is retransmitted. Otherwise, it transmits a response to the node from which it obtains the request.
Due to the traversing of the reply packet toward the local maximum node, all the transitional nodes register the preceding
node from which they receive the reply packet. Thus, the local maximum node can keep a path to a nearby node. However, the
weakness of this scheme is the need for overflowing to detect the non-greedy part of the route. PRB-DV has not been assessed
or compared with the GPSR or AODV protocols. Thus, information about the performance of this protocol is inconclusive.
1.13 Connectivity-aware Routing (CAR) [44]. The task of CAR protocol is to find a route to a destination; it has distinctive
features that enable it to maintain the cache of effective route between innumerable source and destination pairs. CAR can also
predict the position of destination vehicle reformations route in case there is a change in position. The nodes that utilize CAR
protocols transmit periodical Hello beacons which comprise their velocity vector data. A node will register the sender in its
neighbor table and compute its velocity as well as the velocity of its neighbor as soon he that node receives the Hello beacons.
To decrease the loss of bandwidth and congestion, beacons might be piggybacked as well on proceeded data packets. In case
the space between the nodes surpasses the threshold rate, accesses will ran out from the neighbor table. The CAR protocol set
up the notation of a guard which is a geographical marker message; it is buffered and transferred from one vehicle to another to
propagate the information. A guard is a regarded as a temporal message which possesses an ID, a time to live counts, a radius
and some state data. Two models of guards can be presented by CAR which are the standing guard and the traveling guard.
Routing errors may either result from a gap in the communication between the anchor points or due to problems in the guards
themselves. To overcome such inconveniences, CAR protocol devised two recovery strategies.
The first of these recovery strategies is called the time out algorithm with active waiting cycle and the second strategy is
called the walk around error recovery. The CAR protocol has a unique merit that is not found in other protocols which that it is
able to create virtual information in the form of guards.
1.14 Greedy Traffic-aware Routing (GyTAR) [45]. GyTAR is a junction-based routing protocol that effectively chooses the
junction to find potential routes within the town and employs a carry-and-forward approach to recover from the local
maximum. It utilizes a digital map to identify the location of neighboring junctions and efficiently selects the connection based
on the traffic density and curve-metric distance to the destination. A score is given to all neighboring junctions based on the
traffic density Tj and the curve-metric distance Dj to the destination. The junction j with the maximum score is chosen as the
next intersection. The chosen junction is the one near the destination and encompasses the maximum traffic density. The
maximum score is calculated as follows:
= + (1)
where α and β are the weighting factors.
In GyTAR, the greedy routing strategy is employed to send the packet between two implicated junctions. For this reason,
employing GyTAR makes a packet move closer sequentially to the destination along roads where an adequate number of
vehicles offer connectivity [45].
1.15 Landmark Overlays for Urban Vehicular Routing Environments (LOUVRE) [46]. LOUVRE establishes an overlay
network on top of an urban topology and ensures obstacle-free geographic routing on the overlay links. Lee et al. in 2008
classified the geographic greedy overlay routing into two composites. The primary composite is geo-reactive overlay routing,
in which the next superimposed node is determined by the distance to the destination of neighboring nodes. The second
composite is geo-proactive overlay routing, in which the series of overlaid nodes is determined a priori (GSR and A-STAR).
The LOUVRE protocol belongs to the second composite.
LOUVRE considers the fact that above a given density threshold, an overlay link remains connected regardless of the
spatial and temporal distribution of the vehicle on the link. Thus, by considering only the overlay links depending on the
density threshold when the building overlay routes, most routes would slightly use the same overlay links. With these
considerations, geo-proactive overlay routing becomes attractive as it ensures global route optimality and reduces the delay in
establishing overlay routes.
1.16 Diagonal Intersection-based Routing (DIR) [47]. To develop the CAR protocol, Chen et al. in [47] developed a DIR
protocol. The chief modification of CAR and DIR protocols mechanisms from the fact that DIR protocol puts up a sequence of
diagonal intersections between the source and destination vehicles. The DIR protocol is considered to be a geographical
routing protocol. Depending on the geographical routing protocol, the source vehicle geographically advances the data packet
toward the first diagonal intersection, the second and the third diagonal intersection until the last diagonal joint, and lastly it
geographically reaches the destination vehicle. For a specific pair of neighboring diagonal intersections, two or more separate
sub-paths are found between them.
The new feature of DIR protocol is the auto-adjustability; while the auto-adjustability attained that one sub-path with low
data packet delay, between two neighboring diagonal intersections, is vitally chosen to progress data packets. To lessen the
data packet delay, the route is routinely re-routed via the chosen sub-path with the least amount of delay and the DIR protocol
is able to routinely regulate routing path to obtain lower packet delay, compared to CAR protocol.
1.17 Receive on Most Stable Group-Path (ROMSGP) [48]. Taleb et al. in 2007 introduced the ROMSGP routing protocol in a
metropolitan environment to enhance routing consistency. According to them, an unhinged routing typically occurs because of
the loss of connectivity, which normally happens when a vehicle moves out of the coverage area of a neighboring vehicle.
In the ROMSGP protocol, vehicles are divided into four groups depending on the velocity vector. A routing is considered
stable if two vehicles belong in the same group; otherwise, it is considered unsteady. A vehicle belongs to a group if the
velocity vector has the highest projection vector within this cluster.
1.18 Junction-based Adaptive Reactive Routing (JARR) [49]. The network topology of VANET in a metropolitan setting
consists of numerous probable paths and junctions that form the routing paths. The shortest path routing is not practical as
every path must be inhabited by vehicles. A scalable multi-hop routing protocol that is perfectly suitable to a city setting with
swiftly changing network topologies and plenty of detached and intense network conditions is therefore required. JARR aims
to deal with the inadequacies of the present protocols by assessing the density of paths.
The density of a path can be assessed by determining the beaconing rate, which depends on information gathered from one-
hop neighbors. This information also helps assess the density of a path since the beaconing rate is affected by density, and
density is affected by velocity. With information on the velocity of nodes, the density of a path can be assessed.
However, vehicles may move at a slow speed even in a sparse condition. Thus, both the beaconing rate and the velocity of
vehicles are used to assess the density on certain in path. The beaconing rate starts off with an initial rate. Then, the beacon rate
of a node is determined by the surrounding node densities, which means that by obtaining the beaconing rate information of a
node, the density information around that node can also be obtained.
1.19 Edge Node Based Greedy Routing (EBGR) [50]. Prasanth et al. in 2009 proposed a position-based greedy forwarding
strategy routing protocol called EBGR. It employs a unicast forwarding strategy for transmitting the message from any node to
any other node, or a broadcast forwarding strategy for transmitting the message from one node to all other nodes in highly
dynamic networks. This protocol selects the edge node of the limited transmission range as a next hop node for sending a
message from source to destination. In this protocol, a packet is sent to the edge node with consideration of the nodes that
move in the direction of the destination. During packet transmission from source to destination, EBGR uses the following basic
methods:
(1) Neighbor node selection method, which is used to collect information on all direct neighbors within the transmission
range of the source node;
(2) Node direction identification method, which is used to identify the direction of moving nodes toward the direction of
the destination; and
(3) Edge node selection method, which is used to select the edge node as a next hop node within the transmission range to
further forward the packet.
EBGR can be used to minimize the number of hops between source and destination, and to maximize the network throughput.
1.20 Border Node-based Most Forward within Radius (B-MFR) [51]. B-MFR uses the concept of border node within the
sender’s communication range to minimize the number of hops between source and destination. It classifies the nodes into
three categories: interior, border, and outer nodes (Figure 13). The interior nodes refer to all the neighbors inside the circle of
its transmission range. The border nodes refer to the neighbor nodes on the circle (the border of transmission range), and the
nodes outside the circle are the outer nodes.
The source node assumes that it has information about its neighbors through the periodic exchange of beacon packets with
its neighbor’s nodes. All the nodes within the transmission range of the source node are its one-hop neighbor. After obtaining
the list of its one-hop neighbor, the source node chooses the next forwarding node to deliver the packet to the destination. The
border node is selected as the next forwarding node because the border node is the only neighbor node, which is farthest from
the source node and nearest to the destination [51][52]. By projecting all the border nodes on the straight line which connects
the source and destination, B-MFR selects the one farthest from the destination.
As shown in Figure 13, border nodes A and B are projected on the line segment SD which connects source and destination.
From the projection, border node A is clearly closer to the destination compared with border node B. Therefore, among all the
border nodes, node A is selected as the next forwarding node. Similarly, node A selects the border node E as a next forwarding
node because it is closer to the destination than border node F. This greedy approach continues until the destination node is in
the transmission range of the current forwarding node. As shown in the same figure, destination node D is in the transmission
range of node E. Thus, node E delivers the packet to node D and the process is terminated. However, selecting the next
forwarding node out of all the border nodes is difficult sometimes.
Figure 13 B-MFR forwarding method
The B-MFR protocol faces problems with conflicting nodes. For example, if a two-border node is projected to the same
point on the source and destination, and the nodes have the same distance from the source and destination. In this case,
deciding the next forwarding node in B-MFR is difficult.
1.21 Adaptive Movement Aware Routing (AMAR) [52]. Brahmi et al. in 2009 proposed the movement-aware greedy
forwarding routing protocol, which depends on the greedy forwarding scheme, to choose the next-hop node toward the
destination. AMAR solves the problem of B-MFR by employing further information about vehicle movement to choose a
suitable next-hop of the packet that ensures data delivery. In this protocol, a border node is selected out of the two conflicting
nodes by using mobility awareness such as the parameters of speed and direction. It calculates the weight score Wi for border
node i based on the position, speed, and direction as follows:
(2)
where , , and are the weight of the three used metrics Pm, Dm, and Sm which represent the position, direction, and speed
factors, respectively, with 1.
A sorted list of next-hop candidates can be defined based on the computed score Wi. The node with the highest weighted
score among all the border nodes of the current forwarder is selected as the best candidate for the next forwarding node. It also
improves the data delivery.
This protocol solves the problem of B-MFR, but the problem is that if the weighted score of the two border nodes are equal,
then deciding the next forwarding node is difficult.
1.22 Topology-assist Geo-Opportunistic (TO-GO) Routing [53]. Lee et al. in 2009 introduced the TO-GO routing protocol by
improving GpsrJ+ with opportunistic forwarding. In contrast to earlier strategies in which a forwarding region is identified
between the current sender and the destination, TO-GO identifies a forwarding set between the current sender and the anchor
node. Later, it is incorporated in a packet, which enables nodes to compete for forwarding; every single node sets a timer based
on its associated distance to the anchor node. Timer T can be computed as follows:
,
, (3)
where C is a constant that differs with the transmission rate, the processing time, and the system throughput. T is relative to the
distance between the receiving node and the anchor node, and is negatively proportional to the distance between the receiving
node and the sending node. As the nodes closer to the sending nodes consume less time to receive the broadcast compared with
the nodes further away, the closer nodes’ timer should be responsible for the shorter propagation delay so that each node
initiates the timer at approximately the same time. Primary investigations revealed that the functionality of TO-GO is as good
as that of GpsrJ+ and improves as packets are delivered opportunistically toward the destination in an unpredictable, loss
wireless medium.
2. Beaconless
In position-based routing, beaconing enables the exchange of a variety of information between a node and its neighbors.
The VANET environment is characterized by high mobility, which is why the beacon information might quickly become
invalid. This type of protocol does not use regular beacon messages to track the location and the information of the neighbor’s
nodes. The most popular beaconless routing protocol is contention-based forwarding (CBF) [54].
2.1 Contention-based Forwarding (CBF) [54]. CBF is a position-based routing protocol that does not employ beacon
messages. The exclusion of beacon messages saves bandwidth. This protocol transmits the data packet to all direct neighbors
and identifies the one that will forward the packet.
The actual forwarder is selected by using a distributed timer-based contention process which allows the most suitable node
to forward the packet and to suppress other potential forwarders. Receivers of the forwarded data packet would compare their
distance with the destination to the last hop’s distance to the destination. A larger difference indicates greater progress and
shorter time.
The main advantage of this protocol is that it saves bandwidth because it eliminates the periodic forwarding of the beacon
message. The CBF protocol reduces the probability of packet collision and inefficient routing by ignoring inaccurate neighbor
tables. Meanwhile, in a highway destination, CBF is always straightforward, so a local maximum never occurs, and CBF
works well as a result. However, in a city environment, a local maximum frequently occurs because the source and the
destination may lie on different paths.
3.1.2.2 Routing protocols for delay-tolerant networks (DTNs)
DTN is an approach to computer network architecture that seeks to address the technical issues in heterogeneous
networks that may lack continuous network connectivity, which results in a lack of instantaneous end-to-end paths. Examples
of such networks are those which operate in mobile or extreme terrestrial environments, or planned networks in space. The
vehicular routing protocols are developed for VANETs that are considered a form of DTN. Given the challenging
environments of this kind of network, they are affected by regular connection loss. To solve this problem, the packet delivery
is increased by permitting the nodes to store the packets when they lose contact with other nodes, to take the packets for a
certain distance as long as it meets with other nodes, and to forward the packets depending on certain metrics to the
neighboring nodes; this is called the carry-and-forward strategy. The best known of these protocols are SKVR, VADD, and
GeOpps [55-57].
1.Scalable Knowledge-based Vehicular Routing (SKVR) [55]. SKVR divides the network between inter‐domain and
intra‐domain. In inter‐domain routing, the source and destination belong to different routes, whereas in intra‐domain routing,
the source and destination belong to the same route. In the inter‐domain algorithm, the message is forwarded to a vehicle
traveling in the destination domain. Once the destination domain is reached, the intra‐domain message delivery procedure is
followed. In intra‐domain routing, the messages are sent in forward or reverse directions depending on the entries of the
contact list. If the sending vehicle contact list does not contain any vehicle in the destination domain, then the messages are
delivered to the other vehicles in the contact list. When vehicles along the same route encounter one another, a node that
carries a message must decide whether to continue buffering the message or to forward it based on the direction information of
the vehicle.
2. Vehicle-assisted Data Delivery (VADD) [56]. VADD is a vehicular routing strategy intended to enhance routing in
disconnected vehicular networks based on the carry-and-forward concept depending on the use of conventional vehicle
mobility. A vehicle makes a decision at an intersection and chooses the next forwarding path with the negligible packet
delivery delay. A path is only a split road from an intersection. The best path of the packet forwarding is chosen by switching
between three packet modes (Intersection, Straight Way, and Destination). Figure 14 illustrates the mechanism of the packet
forwarding among the three packet modes.
Figure 14 Three-packet mode of VADD [56]
3. Geographical Opportunistic (GeOpps) [57]. GeOpps benefits from the recommended routes of vehicles’ navigation system
to choose vehicles, which may travel closer to the ultimate destination of a packet. It computes the straight distance from the
destination of packets to the nearest point (NP) of the vehicles’ path and approximates the arrival time of a packet to the
destination. The NP computation of neighbors N1 and N2 of Node A is illustrated in Figure 15. Given that N2 offers the closer
NP to the destination, Node A picks N2 to forward its packets.
The packet is forwarded to a vehicle if another vehicle is present, which has a shorter approximate arrival time. The process
is repeated as long as the packet does not reach the destination.
Figure 15 Calculation of the NP from the packet’s destination for N1 and N2
3.1.2.3 Hybrid position-based protocols
Conventionally, the packets are routed through the greedy and recovery modes by geo-routing. In the greedy mode, a
packet is delivered to the destination greedily by selecting a neighbor which has a better movement toward the destination
among all the neighbors. However, given the obstacles, the packet can reach a local maximum where no neighbor is closer to
the destination than itself. In such a case, the recovery mode is employed to extract packets from the local maximum and to
ultimately return to the greedy mode. After a planarization process, packets are delivered across the obstacles toward the
destination. Similarly, the packet delivery is assured provided that the network is connected, but the presumption that the
network is active might not be correct at all times. Owing to the mobile features of VANET, it is normal that the network is
disconnected or partitioned predominantly in sparse networks. The greedy and recovery modes are inadequate in VANET.
Consequently, the non-DTN routing strategy, which is depicted by the two preview modes, is merged with the DTN routing
strategy to recover from this problem. GeoDTN+Nav [58] is one of the most well-known hybrid position-based routing
protocols.
1. GeoDTN+Nav [58]. GeoDTN+Nav is a combination of non-DTN and DTN routing protocols that include the greedy mode,
the perimeter mode, and the DTN mode. It toggles from non-DTN mode to DTN mode by approximating the connectivity of
the network depending on the number of hops a packet has travelled so far, the neighbor’s delivery quality, and the neighbor’s
direction in terms of the destination. The DTN mode can deliver packets even if the network is disconnected or partitioned by
taking advantage of the mobility of vehicles in VANET. In other words, packets are forwarded first in greedy mode and then in
recovery mode when a packet hits a local maximum. If the recovery mode also fails, it finally switches to the DTN mode and
relies on mobility to deliver the packets. Figure16 illustrates the transition diagram between these three modes.
Figure 16 Transition between greedy, perimeter, and DTN modes [58]
The delivery quality of neighbors is obtained through the virtual navigation interface (VNI),which abstracts information
from underlying hardware (e.g., navigation system and event data recorder or EDR) [58] as well as provides necessary
information for GeoDTN+Nav to determine its routing mode and forwarder. In addition to its hybrid approach, VNI offers
users the option to protect private data and at the same time provides the best-effort routing decision.
The simulation results in [58] show that GeoDTN+Nav outperforms GPSR and GPCR because it can estimate network
partitions and then improve the partitions’ reachability by using a store-carry-forward procedure when necessary.
3.1.3 Cluster-based routing protocols
Generally, cluster-based routing protocols are more suitable for network cluster topology. As shown in Figure 17, every
cluster has one cluster head which is responsible for intra- and inter-cluster management purposes. The intra-cluster nodes
interact with one another through direct links, whereas inter-cluster interaction is performed through the cluster headers.
Figure 17 Vehicles from multiple clusters in cluster-based routing [8]
In cluster-based routing protocols, the vehicles close to each other form a cluster. However, in the cluster-based routing
protocols, the configuration of clusters and the choice of the cluster-head is an important issue. Given the high mobility of
VANET, the configuration of the dynamic cluster becomes a major process [8]. The most common cluster-based routing
protocols are CBR, CBDRP, LORA-CBF, COIN, and TIBCRPH [59] [60] [61] [62] [63].
1. Cluster-based Routing (CBR) [59]. In the CBR protocol, the geographic area is separated into square grids. Every single
node computes the best possible neighbor cluster header to deliver data to the next hop based on the geographic information.
The routing overhead is less as it does not need to discover the route and is saved in the routing table. The cluster header
broadcasts a LEAD message to its neighbors with coordinates of its grid and the location of the cluster header. If an RSU is
present in the grid, it will become a cluster header. Every time the header exits the grid, it will broadcast a LEAVE message
that contains its grid position, which will be stored by an intermediate node until a new cluster header is chosen. The new
cluster header uses this information in data routing. This protocol does not consider velocity and direction, which are
significant factors in VANET.
2. Cluster-based Directional Routing Protocol (CBDRP) [60]. In the CBDRP protocol, the vehicles which travel in the same
route are split into several clusters. Each vehicle can communicate via radio with its neighbor clusters. An example of cluster
splitting is illustrated in Figure 18.
Figure 18 Cluster splitting example in CBDRP [60]
The figure shows two clusters. The center position of one cluster is fixed after it is divided. The CBDRP protocol assumes
that the radius of the radio is r, the length of every cluster is d, and the width of half of the highway is w. Given that d>w, the d
is almost equal to r/2. In 802.11 p, the radius of the radio transmission is 1,000 m, so the theoretical length of one cluster can
be as much as 500 m. If the header approaches the center position, the d can be much larger.
The source node in CBDRP forwards the message to its cluster header and then sends the message to the header, which is
in the same cluster as the destination. Ultimately, the destination header forwards the message to the destination. The cluster
header selection and continuance is similar to that of CBR, but it considers the speed and direction of a vehicle. Simulation
results show that the CBDRP can solve the problem of link stability in VANET, thus ensuring reliable and rapid data
transmission.
3. Location Routing Algorithm with Cluster-based Flooding (LORA-CBF) [61]. In the LORA-CBF protocol, each node can
become the cluster head, gateway, or cluster member. All clusters have their own cluster head. The node that links two clusters
is the known gateway. The function of the cluster head is to manage the information of its members and gateways. The
forwarding of the packet is similar to that of greedy strategy routing [54]. However, the cluster head and gateways can only
forward the location request (LREQ) packets, and upon receiving a LREQ, each cluster head checks that the destination is a
member of its cluster. Success triggers a location reply (LREP) packet that returns to the sender by using geographic routing
because each node knows the position of the source and the closest neighbor based on the information received from the LREQ
and the neighbor-sensing mechanism. LORA-CBF is robust in terms of routing overhead, routing load, and delivery ration.
4. Clustering for Open IVC Network (COIN) [62]. Blum et al. in 2003 proposed the COIN protocol for IVC networks. COIN
enhances the stability of a cluster. The clustering approach is designed in a manner that improves scalability. Cluster selection
is based on node mobility, driver behavior, and distance between vehicles instead of ID or relative mobility as in conventional
clustering methods. The relative mobility between a cluster head and a member node should be low, so they stay in the radio
range for as long as possible. With a minimized supplementary control load, this protocol offers the clusters a time to survive,
which is roughly twofold, and minimizes the modifications in cluster association to a minimum of 46%. Initial simulation
results showed that COIN generates more steady structures in IVC networks with just an extra overhead.
5. Traffic Infrastructure-based Cluster Routing Protocol with Handoff (TIBCRPH) [63]. The overlap between clusters
occurs inevitably because of the feature of radio communication in which the high mobility of the node frequently causes
broken routes. The TIBCRPH protocol adopts the concept of handoff in cellular networks to solve this problem. It assists the
transmission of data packets by using the existing traffic infrastructure as cluster head to divide the network into multiple
clusters. Each cluster has a cluster head and a group of members, and the cluster heads of each cluster form a backbone
network. This mode enables the coverage of all roads in the region, which allows the seamless transmission of messages.
Figure 19 Cluster mode in TIBCRPH
The cluster mode shown in Figure 19 leads to overlaps (interference regions) between the clusters (striped region in the
same figure). When vehicles drive across the interference regions, the handoff metric is produced by calculating the dot
product of velocity vector of a vehicle and its two neighboring cluster heads’ direction vector. Through the value of the dot
product, we can estimate the ID of one cluster head, which is more suitable for the vehicle. Thereafter, the node of this ID, as
the cluster head of the vehicle, is selected; then, the vehicle sends or receives messages only through the selected cluster head
inside a cluster, the members communicate with each other directly, and the cluster head can broadcast data packets to the
members or directly send packets to one node.
TIBCRPH is compared with six other typical routing protocols in [63] in terms of node density and speed. The results show
that TIBCRPH always performs well no matter how node density and speed change, which makes it better than several other
traditional routing protocols.
3.1.4 Geocast-based routing protocols
Fundamentally, the geocast-based routing is a position-based multicast routing employed to forward a message to all the
vehicles in a fixed topographical area. The main goal of this approach is to distribute the packet from the source node to all the
other nodes in a particular geographical area or zone of relevance (ZOR). The ZOR is defined [13] as a geographic region
where vehicles should receive the geocast messages.
In geocast routing (Figure 20), vehicles outside the ZOR are not alerted to avoid unnecessary quick reactions. Geocast is
considered a multicast service within a specific geographic region. It normally defines a forwarding zone where it directs the
flooding of packets to reduce message overhead and network congestion caused by simply flooding packets everywhere. In the
destination zone, unicast routing can be used to forward the packet.
Figure 20 Geocast routing
One drawback of geocast is network partitioning and the presence of unfavourable neighbors, which may hinder the proper
forwarding of messages. The various geocast routing protocols are IVG, CGR, AGR, ROVER, and Mobicast [64] [65] [66]
[67].
1. Inter-Vehicle Geocast (IVG) [64]. Bachir et al. in 2003 introduced the IVG protocol to disseminate information associated
with security, such as the occurrence of accidents, floods, or other natural calamities, to vehicles on highways. In the IVG
protocol, risk areas are established with respect to the driving direction and the position of the vehicles, which are detected
with the help of the GPS. However, the messages received by the vehicle must not be instantly rebroadcasted but have to be
stored for a certain period (known as defer time) to determine if the messages should be rebroadcasted. When the defer time
ends and if it does not receive a similar alarm message from another vehicle behind it, the protocol detects no relay nodes.
Consequently, it needs to assign itself as a relay and starts broadcasting alarm messages to notify the vehicles that might be
behind the vehicle. The defer time of node x when receiving a message from another node s is inversely symmetrical to the
distance that isolates them from the preferred farthest node to wait for less time and to rebroadcast faster.
The GPS enables IVG to effectively constrain the alarm message dissemination to appropriate areas. These areas determine
the associates of the multicast group geographically. That is, based on double advantage, in the beginning, VG can prevent
operations of the maintenance of the multicast tree such as routing and neighbor computation. However, these procedures are
extremely costly in highly dynamic environments such as transportation systems. The simulation results in [64] proved the
reliability and scalability of the IVG protocol.
2. Cached Geocast Routing (CGR) [65]. CGR is another geocast protocol. The major idea behind the CGR is to add a small
cache to the routing layer that holds the packets which a node cannot forward distantly because of a local minimum. As soon as
a new neighbor arrives within reach or when the recognized neighbors relocate, the cached message can be submitted to the
recently identified node. Their distance-aware neighborhood approach considers frequent neighborhood changes. It selects the
nearest node to the destination, which is within the range (smaller than the transmission range), instead of the node
transmission range in the general greedy routing mode. The enhanced neighborhood selection considers persistent
neighborhood changes, which considerably decreases the network load and decreases the end-to-end delivery delay.
3. Abiding Geocast Routing (AGR) [66]. AG is a special geocast routing protocol apart from the classical geocast routing,
which requires that the packets be sent to all nodes during the geocast lifetime (a certain period) inside the geocast destination
region. Services such as position-based advertising, position-based publish-and-subscribe, and many other location-based
services benefit from AGR. For VANETs, AGR allows the realization of information and safety applications such as virtual
warning signs. Similar to real traffic of warning signs, they are attached to a certain geographical position or area. As soon as a
vehicle enters such an area, the virtual warning sign is displayed to the driver. The following three solutions are suggested:
(1) A server is used to store the geocast messages,
(2) An elected node inside the geocast region stores the messages, and
(3) Each node stores all geocast packets destined for its location and keeps the neighbor information.
4. Robust Vehicular Routing (ROVER) [67]. ROVER is a reliable geographical multicast protocol in which only control
packets are broadcasted in the network, and the data packets are unicasted. This protocol aims to send a message to all other
vehicles within a specified ZOR. A message is defined by the triplet A, M, and Z, which indicates the specified application,
message, and identity of a zone, respectively. When a vehicle receives a message, it accepts the message if it is within the
ZOR. It also defines a ZOF, which comprises the source and the ZOR. In the routing process, all vehicles in the ZOF are used.
The protocol uses a reactive route discovery process within a ZOR.
This protocol creates numerous redundant messages in the network, which leads to congestion and a long data transfer
delay. To address this problem, the authors in [68] proposed a two-zone dissemination protocol for the VANET that uses the
hop count in the packet and is reduced as soon as the packet is delivered. If the hop count reaches zero, the packet is removed.
It triggers the nodes that are close to the sender to send a packet numerous times. To prevent redundant messages, the sequence
number for every single packet is presented to recognize whether or not a packet has been received already.
5. Mobicast [69]. Chen et al. in 2010 proposed the Mobicast protocol as a spatiotemporal multicast/geocast routing protocol.
Mobicast considers not only the space but also the time factor in the geocast routing. As shown in Figure 21, the goal of this
protocol is to transmit the Mobicast message from a source node (V1) to all nodes which belong to the ZOR at time t called
ZORt. To achieve this goal, the authors suggested a method to estimate the accurate ZOF to solve temporal network
fragmentation. This method entails the use of an adaptable zone of approaching (ZOA) to dynamically form the flexible ZOF
and to disseminate the Mobicast message to the ZOR at the appropriate time.
Figure 21 Mobicast ZOR
3.1.5 Multicast-based routing protocols
Multicast transmission in the VANET is typically a transmission from a single source to multiple destinations within a specific
geographic region and is usually conducted via geocast routing [70]. Traditional multicast protocols were designed for wired
networks that have a stable network topology. VANETs are considerably different from such networks. For this reason,
traditional multicast protocols do not perform efficiently for vehicular environments. VANET multicast protocols have to
adapt to the characteristics of these types of networks by considering high node mobility, high movement speed, frequent
topological changes that necessitate constant delivery path updates and have to keep link state information as accurate as
possible. VANETs are beneficial for multicast protocols because of their wireless nature, which allows a message sent by a
node to be broadcast to all nodes within range. VANET nodes do not have to conserve power because vehicles provide
substantial power supply for long durations, which enable the nodes to perform complex computational tasks.
Several multicast protocols have been proposed for MANET [71], most of which are also valid for VANETs. In wireless
MANET multicast routing protocols, a multicast group is composed of senders and receivers. To connect senders and
receivers, each protocol constructs either a tree or a mesh as the routing structure. Forwarding nodes in the routing structure are
not interested in multicast packets but function as routers to forward such packets to receivers. Group members (senders and
receivers) and forwarding nodes are also called tree or mesh nodes depending on the routing structure. Based on the multicast
routing structure, the multicast routing protocols can be classified as either tree-based or mesh-based. The most common
multicast routing protocols are multicast ad hoc on-demand distance vector (MAODV), adaptive demand-driven multicast
routing (ADMR), and multicast with ant colony optimization for VANETs based on MAODV (MAV-MAODV), MOLSR,
ODMRP, and D-ODMRP [72] [73] [74, 75] [76] [77].
3.1.5.1 Tree-based multicast routing protocols
A tree-based multicast routing protocol establishes and maintains a shared multicast routing tree to deliver data from a source
to receivers in a multicast group. Tree-based protocols function poorly on VANETs because they have to rebuild the
distribution tree frequently as a result of high node mobility, which leads to continuous service disruptions. However, certain
tree-based protocols attempt to provide multicast for mobile networks. Among the well-known tree-based multicast routing
protocols are MAODV, ADMR, and MAV-MAODV.
1. Multicast Ad Hoc On-Demand Distance Vector (MAODV) protocol [72]. The MAODV protocol is an extension of AODV.
MAODV conducts broadcast to discover on-demand new routes. As shown in Figure 22a, when a sender node S wants to join a
multicast group or has data to send but does not have a route to the multicast tree, this sender node broadcasts a RREQ
message. The rest of the nodes rebroadcast the message to their neighbors until it reaches a node that is part of the multicast
group tree. To establish a reverse route to the source of the RREQ, these nodes save the address of the node that has sent them
the message in their routing tables. When a multicast group member node receives the message, it sends back an RREP via
unicast, as illustrated in Figure 22b. The message originator may receive more than one RREP. In this case, the message
originator selects the shortest route (by using hop count as metric) and sends a multicast activation (MACT) message along this
path, as shown in Figure 22c.
Figure 22 MAODV join operation: (a) RREQ message, (b) RREP message, and (c) MACT message
By acknowledging the MACT message exchange, the node becomes a member of the multicast group, and all the nodes
along the selected path from this particular node to the one that receives the MACT message become forwarding nodes. The
tree link status is monitored by the multicast group leader, the first member of the group, by sending Group Hello beacons
along the tree. If a link breaks, the involved forwarding nodes are pruned, and a new multicast tree is formed by the nodes that
are disconnected from the tree of the leader. The nodes select a new leader, and if they later receive a Group Hello from a
different leader, then a reconnection to the main tree is initiated.
The main disadvantages of MAODV are long delays and high overheads related to the need to fix broken links in high
mobility and traffic load situations. Moreover, MAODV has a low packet delivery ratio in high mobility situations, numerous
members, or heavy traffic load. Given its dependence on AODV, MAODV is not resilient and suffers from a single point of
failure, which is the multicast group leader.
2. Adaptive Demand-driven Multicast Routing (ADMR) [73]. In ADMR protocol, a tree for every source–multicast pair is
maintained by ADMR. Each tree is preserved by a recurring deluge of keep-alive packets within it. The multicast routing state
in ADMR is dynamically organized and preserved only for active groups that consist of at least one receiver and one active
sender in the network. Every multicast data packet is transmitted from the sender to the receivers through the shortest delay
path via the multicast forwarding state. Senders do not need to start or cease sending data to the group or to enter the group to
which they want to send a packet. Receivers adjust dynamically to the transmission pattern of senders and mobility in the
network. ADMR also traces when mobility in the network is extremely high to maintain the multicast routing state efficiently,
and then the protocol reverts to deluging for a short span of time if it can determine that the high mobility has abated. ADMR
governs the traffic pattern of the multicast source application depending on which protocol can trace link breaks in the tree,
sources that are not sending any data and inactive sources. In case of breaks, the protocol starts local repair processes and
initiates global repair if local repair fails.
ADMR adjusts effectively to the network load and eliminates redundancy. The disadvantage of this protocol is that a
significant amount of state information has to be preserved at every node for every group source. Joining a group is extremely
costly. A receiver should first send a flood. Thereafter, every source must respond to the new receiver. The receiver must then
send an approval to every source. In case the tree breaks regularly and the receiver continuously attempts to join the group, the
process becomes even more expensive. Eventually, the protocol signifies how the source switches to flooding mode for high
mobility but does not indicate how it reverts to a lower mode in case mobility declines.
3. Multicast with Ant Colony Optimization for VANETs based on MAODV (MAV-AODV) protocol [74]. The MAV-AODV
protocol is based on the MAODV protocol in the sense that MAV-AODV uses vehicular mobility information to build stable
trees. It uses the principles of the ant colony optimization meta-heuristic [78] to qualify routes based on the deposited
pheromone amount and to evaluate route stability. To determine this stability, the MAV-AODV protocol employs beacons to
make vehicles aware of information about other vehicles in its group. When a beacon message arrives at a node, it calculates
the estimated link lifetime. By using this mobility information, the protocol defines the procedures for the route request to the
multicast tree and for the multicast route reply via two messages: route request message (Ant-RREQ-J) and reply message
(Ant-RREP).
Ant-RREQ-J is sent to the destination node through broadcast. Each Ant-RREQ-J sent loads the lower lifetime of a link of
the route (lifetime field) and the hop count to the destination node (hop count field). The link lifetime is estimated by using
beacon messages and is constantly updated to make it valid and usable. This Ant-RREQ-J message consists of all RREQ
message fields of the MAODV protocol and the lifetime field, which is new. If a node receives an Ant-RREQ-J to a multicast
group, it can reply if it has a route to the multicast tree, and its sequence number registered for the multicast group is greater
than or equal to that contained in the Ant-RREQ-J. The node that replies to an Ant-RREQ-J inserts the next hop information
from the routing tables and generates an Ant-RREP.
Before generating an Ant-RREP, the node examines two fields in the Ant-RREQ-J: lifetime and hop count. Based on these
values, the node calculates the pheromones on the route travelled by the Ant-RREQ-J. Figure 23 illustrates the path of Ant-
RREPs (represented as ants) to the source node. The Ant-RREPs deposit pheromones in the multicast routing tables of the
nodes (pheromone field) all the way back to the source node. The multicast tree built by using MAV-AODV uses paths with
longer lifetimes and smaller hop counts. Such paths are chosen probabilistically depending on the amount of pheromones
deposited by routing messages that function as ants. MAC-AODV performs better than MAODV in terms of packet delivery
ratio, routing overhead, and maximum end-to-end delay metrics.
Figure 23 MAV-AODV route reply operation [74]
4. The Multicast Optimized Link State routing (MOLSR) protocol [75]: MOLSR belongs to the source tree based family. It
takes advantage of the topology knowledge gathered by the OLSR. MOLSR is the extended version of OLSR protocol. A
multicast tree is built and maintained per pair (source, multicast group) in a distributed manner, devoid of any central entity
and offers shortest routes from the source to multicast group members. When the topology changes are deleted, the trees are
updated.
Once a source node needs to transmit data to a particular multicast group, it sends a SOURCE_CLAIM message permitting
node, which includes members of this group, to identify its existence and to attach themselves, to the associated multicast tree.
These messages are flooded within the ad hoc network, by using the optimized flooding technique of OLSR. Branches are
built, hop by hop in a backward mode as follows. When a group member gets a SOURCE CLAIM message and it is not
previously a participator of this (source, multicast group) tree, it attaches itself to the tree, and assigns the subsequent hop to
arrive at the source in a shortest path as its parent in the multicast tree and transmits a CONFIRM PARENT message to it.
Upon receiving this message, the parent node attaches itself to the (source, multicast group) tree, if it is not previously a
participator to this tree. The trees are regularly renewed, based on the SOURCE CLAIM message and the CONFIRM
PARENT message. It is noteworthy that, the changes of topology are still discovered by the swapping of topology control
messages, which is generally, carried out by OLSR. Thus, updates of trees are induced by the detecting the changes of
topology.
However for the non-duplicate packets received, the MOLSR node ascertains, if it must forward the packet or eliminate it.
On the other hand, if this node is attached to the associated tree as a parent, then it forwards the packet (with neighbor-cast), or
else the packet is not routed.
Due to the reactive nature of this protocol, minimal control overhead is produced for maintaining multicast trees; however,
initially few packets, which are disseminated while forming a multicast tree, might encounter some delay and packet loss.
These kinds of delay and packet loss might not be suitable particularly for inter-vehicle safety and emergency applications,
which demand rapid and trusted dissemination of information.
3.1.5.2 Mesh-based multicast routing protocols
A mesh-based multicast routing protocol sustains a mesh that consists of a connected component of the network that contains
all receivers in a group. Examples of mesh-based multicast routing approaches are on-demand multicast routing protocol
(ODMRP) and destination-driven on-demand multicast routing protocol (D-ODMRP).
1. On-demand Multicast Routing Protocol (ODMRP) [76]. ODMRP is based on a forwarding group concept in which the
multicast packets are forwarded via scoped flooding by a subset of nodes. As illustrated in Figure 24a, when a sender node S
wants to join a multicast group or has data to send, it periodically broadcasts a JOIN QUERY message to the network. This
message refreshes routing and group membership information because ODMRP uses the soft-state approach to keep its
multicast routing information. When a node that is not a member of any multicast group receives a JOIN QUERY, the protocol
checks if the node is a duplicate; otherwise, the protocol stores the ID of the upstream node and then rebroadcasts the message.
When the message reaches a multicast group member, the receivers (R1, R2, and R3) create an entry in its member table or
update the correspondent entry if the message already exists, and then broadcast the message to their neighbors in a JOIN
REPLY message, as shown in Figure 24b. Every node that receives the message and reads its ID on the table knows that the
message is on the path, so the node establishes itself as a forwarding node (by setting the forwarding group flag) and
propagates a new JOIN REPLY message. This process creates a mesh that connects the source to every multicast group
member. When a node wants to leave the multicast group, it simply has to stop sending JOIN QUERIES because, as
mentioned, the ODMRP protocol uses the soft-state approach through which a route that is not refreshed is deleted.
Figure 24 ODMRP join operation: (a) JOIN QUERY and (b) JOIN REPLY
2. Destination driven on demand multicast routing protocol (D-ODMRP) [77]. D-ODMRP is an extension of ODMRP. D-
ODMRP aims to improve multicast forwarding efficiency. To achieve this goal, the path to reach a multicast destination (from
source to receiver) is biased toward directions that pass through another multicast destination. If several paths are available, the
protocol selects the least costly path. In other words, D-ODMRP is employed to reduce the number of nodes to be added to the
forwarding group.
D-ODMRP introduces destination-driven features to the existing on-demand process of the multicast forwarding structure. The
protocol modifies the regular route request flooding process to build a multicast tree by intensely adding a deferring time at
each intermediate node to forward every route request based on the distance of this intermediate node from the last multicast
group member visited by the received route request; a greater distance increases the deferring time. In considering the
deferring time, the route request introduces the least extra cost by identifying nodes that travel faster than those that have
longer deferring times. As a result, a cost-effective forwarding structure can be created. No extra overhead is incurred, unlike
with an existing on-demand routing protocol such as ODMRP.
3.1.6 Broadcast-based routing protocols
Broadcast-based routing is normally used in VANET to share information about road conditions, the climate, and urgent
situations with vehicles, and for advertising and announcements[13]. Broadcast-based routing protocols follow the simple
broadcast method by flooding, in which each node retransmits the message to other nodes. This process guarantees the arrival
of the message to all destinations but has a higher overhead cost. Furthermore, it is suitable only for a smaller number of nodes
in the network. A greater node density results in more message broadcasts that lead to collisions, higher bandwidth utilization,
and a decrease in the overall performance of the system [13]. Broadcast-based routing protocols include BROADCOMM,
UMB, DV-CAST, EAEP, HyDi and DECA [79] [12] [80] [81] [82] [83].
1. BROADCOMM [79]. BROADCOMM is based on a hierarchal structure for a highway network. In BROADCOMM, the
highway is divided into virtual cells that move like vehicles. These cells have a similar and equivalent optimal length, which
facilitates the best possible transmission and reception. The nodes in the highway are organized into two levels: the first level
includes all the nodes in a cell, and the second level is represented by cell reflectors, where few nodes are located close to the
geographical center of the cell.
The cell functions for a specific interval of time as the cluster head and handles emergency messages sent by the same
members of the cell or by a nearby neighbor. This protocol works in a similar manner as flooding-based routing protocols for
message broadcasting and routing overhead.
2. Urban Multi-hop Broadcast (UMB) [12]. UMB is designed to address (i) the broadcast storm, (ii) the hidden node, and (iii)
reliability problems of multi-hop broadcast in urban areas. This protocol assigns the responsibility of forwarding and
acknowledging broadcasts to only one vehicle by splitting the sections of the road inside the transmission range into portions
and selects the vehicle in the furthest non-empty segment without a priori topological information. When an intersection is
present in the path of the message dissemination, a new directional broadcast is initiated by repeaters located at
the intersections.
Figure 25 illustrates an example of intersection handling in UMB, where the directional broadcast is used by the vehicle A
to reach vehicle B. Vehicle A is out of the transmission range of the repeater C. Meanwhile, vehicle B is in the transmission
range of repeater C. Therefore, vehicle C uses the IEEE 802.11 protocol [84] to communicate with repeater C. Once repeater C
receives the message, it initiates directional broadcasts to the north and south directions. Given that repeater D is in the
transmission range of repeater C, it also sends the packet to repeater D by using the IEEE 802.11 protocol.
Figure 25 Intersection handling in UMB protocol [12]
The simulations in [12] show that UMB has an extremely high success rate and efficient channel utilization compared with
other flooding-based protocols.
3. Distributed Vehicular Broadcast (DV-CAST) [80]. In DV-CAST, all the vehicles use a flag variable to confirm the
redundancy of packets. It uses information of restricted topology by employing cyclic Hello beacons to send the information.
DV-CAST categorizes the vehicles into three types, namely:
(1) Well connected,
(2) Sparsely connected, and
(3) Totally disconnected neighborhood.
The first type uses a scheme which involves weighted perseverance and slotted land perseverance. In the second type, after
receiving the broadcast message, the vehicles can instantly rebroadcast within vehicles moving in the same direction. In a
completely separated neighborhood, the vehicles store the transmitted message until another vehicle enters the transmission
range. Otherwise, the packet is discarded when the time expires.
4. Edge-aware Epidemic Protocol (EAEP) [81]. EAEP is a reliable and bandwidth-efficient information dissemination method
based on the highly dynamic VANET protocol. It minimizes the control packet expense by eliminating the swapping of added
Hello beacons for transmitting messages between various clusters of vehicles, and relieves cluster management. All the
vehicles carry their own topographical position to send messages to clear beacon messages. After receiving a new rebroadcast
message, the EAEP employs different types of transmission front and back nodes within a specific period to compute the
possibility of making a decision on whether or not the nodes will resend the message.
5. Hybrid Data Dissemination (HyDi) [82]. Maia et al. in 2012proposed the HyDi protocol, a data dissemination protocol for
highway scenarios that flawlessly functions in both well-connected and intermittently connected VANETs. HyDi is intended to
perform directional data dissemination in extreme circumstances, such as those commonly observed in highways where traffic
flows in both directions.
In well-connected networks, the sender- and receiver-based methods in HyDi merge to handle the broadcast storm problem
which is characterized by substantial packet loss, contention, and delay at the link layer. In a sender-based approach, a node
selects a priori the next node to receive a message. However, in a receiver-based approach, one of the nodes that receive the
message is accountable for handling the message; this is an a posteriori decision. If a vehicle finds no other vehicles to which
the message can be delivered, it implements carry-and-forward techniques by keeping the message until a new connection is
recognized with a vehicle that can return the message dissemination to its pre-recognized route. Later, the vehicle broadcasts
the message, and the carry-and-forward state occurs.
Simulation results show that HyDi has an overhead similar to that of DV-CAST, outperforms DV-CAST in terms of
average delay under heavy traffic conditions, decreases the average number of hops to deliver messages, and delivers data to
almost all nodes in a given region.
6. Density-aware Reliable Broadcasting (DECA) [83]. DECA is a broadcasting routing protocol that does not need position
knowledge in its routing operation. It uses only local density information of x-hop neighbors acquired by beaconing. Prior to
broadcasting, a node chooses one neighbor that comprises the uppermost local density information to be the next rebroadcast
node. Other nodes arbitrarily establish their waiting timeout. If they do not hear anyone rebroadcast the message before the
timeout expiration, they will rebroadcast the message. In addition, identifiers of the received broadcast messages are integrated
in periodic beacons so that a node can locate its neighbors that have never received the messages and subsequently rebroadcast
the messages for those neighbors. DECA is more flexible, which allows it to suit any operating environment because it does
not require position knowledge to operate.
3.2 Vehicle-to-infrastructure-based (V2I) routing protocols
VANET routing protocols improve their performance to a certain extent but suffer from network partitioning because of
high mobility. Current research tends to combine both approaches (V2V, V2I) to obtain the desired result; a hybrid network is
much more efficient. This section explains a few vehicular routing protocols that exploit both the V2V and V2I forms of
communication. Vehicular networks are highly dynamic in nature, and this characteristic causes frequent topological changes
that affect routing and packet delivery ratio. In addition, the performance of vehicular routing protocols is susceptible to
vehicular density. Vehicular routing protocols show a significant performance variation under sparse and dense networks.
Given all the traffic-related factors, VANETs cannot deal with network partitioning. One solution is to deploy the access points
along the road to make vehicular communication more reliable and reduce unwanted delay in different vehicular applications
[85]. Unlike ad hoc and sensor networks, energy is not an issue because vehicles have a rechargeable energy source. Thus,
deploying communication infrastructure along the road increases the packet delivery ratio and decreases the delay. These
protocols can be categorized into static infrastructure-based and mobile infrastructure-based routing protocols. The following
protocols are infrastructure-based, as they depend on permanent infrastructure on their routing algorithms.
3.2.1 Static infrastructure-based routing protocols
As shown in Figure 26, the protocols in this category use RSUs in junctions and along the roads to route packets to
reachable vehicles within the transmission range [13], [86].
Figure 26 RSUs in static infrastructure-based communication [13]
Placing the fixed RSUs, which are linked to the backbone network in precise positions, is necessary for communication.
The number and distribution of RSUs depend on the communication protocol which has to be employed. For instance, certain
protocols require RSUs to be uniformly allocated all over the entire road network, whereas several others require RSUs only at
intersections, and others require RSUs only at region borders. One can assume that infrastructure prevails to a certain level and
vehicles have access to it occasionally.
The use of RSUs for VANET provides two prospective benefits. In the first case, the higher antenna height increases the
range and reliability of vehicular-to-infrastructure communications compared with IVCs. In addition, the deployed RSUs are
connected to a higher bandwidth and a more reliable backbone network to provide traffic authorities with centralized access
and to enable the configuration and maintenance of these units. The best-known static infrastructure-based routing protocols
areSADV, RAR, VGPR, IAGR, and MOVE [87-91].
1.Static Node-assisted Adaptive (SADV) Routing Protocol [87]. SADV is aimed at minimizing message delivery delay in
sparse networks and adjusts to varying traffic densities by enabling each node to estimate the amount of time for message
delivery. SADV assumes that every single vehicle is aware of its location through GPS and that each one has access to an
external static street map. SADV has three different modules:
(1) Static node assisted routing (SNAR),
(2) Link delay update (LDU), and
(3) Multipath data dissemination (MPDD).
SADV operates in road mode and intersection mode [87]. SNAR uses optimal paths which are determined based on a
graph abstracted from a road map. LDU maintains the delay matrix dynamically by measuring the delay of message delivery
between static nodes. MPDD assists in multipath routing.
The static node in SADV has a similar concept as in Throwbox [92], a device that can store and forward data in DTN
routing protocols, in that SADV can store and forward data when necessary. However, SADV is different from Throwback
because SADV does not require the node contact options, but needs both the implementation of static nodes and the routing
algorithm to utilize the street map structure and the vehicle densities of each road in vehicular networks.
The static node can store the packet for a certain period until the shortest delay path becomes available. As illustrated in
Figure 27a, a packet is sent from A to a remote location. Once the packet is transmitted from A to B, the latter has to determine
the next vehicle to forward the packet to. For example, let us assume that the shortest delay path to deliver the packet is
northward, but no vehicle within the communication range of B can transmit the packet along the direction. Therefore, B
transports the packet to the static node. The static node stores the packet for a certain period and sends it to C when it travels
through the junction and moves toward the northward road, as shown in right side of Figure 27b. Without support from the
static node, the packet will be moved by B to the eastward road. However, if B does not meet C at the junction, the result may
be a considerably longer packet delivery path.
Figure 27 SNAR in VANET [87]
2.Roadside-Aided Routing (RAR) [88]. In the proposed RAR protocol, a vehicle is affiliated with an area, called a sector,
which is bounded by RSUs (see Figure 28). To enter or leave a sector, a vehicle must pass by an RSU. Specifically, the
affiliation change between sectors happens in the radio coverage of RSUs. Therefore, modification of affiliation can be
recognized within a single hop of RSUs, thus preventing the multi-hop broadcasting of an agent’s advertisement.
Figure 28 Architecture of the RAR protocol [88]
Consequently, the overhead of affiliation is considerably minimized. A single-phase routing without hierarchical
addressing is proposed in RAR, assuming that VANETs and infrastructure networks are governed in various subnets. RSUs
receive route discovery requests and forward them to infrastructure networks or destination sectors as shortcuts because
addressing is regulated and all vehicles are affiliated with sectors. If the destination is in the same sector, direct ad hoc routing
is employed. The best route is discovered in a single phase without hierarchical addressing. Thus, overhead and delay are both
reduced.
3.Vertex-based Predictive Greedy Routing (VGPR) [89]. VGPR is a multi-hop vehicle-to-infrastructure routing protocol for
the city environment. It estimates a sequence of valid junctions from a source node to fixed infrastructure, and then transmits
the message to the fixed infrastructure through the sequence of junctions. It employs the position, velocity, and direction of
vehicles for computing both the sequence of valid junctions and greedy forwarding. While calculating a sequence of valid
junctions, a source node computes the shortest path between itself and its nearest fixed infrastructure with the help of a
navigation system such as GPS. If the source node obtains more than one route to the fixed infrastructure with an identical
number of junctions, then it randomly chooses one route from these junctions. It employs PDGR [39] to forward data from the
source node to the nearest fixed infrastructure. Each vehicle maintains a table which contains the ID, position, velocity, and
direction of its two-hop neighbors. The table is regularly upgraded by interchanging beacon messages among neighboring
vehicles. With the help of a table, the source node calculates the weighted score for itself, for the current packet carrier, and for
the two-hop neighbors. The source vehicle forwards the packet to the neighbor only if a neighbor rather than the existing
packet carrier has a higher score, or else the existing packet carries the packet until it discovers that its neighbor has a higher
weighted score than itself. VGPR has less overhead control, reduces packet retransmissions, increases the reliability of packet
delivery, and minimizes end-to-end delay.
4. Infrastructure-assisted Geo-Routing (IAGR) [90]. Borsetti and Gozalvezin 2010 proposed IAGR to improve the operation,
reliability, and performance of multi-hop geo-routing protocols by exploiting the reliable interconnection of RSUs. Even
though the infrastructure may add some latency, this protocol mainly focuses on traffic management and infotainment
applications in which the delay requirements are not as severe as those required in safety applications. IAGR uses GSR, which
is basically designed for V2V communication, to analyse the potential advantage of V2I over V2V communication. The
routing algorithm assumes the presence of digital maps and location servers. It modifies the existing network graph of GSR by
including RSUs. In the network graph, a node can represent either a roadside unit or a junction. Initially, and following a
traditional topology-aware approach, nodes are connected following the road map topology and the intersection anchor points.
The weights of the graph are then initially calculated following the metric employed by the considered topology-aware routing
protocol, such as the distance between two consecutive nodes in the case of the GSR protocol. Given that RSUs can be
considered to be interconnected through a reliable and high-bandwidth backbone network, the geographical distance among
infrastructure units should be neglected from the perspective of the geo-routing algorithm. To enclose this property into the
network graph, all nodes that represent an RSU can be merged into a unique graph node referred to as the backbone gate.
Consequently, vehicles will recognize all the RSUs as a unique graph node, and the shortest routes can be computed by using
this unique property that characterizes infrastructure nodes.
5. Motion Vector Routing Algorithm (MOVE) [91]. MOVE was proposed for vehicle-to-infrastructure VANET. It considers a
sparse network where the prediction has to be made in advance for rare opportunistic routing. It assumes that every vehicle
understands its own position and direction, and that the destination is a fixed globally known location. The current vehicular
node finds the closest distance between the vehicle and the message destination along its trajectory. The current vehicular node
periodically sends a Hello beacon. The RESPONSE message is sent by neighboring nodes to reveal themselves to the current
vehicular node. Considering the direction where the neighboring node is headed, the current vehicle establishes the short route
to the location along the trajectory of the neighboring vehicle, then the current vehicle decides to forward the message while
establishing the current distance of each vehicle from the destination. This algorithm has a higher data delivery rate for a sparse
network compared with greedy position-based routing and uses less system buffer space. Based on a performance evaluation,
Lebrun J. et al. identified that if routes are constant and uniform, greedy position-based routing can perform better than
MOVE.
3.2.2 Mobile infrastructure routing protocols
RSUs minimize the end-to-end delay significantly, but again, the inherited problem of RSUs is the number needed to cover
an area and the cost associated with each RSU. Such cost includes the hardware, installation, operational, and maintenance
costs. Another issue is that protocols based on fixed RSUs can only provide connectivity in areas where they have been
deployed. Areas where access points are not installed are out of range, and thus, information cannot be collected or provided
[85]. Mobile infrastructure-based routing protocols overcome the restriction of fixed RSUs. Mobile infrastructure routing
protocols exploit the concept of mobile gateways where RSUs are replaced with mobile vehicles that function as mobile
gateways. The following protocols are mobile infrastructure-based MIBR, MGRP, and PBR [93-95], which depend on mobile
infrastructure for their routing algorithms.
1. Mobile Infrastructure-based VANET Routing (MIBR) [95]. MIBR is a position-based reactive routing protocol in which
buses are used as an essential factor during route selection and data transfer. The quality of transmission for every single road
segment and the various transmission capabilities of different vehicles are considered in the design of the protocol. The design
estimates the density of every road segment based on bus line information. In MIBR, the source node employs a GPS system to
obtain the destination information. Every single bus includes two assorted wireless interfaces, and a single interface comprises
other vehicles. During routing, the protocol approximates the next road segment and the hop counts, and stores the information
in a route table. When the packet is near a junction, the subsequent road segment is selected. This process consumes less
bandwidth. Figure 29 shows the selection mechanism of the route where the source vehicle S selects the road segment one after
another by considering both the hop count and the distance to the destination. The striped road segment represents the selected
segment. In a dense network (i.e., with a density of buses), the packet is forwarded from one road junction to another by using
buses; in a sparse network, both vehicles (i.e., buses and cars) work together to carry out the forwarding process. Typically,
buses take a higher priority during the forwarding phase of MIBR, and thus, the forwarding strategy is called “Bus First.” In
the Bus First strategy, the forwarding vehicle first tries to select a bus as the next forwarder, and if no bus is available in the
neighbor table of the forwarding vehicle, it then selects a car.
Figure 29 Route selection in MIBR
2. Mobile Gateway Routing Protocol (MGRP) [94]. Pan et al. in 2010recommended MGRP to increase the packet delivery
ratio and decrease the average hop count by exploit both inter-vehicle-based and infrastructure-based communication to route
packets. Like other position-based routing protocols, MGRP assumes the presence of a GPS and a digital map so that each
vehicle builds its neighbor table (including neighboring vehicles, directions, and speeds) that would assist in routing.
Furthermore, digital maps indicate the traffic load condition of roads. The fixed RSUs are replaced with mobile gateways to
provide connectivity in a considerably larger region. Mobile gateways are equipped with two interfaces: IEEE 802.11(IVC)
and 3G interfaces (vehicle-to-infrastructure communication). MGRP is based on the concept of mobile gateways proposed in
MIBR, which utilizes buses as a mobile gateway with a fixed route. However, their connectivity is limited by their scheduling
time and within the region covered by the bus routes. Unlike MIBR, it uses vehicles such as taxis as mobile gateways. The
IEEE 802.11 interface is used for IVC with nearby vehicles that do not have a 3G interface or vehicles that are not mobile
gateways. Figure 30 illustrates the basic architecture of MGRP.
Figure 30 Mobile gateway architecture of MGRP [85]
Upon receiving packets from the IEEE 802.11 interface, mobile gateways forward the packet to the base station via the 3G
interface. In turn, the base station forwards the packets to the gateway controller. The gateway controller finds the position of
the destination vehicle and forwards the packet to each of the mobile gateways that are closest to the destination vehicle via the
base station. Upon receiving a packet from the gateway controller, mobile gateways forward the packet to the destination
vehicle by using the IEEE 802.11 interface.
3. Prediction-based Routing (PBR) [93]. PBR is focused on providing Internet connectivity to vehicles. Constructing an
infrastructure of the roadside static gateways is expensive, particularly outside city areas. Alternatively, PBR explores the
probability of mobile gateways with wireless WAN connections, which function as Internet gateways for other vehicles,
particularly focusing on highway scenarios.
The PBR algorithm assumes that every single vehicle is aware of its individual location via GPS or other services. The
protocol benefits from the significantly less inconsistency in vehicle movement patterns on highways to predict the duration
and expiration of a route from a client vehicle to a mobile gateway vehicle. Before a route malfunction is predicted, PBR pre-
emptively looks for a new direction to prevent loss of service. To communicate to a location on the Internet, a node checks its
routing table for a current route. In accordance with reactive routing protocols, if the node finds no existing route, the node
broadcasts a RREQ message with a minimum number of hops. When the RREQ reaches a mobile gateway, a RREP message is
sent back to the source node by the sequence of nodes stored in the RREQ. Given the short lifetime of vehicular routes,
intermediate nodes with cached gateway routes do not send RREP messages. If numerous gateways are found, the source node
selects the route with the shortest number of hops, where most of the hops move in the same direction as the source node. If
multiple routes to the same gateway are identified, then the route with the longest predicted lifetime is selected. As soon as the
route to the gateway is recognized, communication is initiated, and PBR recognizes a new route to a gateway if the route is
predicted to fail.
The RREP message is used to predict the lifetime of the route. When the gateway receives the RREQ message, it writes its
position, velocity information, and a maximum lifetime value in the RREP message that it returns to the source node. As each
intermediate node deals with the RREP message, which predicts the lifetime of the link between itself and its predecessor
depending on the wireless communication range, direction of travel, velocities of nodes, and distance between the nodes. If
that lifetime is less than that saved in the RREP message, the message is later updated with a shorter lifetime. When the source
node receives the RREP message, it consists of the projected lifetime of the route. Regardless of the efficiency of the PBR
algorithm in the mobile gateway scenario, it is unclear how realistic the situation is. The issue of how a vehicle could be
encouraged to reveal its wireless Internet connection with others when that connection is probably expensive is ambiguous,
Comment [DR1]: Added
although the incorporation of micropayments might be an interesting field of research. Internet providers charging for the use
of their roaming WAN-connected vehicles might be financially feasible, but additional evaluation might be needed to
determine its feasibility. Furthermore, the mobile gateway’s wireless WAN connections must have sufficient bandwidths to
support the demand of numerous client vehicles.
4 VANET mobility (building) models
Based on the review of various routing protocols, the authentic mobility model and decision parameters of nodes to send
the packets to other nodes are the main factors in developing an appropriate routing protocol to VANET. As suggested by
various studies of routing protocols [45] [31] [21] [44] establishing a realistic mobility model involves considering constraints,
including street map structure, vehicle density and speed, vehicular movements and inter-vehicle behavior in urban or
topographical circumstances, and obstructions such as architectural structures and trees. These constraints are divided into two
parts that are dealt with separately [96], as follows:
(1) Macroscopic, including node mobility constraints such as streets, lights, roads, and buildings.
(2) Microscopic, including vehicle movement constraints and their behaviors.
The mobility model can also be regarded as traffic and a motion generator. Motion constraints are influenced by car driver
habits, cars, and pedestrians, and describe each vehicle movement. The traffic generator creates random topologies from maps
and defines the vehicular behavior under a particular environment. A mobility model is also defined by a framework that
includes topological maps such as lanes, roads, streets, obstacles in mobility, communication model, and car velocities based
on traffic densities related to how the simulation time could be varied as well as vehicular distribution on roads and intelligent
driving patterns. This framework is illustrated in Figure 31.
Figure 31 VANET mobility model framework [97]
To create a real-world simulation, a mobility model must be generated. One way to capture a realistic model is to generate
patterns from mobility traces. Various models can generate mobility patterns based on certain criteria. The following are the
most popular models that generate mobility traces.
4.1 Survey models
Realistic human behavior in urban mesh environments is represented by survey models. Data gathered from surveys are
used in these models. The survey conducted by the U.S. Department of Labor is considered as one of the largest surveys [98]
that recorded the behavior and activities of workers at lunch time, their mode of communication among themselves and with
other people, and their lunch and break times. These data were later used to develop a generic mobility model. The UDel
mobility model [99] is a tool that simulates urban mesh networks. This model consists of impediments in mobile nodes and
produces a graph of the urban area. The mobile nodes and their behavior are then placed on the graph.
4.2 Event-driven models
Event-driven models are also known as trace models. Such models facilitate the analysis of the movement of human beings
and vehicles, assess such movements, and produce traces depending on their mobility. In [100], the authors proposed a WLAN
mobility model which assesses the traffic characteristics of WLAN users throughout a campus. In [97], the authors identified
how WLAN users connect to the infrastructure network. Event-driven models might be compiled to build a probabilistic
mobility model that demonstrates actual movement on the map. This probabilistic mobility model facilitates the construction of
a distinct occurrence Markov chain, which considers the source, destination paths, and the present and previous locations.
However, the limitation of this model is that it only recognizes the features of mobile nodes with access points and does not
consider the association between the nodes. Consequently, probabilistic models cannot assist the ad hoc mode of VANET.
4.3 Software-oriented models
Simulators such as VisSim [101] and CORSIM [102] can generate the traces of urban microscopic traffic. VanetMobiSim
[103] employs the topologically integrated geographic encoding and referencing (TIGER) database [104] and Voronoi graphs
[105] to acquire road topologies, maps, streets, and other details for the network simulators. These simulators have
disadvantages in that they can only function at the traffic level and cannot produce realistic details. Furthermore, the
interoperability with other network simulators and the produced level of particulars are insufficient for other network
simulators.
4.4 Synthetic models
Extensive research has been conducted in the field of synthetic modeling. All models in this class employ mathematical
equations to build realistic mobility models. The efficiency of mathematical models is verified by comparing them with real
mobility models. According to [106], synthetic models can be divided into five main categories:
• Stochastic model, which deals with totally random motion;
• Traffic stream model, which examines the mechanical properties of the mobility model;
• Car following model, which monitors the behavior of car-to-car interaction;
• Queue model, which considers cars as standing in queues and roads as queue buffers; and
• Behavioral models, which examine how movement is influenced by social interaction.
For instance, the movement of a mobile node in an area can proceed either in a permanent line or in an arbitrary path. The
weighted way point where the detonation is selected based on the present location and time, and the random way point where
the detonation is selected arbitrarily via mobility algorithms computes the routine of the mobility of a node by determining a
number of mathematical equations. The synthetic model enforces specific restrictions, such as eliminating a real human
behavioral model. Consequently, developing random topologies by using this model is difficult.
5 Comparison and discussion of issues related to VANET routing protocols
Designing an efficient routing protocol that can deliver a packet within the shortest time and with few dropped packets is a
critical challenge in VANET technology given the high mobility of nodes and rapid topological changes. Several researchers
have focused on designing a routing protocol suitable for highly dense environments in which numerous vehicles have close
distances between them. Generally, the process of designing an efficient routing protocol helps improve several factors. The
first is enhancing system reliability by leveraging the packet delivery percentage. The second factor is reducing the extent of
interference caused by high buildings in a city environment. The third factor is considering scalability, which is necessary to
prevent conflict when a simultaneous operation of routing requests has been initiated. The fourth factor, which is crucial [107]
[70], is delivering a packet within the shortest possible time, particularly during an emergency situation.
In this paper, we present a comprehensive investigation of VANET routing protocols. A number of these protocols intend
to provide V2V services, and others focus on V2I communication. As indicated in the criteria presented in Table 3, the
VANET protocol is categorized based on the various strategies used by each protocol. From the study and comparison among
the characteristics of the listed VANET routing protocols, we can identify several relevant issues such as traffic awareness,
routing strategy, prediction, and network connectivity.
5.1 Traffic awareness
Traffic awareness refers to the ability of a protocol to use traffic information to select an efficient route. Certain routing
protocols such as A-STAR and VADD make probabilistic assumptions about traffic density by using static knowledge,
including information on bus routes and lanes. Other protocols, such as STAR and CAR, determine traffic density via real-time
measurement. These traffic-awareness routing protocols have employed various methods to calculate traffic density, but none
of them have described how they obtain information on actual vehicular traffic. The probabilistic assumptions used by A-
STAR and VADD do not provide accurate vehicle density because they do not consider real-time traffic flows. The real-time
measurements used in the STAR and CAR protocols have attempted to overcome this problem, but they still have a drawback
in that the connection (presence) between vehicles is extremely poor in sparse traffic flow. This condition results in inaccurate
calculation of road density. Thus, all choices made in accordance with the vehicular traffic density would be incorrect, which
would result in an invalid execution of routing protocols. To address this inconvenience, we propose two solutions. The first is
to employ RSUs along the road to ensure regular updates on the traffic density; this method is easy but costly. The other
solution is to develop a new traffic-density estimating algorithm based on the dynamic transmission range concept [108] by
considering real-time traffic flow in dense and sparse conditions.
5.2 Routing strategy
Through intensive investigation, we find two main strategies that can be applied to several routing protocols: forwarding
and buffering (carry-and-forward). The forwarding strategy describes the first routing decision of the protocol when certain
packets have to be forwarded. To provide greater distribution in sparse networks and enhance the reliability of the system, a
buffering strategy is commonly employed. In this strategy, a node may maintain a packet in a local buffer until a forwarding
prospect is available rather than merely dropping the packet. The buffering strategy in VANET routing protocols such as
VADD and SKVR improves the packet delivery ratio. In turn, the end-to-end delay increases, which is a critical factor in
ensuring the functions of emergency applications. Therefore, these protocols require further improvement to decrease end-to-
end delays. One solution is to increase the transmission power of vehicles dynamically in areas with sparse traffic. This
approach may help raise the probability of vehicle connection and reduce packet buffering time.
5.3 Prediction
We have also identified a number of predictive protocols that assume the characteristics of mobile vehicles in a VANET
environment involved in a routing decision. Current locations of vehicles based on their last known position and velocity are
predicted by using algorithms such as MURU, PDGR, and GyTAR. However, other algorithms, such as RAODV, employ the
same information to predict the stability or estimated lifetime of a route. Given that such predictive protocols involve the
characteristics of mobile vehicles in their routing decisions, adopting the behavior of a driver in the prediction algorithms may
be appropriate to ensure more efficient and realistic predictions. Driver behavior has a significant effect on mobility patterns in
both near and long terms. In the near term, vehicle movements vary significantly based on individual lane changing, braking,
and passing behaviors. In the long term, mobility is affected by variations in the intended destination of a driver. Thus, the
development of new algorithms that can predict driver intentions may contribute to accurate predictions of an efficient route.
5.4 Network connectivity
In analyzing several routing protocols by considering all traffic-related factors and various techniques used in V2V routing
protocols, VANETs cannot cope with network segmentation because of frequently disconnected networks in sparse
environments. Several solutions have been proposed in the literature, such as integrating the VANET with other infrastructure
networks, for example, cellular networks, WLAN, and the Internet. V2I communication can expand the network coverage area
and offer certain services that the VANET does not provide. This method of integration makes vehicular communication more
reliable and reduces unwanted delays in different vehicular applications.
Two different virtual infrastructures are discovered in this comparison, namely, static RSUs and mobile infrastructure.
These infrastructures have been employed to connect seamlessly through V2I communication. Protocols based on static RSUs,
such as VGPR, MOVE, and RAR, can provide connectivity only in areas where they are deployed. Thus, the distribution and
the requirement of static RSUs is the main drawback of these protocols. To solve this problem, the mobile gateway concept has
been exploited in which RSUs are replaced with mobile vehicles that function as mobile gateways. In MIBR, buses are
employed as mobile gateways that have fixed routes. The problem with MIBR is limited connectivity to the region covered by
the bus routes. MGRP uses taxis, which do not have a fixed route, as mobile gateways. The IEEE 802.11 interface is used in
MGRP to enable vehicles to communicate with one another, and the 3G interface is installed in mobile gateways (taxis) to
enable them to communicate with the base station. Indeed, the mobile gateway selection, such as taxis in the MGRP routing
protocol, is not efficient because the routing decision is restricted by the presence of such vehicles in the network. A more
efficient approach is selecting the mobile gateway dynamically based on certain factors, such as route stability with both the
neighbors and the roadside base station. The PBR routing protocol explores the probability of mobile gateways with wireless
WAN connections, which function as Internet gateways for other vehicles. This protocol uses the link lifetime between the
source vehicle and the mobile gateway vehicle to select the optimal mobile gateway. The optimal Internet gateway selection in
PBR does not consider quality of service (QoS) metrics that define a guarantee provided by the network to satisfy a set of
predetermined service performance constraints for the user in terms of end-to-end delay, jitter, and available bandwidth.
To obtain an optimal connection anytime and anywhere, certain critical factors should be considered in designing future
V2I routing protocols. The main environment of V2I communication should be divided into two main parts: indoor and
outdoor. The indoor environment corresponds to the communication environment within the transmission range of the roadside
base station. The outdoor environment corresponds to the communication environment in which the vehicles are located and
out of the coverage of the roadside base station. With regard to the indoor environment, two main issues should be addressed in
the design of V2I routing protocols.
The first issue is the optimality of mobile gateway selection, which involves numerous factors, to select an optimal mobile
gateway for other vehicles. In other words, special VANET nodes exist, namely, roadside base station units that are static
nodes, which may provide Internet access to the VANET. These nodes may have access to a complete view of the network
topology and to additional information that may improve not only routing but also network topology construction. The second
issue is the handover process in an indoor environment. With regard to the outdoor environment, continuous connection with
the indoor environment should be ensured by deploying the most effective inter-vehicle routing strategies for this case.
Currently, we are developing a solution based on V2I communication by specifying the number of mobile gateways based
on the real traffic density. In this solution, the QoS metrics are considered in optimal gateway selection to meet the
requirements of various VANET applications.
Table 3 summarizes the qualitative comparison of various VANET routing protocols based on the following points:
(1) Characteristic parameters such as traffic awareness, forwarding strategy, buffering (carry-and-forward) strategy,
overlay or non-overlay routing, and predictive routing;
(2) Requirements to be met by the routing strategy, such as virtual infrastructure, location services, and digital map; and
(3) The most suitable communication environments such as cities and highways.
Most VANET routing protocols are new and most of them are not yet applied to network simulators. Thus, to make our
study more useful for researchers, Table 4 presents all previous VANET routing protocols, the simulators employed to evaluate
each of them, and the availability of these protocols in common network simulators.
Table 3 Comparison among various VANET routing protocols
VANET Routing Protocols
Characteristic parameters
Routing requirements
Communication
environments
Tra
ffic
-aw
aren
ess
Forw
ardi
ng
stra
tegy
Buf
feri
ng
(c
arry
and
fo
rwar
d)
stra
tegy
Ove
rlay
or
Non
-ove
rlay
Pred
ictiv
e
Vir
tual
in
fras
truc
ture
-req
uire
d
Map
-bas
ed
requ
ired
Posi
tioni
ng
syst
em-
requ
ired
Loc
atio
n se
rvic
es
requ
ired
DSDV, GSRP, FSR, OLSR, WRP, TBRPF, ZRP, HARP No Multi-
hop No No No No No No No City
TORA, AODV, DSR, AODV+PGB No Multi-
hop Yes No No No No No No City
PRAODV No Multi-hop Yes No Yes No No No No City
GPSR No Greedy No No No No No Yes Yes Highway GPSR+AGF No Greedy No No No No No Yes Yes City GSR No Greedy No Yes No No Yes Yes Yes City SAR No Greedy No No No No Yes Yes Yes City A-STAR, STAR Yes Greedy No Yes No No Yes Yes Yes City MURU No Greedy No No Yes No Yes Yes Yes City GPCR No Greedy No Yes No No Yes Yes Yes City GpsrJ+ No Greedy No Yes Yes No No Yes Yes City GPGR No Greedy No No Yes No Yes Yes Yes City PBR-DV No Greedy No No No No No Yes Yes City CAR Yes Greedy No Yes No No Yes Yes Yes City GyTAR, JARR, LOUVRE Yes Greedy No Yes Yes No Yes Yes Yes City DIR,ROMSGP, AMAR, EBGR, B-MFR No Greedy No No No No Yes Yes Yes City
TO-GO Yes Greedy No No Yes No Yes Yes Yes City CBF No Greedy No No No No No No Yes City SKVR No Greedy Yes No No No No No No City VADD Yes Greedy Yes No Yes No Yes Yes Yes City
GeOpps, GeoDTN+Nav, LORA-CBF No Greedy Yes No No No Yes No Yes City
CBR, CBDRP, COIN, TIBCRPH No Multi-
hop Yes No No No Yes Yes No City
IVG No Multi-hop No No No No No Yes No Highway
CGR, AGR, ROVER, Mobicast No Multi-hop No No No No No Yes No City
MAV-AODV No Multi-hop No No No No No Yes Yes City
MAODV, ADMR, MOLSR, ODMRP, D-ODMRP No Multi-
hop No No No No No No No City
BROADCOM, UMB, DV-CAST, EAEP, HyDi, DECA. No Multi-
hop Yes No No No No No No Highway
SADV, RAR, IAGR Yes Store & forward No No No Static
RSUs Yes Yes Yes City
VGPR, MOVE, No Store & forward No No Yes Static
RSUs No Yes No City
MIBR, MGRP Yes Store & forward
No No Yes
Mobileinfrastructure
Yes Yes Yes City
PBR No Store & forward No No Yes
Mobileinfrastructure
Yes Yes Yes Highway
Table 4 Availability and Evaluation of VANET routing protocols
VANET routing
protocols
Simulators VANET routing protocols Simulators
ADMR NS2* [109] HyDi OMNeT++* AGR N/A¥ IAGR NS2* AMAR NS2* IVG GloMoSim*
AODV
OPNET† [110], OMNeT++* [111] PARSIC* [112] NS2†, NCTUns†
[113], GloMoSim† [114], QualNet† [115], EstiNet† [116], NCTUns
JARR NS2*
AODV-PGB NS2* LOUVRE QualNet* A-STAR NS2* LORA-CBF OPNET* B-MFR NS2* MAODV PARSIC* BROADCOMM N/A¥ MAV-AODV NS2* CAR NS2* MGRP NS2* CBDRP NS2* MIBR NS2* CBF NS2* Mobicast NCTUns* CBR N/A¥ MOLSR N/A¥ CGR NS2* MOVE QualNet* COIN Own± MURU NS2* DECA TraNS* [99] ODMRP GloMoSim*, QualNet† DIR NCTUns* OLSR OPNET†, QualNet† D-ODMRP GloMoSim* PBR Own± DSDV NCTUns†, NS2†, GloMoSim† PBR-DV Own±
DSR Own±,NS2†, QualNet†, NCTUns†, GloMoSim†, OPNET†, OMNeT++†
PDGR NS2*
DV-CAST Own± RAR NS2* EAEP JAVA ROMSGP N/A¥ EBGR NCTUns* ROVER JiST/SWANS* [117] FSR GloMoSim* SADV Own± GeOpps OMNeT++* SAR NS2* GeOpps+Nav QualNet* SKVR NS2*
GPCR NS2* STAR NS2, QualNet† GPGR NS2* TBRPF N/A¥ GPSR NS2* TIBCRPH NS2* TO-GO QualNet* GpsrJ+ QualNet* TORA NS2† ,OPNET† GPSR+AGF NS2* UMB Own± GSR NS2* VADD NS2* GSRP C++ VGPR NS2* GYTAR QualNet* WRP Drama* [118] HARP N/A¥ ZRP QualNet†, GloMoSim† *: The authors use the common network simulator package to implement and evaluate the protocol. †: The protocol implementation is available in the common network simulator package. N/A¥: No simulation implementation and evaluation for the protocol. Own±: The authors have built their own simulator to implement and evaluate the protocol.
6 Conclusion and directions for future research
A. Conclusion
Vehicle communication technology has become crucial in designing vehicles for the future. VANET offers communication
services among vehicles or with roadside infrastructure. In this study, we discussed the prospective applications and the
problems involved in designing routing protocols in VANETs as well as surveyed and analyzed a large number of routing
protocols. We also proposed a taxonomy of protocols based on the VANET features and classified these protocols into two
main categories: (1) vehicle-to-vehicle-based routing protocols and (2) vehicle-to-infrastructure-based routing protocols. This
study discussed the characteristics, routing metrics and routing philosophies of each of these protocols selected from a class of
similar approaches, which can reflect state-of-the-art research on VANET routing protocols. The classification of the primary
routing selection principles can simplify the task of a network designer in deciding the VANET routing strategies to be adopted
in a given condition. We believe that our survey will be useful to the research community and will serve as suitable
introductory material for individuals who want to pursue the study and application of VANETs.
Through this extensive survey, we can conclude that the main distinguishing factor among the various VANET protocols is
the means of identifying and organizing routes between the source and destination pairs. A number of routing protocols have
been proposed to solve the most critical problems in VANET technology. Most of these protocols cannot address the highly
dynamic topology and frequently disconnected network, which are considered as major challenges. We highlighted certain
issues related to these protocols and proposed corresponding solutions. Generally, position-based routing and geo-casting are
more efficient than the other routing protocols for VANETs because of environmental limitations. Moreover, infrastructure-
based routing protocols are most promising for VANET communication.
B. Directions for future research
As stated, research on VANETs in general and on routing protocols in particular is not comprehensive. Numerous studies
have focused on developing VANET routing protocols to support effective and efficient communication between vehicles.
However, we found that several key challenges have not been resolved and therefore require extensive research. For example,
in terms of VANET communication environments, most of the routing protocols were proposed by considering city
environments. Such protocols have not functioned efficiently in highway environments, and vice versa. Moreover, VANETs
cannot cope with segmentation in sparse networks. To overcome such unique challenges, integrating VANET and
infrastructure networks should be considered further in designing VANET routing protocols. Moreover, the following
challenges require further examination in future research:
(1) Network disconnection: The disconnected network is a major problem for inter-vehicle routing protocols. The
infrastructure-based routing protocol can help solve this problem. Future VANET routing protocols should support
the coexistence and interoperability of heterogeneous wireless technologies with varying requirements. Handover
issues should also be addressed in the design of such protocols.
(2) QoS metrics: QoS metrics identify a guarantee provided by the network to meet a group of prearranged service
performance limitations for the user with regard to end-to-end delay, jitter, and available bandwidth. To design a
feasible routing protocol that is suitable for all VANET applications, we should improve adaptive QoS routing
approaches to ensure that these applications can install new routes rapidly and efficiently in case current routing paths
become inaccessible as a result of alterations in node velocity, node positioning, network topology, or distance
between vehicular nodes.
(3) Benchmarking protocol: Benchmarks for various VANET routing protocols should be proposed. We have observed
from all previous studies that no benchmarking protocol exists for evaluating these protocols. GPSR protocol is a
widely accepted benchmarking protocol for most of these studies, but it is no longer adequate. The benchmark should
not only include a standard routing protocol but also consider the simulation environment.
(4) Bandwidth limitations: Bandwidth limitations represent another key issue in VANET routing protocols, specifically
the absence of a central coordinator that controls the communications between nodes and which is responsible for
managing the bandwidth and contention operation. Therefore, the available bandwidth has to be used efficiently.
(5) Security: With regard to VANET applications, security is a critical routing issue because numerous applications may
influence life-or-death decisions, and illegal interference may have disastrous consequences. The features of VANETs
make secure routing a more significant challenge than in the case of other communication networks. Therefore,
further research is necessary to develop methods of protecting the network routing protocol itself from malicious or
compromised nodes.
(6) Scalability: Given the number of vehicles that could be incorporated into vehicular networks, the VANET may
become the largest type of ad hoc network in history. Therefore, scalability is a critical factor to be considered in
future research of VANET routing protocols.
(7) Efficient routing: Although numerous routing protocols for VANETs have been proposed, methods of effective
routing in partitioned networks and methods of addressing dynamic topology in VANETs both require further
examination.
(8) Driver behavior: Another challenge related to inter-vehicle routing protocols is that of DTNs. The main approach
used by these routing protocols to deliver packets depends on carry-and-forward strategies, but these protocols should
also consider the driver behavior, which can significantly influence successful packet delivery.
(9) VANET applications: In order to expand the VANET applications to be encompassing the health-care of driver
inside the vehicles, the cooperation among inter-vehicular networks and sensor networks need to be further
investigated. Moreover, some of future trends need to be focusing on designing an efficient routing protocol by
considering the requirement of emergency vehicular health-care system.
Acknowledgments
This study was funded in part by the University Kebangsaan Malaysia under Grant Nos.UKM-GUP-2011-252 and
FRGS/1/2012/SG05/UKM/02/7.
References
[1] S. Misra, I. Woungang, and S. C. Misra, Guide to Wireless Ad Hoc Networks: Springer, 2009. [2] H. Hartenstein and K. P. Laberteaux, "A tutorial survey on vehicular ad hoc networks," Communications Magazine,
IEEE, vol. 46, pp. 164-171, 2008. [3] I. F. Akyildiz, D. M. Gutierrez-Estevez, and E. C. Reyes, "The evolution to 4G cellular systems: LTE-Advanced,"
Physical Communication, vol. 3, pp. 217-244, 2010. [4] Z. T. Sharef, A. E. Alaradi, and B. T. Sharef, "Performance Evaluation for WiMAX 802.16 e OFDMA Physical Layer,"
in Computational Intelligence, Communication Systems and Networks (CICSyN), 2012 Fourth International Conference on, 2012, pp. 351-355.
[5] V. Namboodiri, M. Agarwal, and L. Gao, "A study on the feasibility of mobile gateways for vehicular ad-hoc networks," in Proceedings of the 1st ACM international workshop on Vehicular ad hoc networks, 2004, pp. 66-75.
[6] U. Lee, B. Zhou, M. Gerla, E. Magistretti, P. Bellavista, and A. Corradi, "Mobeyes: smart mobs for urban monitoring with a vehicular sensor network," Wireless Communications, IEEE, vol. 13, pp. 52-57, 2006.
[7] H. Hartenstein and K. Laberteaux, VANET: vehicular applications and inter-networking technologies vol. 1: Wiley Online Library, 2010.
[8] F. Li and Y. Wang, "Routing in vehicular ad hoc networks: A survey," Vehicular Technology Magazine, IEEE, vol. 2, pp. 12-22, 2007.
[9] R. A. A. Baraa T. Sharef, Mahamod Ismail and Sardar Muhammad Bilal, "A Comparison of Various Vehicular ad hoc Routing Protocols Based on Communication Environments," The 7th International Conference on Ubiquitous Information Management and Communication, pp. Kota Kinabalu, Malaysia 17-19 January 2013.
[10] D. Antolino Rivas, J. M. Barceló-Ordinas, M. Guerrero Zapata, and J. D. Morillo-Pozo, "Security on VANETs: Privacy, misbehaving nodes, false information and secure data aggregation," Journal of Network and Computer Applications, vol. 34, pp. p. 1942-1955, 2011.
[11] E. Fonseca and A. Festag, "A survey of existing approaches for secure ad hoc routing and their applicability to VANETS," NEC network laboratories, 2006.
[12] G. Korkmaz, E. Ekici, F. Özgüner, and Ü. Özgüner, "Urban multi-hop broadcast protocol for inter-vehicle communication systems," in Proceedings of the 1st ACM international workshop on Vehicular ad hoc networks, 2004, pp. 76-85.
[13] S. Zeadally, R. Hunt, Y. S. Chen, A. Irwin, and A. Hassan, "Vehicular ad hoc networks (VANETs): status, results, and challenges," Telecommunication Systems, pp. 1-25, 2010.
[14] R. A. Alsaqour, M. S. Abdelhaq, and O. A. Alsukour, "Effect of network parameters on neighbor wireless link breaks in GPSR protocol and enhancement using mobility prediction model," EURASIP Journal on Wireless Communications and Networking, vol. 2012, p. 171, 2012.
[15] A. Fonseca and T. Vazão, "Applicability of position-based routing for vanet in highways and urban environment," Journal of Network and Computer Applications, vol. 36, pp. 961–973, 2013.
[16] F. B. Zhan and C. E. Noon, "Shortest path algorithms: an evaluation using real road networks," Transportation Science, vol. 32, pp. 65-73, 1998.
[17] C. E. Perkins and P. Bhagwat, "Highly dynamic destination-sequenced distance-vector routing (DSDV) for mobile computers," ACM SIGCOMM Computer Communication Review, vol. 24, pp. 234-244, 1994.
[18] T. Clausen, G. Hansen, L. Christensen, and G. Behrmann, "The optimized link state routing protocol, evaluation through experiments and simulation," in IEEE Symposium on" Wireless Personal Mobile Communications, 2001.
[19] T. W. Chen and M. Gerla, "Global state routing: A new routing scheme for ad-hoc wireless networks," in Communications, 1998. ICC 98. Conference Record. 1998 IEEE International Conference on, 1998, pp. 171-175.
[20] G. Pei, M. Gerla, and T. W. Chen, "Fisheye state routing: A routing scheme for ad hoc wireless networks," in Communications, 2000. ICC 2000. 2000 IEEE International Conference on, 2000, pp. 70-74.
[21] S. Murthy and J. J. Garcia-Luna-Aceves, "An efficient routing protocol for wireless networks," Mobile Networks and Applications, vol. 1, pp. 183-197, 1996.
[22] R. Ogier, F. Templin, and M. Lewis, "Topology dissemination based on reverse-path forwarding (TBRPF)," RFC Editor2004.
[23] T. H. Cormen, C. E. Leiserson, R. L. Rivest, and C. Stein, Introduction to algorithms: MIT press, 2001. [24] V. D. Park and M. S. Corson, "A highly adaptive distributed routing algorithm for mobile wireless networks," in
INFOCOM'97. Sixteenth Annual Joint Conference of the IEEE Computer and Communications Societies. Proceedings IEEE, 1997, pp. 1405-1413.
[25] C. E. Perkins and E. M. Royer, "Ad-hoc on-demand distance vector routing," in Mobile Computing Systems and Applications, 1999. Proceedings. WMCSA'99. Second IEEE Workshop on, 1999, pp. 90-100.
[26] D. B. Johnson and D. A. Maltz, "Dynamic source routing in ad hoc wireless networks," Mobile computing, pp. 153-181, 1996.
[27] V. Naumov, R. Baumann, and T. Gross, "An evaluation of inter-vehicle ad hoc networks based on realistic vehicular traces," in Proceedings of the 7th ACM international symposium on Mobile ad hoc networking and computing, 2006, pp. 108-119.
[28] Z. J. Haas, "A new routing protocol for the reconfigurable wireless networks," in Universal Personal Communications Record, 1997. Conference Record., 1997 IEEE 6th International Conference on, 1997, pp. 562- 566.
[29] N. Nikaein and C. Bonnet, "Harp-hybrid ad hoc routing protocol," in Proceedings of International Symposium on Telecommunications (IST), 2001, pp. 56-67.
[30] S. Yousefi, M. Fathy, and A. Benslimane, "Performance of beacon safety message dissemination in Vehicular Ad hoc NETworks (VANETs)," Journal of Zhejiang University-Science A, vol. 8, pp. 1990-2004, 2007.
[31] B. Karp and H. T. Kung, "GPSR: Greedy perimeter stateless routing for wireless networks," in Proceedings of the 6th annual international conference on Mobile computing and networking, 2000, pp. 243-254.
[32] C. Lochert, H. Hartenstein, J. Tian, H. Fussler, D. Hermann, and M. Mauve, "A routing strategy for vehicular ad hoc networks in city environments," in Intelligent Vehicles Symposium, 2003. Proceedings. IEEE, 2003, pp. 156-161.
[33] T. Camp, J. Boleng, and L. Wilcox, "Location information services in mobile ad hoc networks," in Communications, 2002. ICC 2002. IEEE International Conference on, 2002, pp. 3318-3324.
[34] D. Bertsekas, "Dynamic behavior of shortest path routing algorithms for communication networks," Automatic Control, IEEE Transactions on, vol. 27, pp. 60-74, 1982.
[35] J. Tian, L. Han, and K. Rothermel, "Spatially aware packet routing for mobile ad hoc inter-vehicle radio networks," in Intelligent Transportation Systems, 2003. Proceedings. 2003 IEEE, 2003, pp. 1546-1551.
[36] B. C. Seet, G. Liu, B. S. Lee, C. H. Foh, K. J. Wong, and K. K. Lee, "A-STAR: A mobile ad hoc routing strategy for metropolis vehicular communications," NETWORKING 2004. Networking Technologies, Services, and Protocols; Performance of Computer and Communication Networks; Mobile and Wireless Communications, pp. 989-999, 2004.
[37] F. Giudici and E. Pagani, "Spatial and traffic-aware routing (star) for vehicular systems," High Performance Computing and Communications, pp. 77-86, 2005.
[38] Z. Mo, H. Zhu, K. Makki, and N. Pissinou, "MURU: A multi-hop routing protocol for urban vehicular ad hoc networks," in Mobile and Ubiquitous Systems: Networking & Services, 2006 Third Annual International Conference on, 2006, pp. 1-8.
[39] J. Gong, C. Z. Xu, and J. Holle, "Predictive directional greedy routing in vehicular ad hoc networks," in Distributed Computing Systems Workshops, 2007. ICDCSW'07. 27th International Conference on, 2007, pp. 2-2.
[40] C. Lochert, M. Mauve, H. Füßler, and H. Hartenstein, "Geographic routing in city scenarios," ACM SIGMOBILE Mobile Computing and Communications Review, vol. 9, pp. 69-72, 2005.
[41] K. C. Lee, J. Härri, U. Lee, and M. Gerla, "Enhanced perimeter routing for geographic forwarding protocols in urban vehicular scenarios," in Globecom Workshops, 2007 IEEE, 2007, pp. 1-10.
[42] S. H. Cha, K. W. Lee, and H. S. Cho, "Grid-Based Predictive Geographical Routing for Inter-Vehicle Communication in Urban Areas," International Journal of Distributed Sensor Networks, vol. 2012, 2012.
[43] K. C. Lee, U. Lee, and M. Gerla, "Survey of routing protocols in vehicular ad hoc networks," Advances in Vehicular Ad-Hoc Networks: Developments and Challenges, IGI Global, vol. 21, 2009.
[44] V. Naumov and T. R. Gross, "Connectivity-aware routing (CAR) in vehicular ad-hoc networks," in INFOCOM 2007. 26th IEEE International Conference on Computer Communications. IEEE, 2007, pp. 1919-1927.
[45] M. Jerbi, S. M. Senouci, R. Meraihi, and Y. Ghamri-Doudane, "An improved vehicular ad hoc routing protocol for city environments," in Communications, 2007. ICC'07. IEEE International Conference on, 2007, pp. 3972-3979.
[46] K. C. Lee, M. Le, J. Harri, and M. Gerla, "Louvre: Landmark overlays for urban vehicular routing environments," in Vehicular Technology Conference, 2008. VTC 2008-Fall. IEEE 68th, 2008, pp. 1-5.
[47] Y. S. Chen, Y. W. Lin, and C. Y. Pan, "DIR: diagonal-intersection-based routing protocol for vehicular ad hoc networks," Telecommunication Systems, vol. 46, pp. 299-316, 2011.
[48] T. Taleb, E. Sakhaee, A. Jamalipour, K. Hashimoto, N. Kato, and Y. Nemoto, "A stable routing protocol to support ITS services in VANET networks," Vehicular Technology, IEEE Transactions on, vol. 56, pp. 3337-3347, 2007.
[49] C. Tee and A. Lee, "Adaptive reactive routing for VANET in city environments," in Pervasive Systems, Algorithms, and Networks (ISPAN), 2009 10th International Symposium on, 2009, pp. 610-614.
[50] K. Prasanth, K. Duraiswamy, K. Jayasudha, and C. Chandrasekar, "Minimizing end-to-end delay in vehicular ad hoc network using edge node based greedy routing," in Advanced Computing, 2009. ICAC 2009. First International Conference on, 2009, pp. 135-140.
[51] R. S. Raw and D. Lobiyal, "B-MFR routing protocol for vehicular ad hoc networks," in Networking and Information Technology (ICNIT), 2010 International Conference on, 2010, pp. 420-423.
[52] N. Brahmi, M. Boussedjra, J. Mouzna, and M. Bayart, "Adaptative movement aware routing for vehicular ad hoc networks," presented at the Proceedings of the 2009 International Conference on Wireless Communications and Mobile Computing: Connecting the World Wirelessly, Leipzig, Germany, 2009.
[53] K. C. Lee, U. Lee, and M. Gerla, "TO-GO: TOpology-assist geo-opportunistic routing in urban vehicular grids," in Wireless On-Demand Network Systems and Services, 2009. WONS 2009. Sixth International Conference on, 2009, pp. 11-18.
[54] H. Füßler, H. Hartenstein, J. Widmer, M. Mauve, and W. Effelsberg, "Contention-based forwarding for street scenarios," in 1st International Workshop in Intelligent Transportation (WIT 2004), 2004, pp. 155-159.
[55] S. Ahmed and S. S. Kanere, "SKVR: scalable knowledge-based routing architecture for public transport networks," in Proceedings of the 3rd international workshop on Vehicular ad hoc networks, 2006, pp. 92-93.
[56] J. Zhao and G. Cao, "VADD: Vehicle-Assisted Data Delivery in Vehicular Ad Hoc Networks," Vehicular Technology, IEEE Transactions on, vol. 57, pp. 1910-1922, 2008.
[57] I. Leontiadis and C. Mascolo, "Geopps: Geographical opportunistic routing for vehicular networks," in World of Wireless, Mobile and Multimedia Networks, 2007. WoWMoM 2007. IEEE International Symposium on a, 2007, pp. 1-6.
[58] P. C. Cheng, J. T. Weng, L. C. Tung, K. C. Lee, M. Gerla, and J. Haerri, "GeoDTN+ Nav: A hybrid geographic and DTN routing with navigation assistance in urban vehicular networks," MobiQuitous/ISVCS, 2008.
[59] Y. Luo, W. Zhang, and Y. Hu, "A new cluster based routing protocol for VANET," in Networks Security Wireless Communications and Trusted Computing (NSWCTC), 2010 Second International Conference on, 2010, pp. 176-180.
[60] T. Song, W. Xia, and L. Shen, "A cluster-based directional routing protocol in VANET," in Communication Technology (ICCT), 2010 12th IEEE International Conference on, 2010, pp. 1172-1175.
[61] R. Santos, A. Edwards, R. Edwards, and N. Seed, "Performance evaluation of routing protocols in vehicular ad-hoc networks," International Journal of Ad Hoc and Ubiquitous Computing, vol. 1, pp. 80-91, 2005.
[62] J. Blum, A. Eskandarian, and L. Hoffman, "Mobility management in IVC networks," in Intelligent Vehicles Symposium, 2003. Proceedings. IEEE, 2003, pp. 150-155.
[63] T. Wang and G. Wang, "TIBCRPH: Traffic Infrastructure Based Cluster Routing Protocol with Handoff in VANET," in Wireless and Optical Communications Conference (WOCC), 2010 19th Annual, 2010, pp. 1-5.
[64] A. Bachir and A. Benslimane, "A multicast protocol in ad hoc networks inter-vehicle geocast," in Vehicular Technology Conference, 2003. VTC 2003-Spring. The 57th IEEE Semiannual, 2003, pp. 2456-2460.
[65] C. Maihofer and R. Eberhardt, "Geocast in vehicular environments: caching and transmission range control for improved efficiency," in Intelligent Vehicles Symposium, 2004 IEEE, 2004, pp. 951-956.
[66] C. Maihöfer, T. Leinmüller, and E. Schoch, "Abiding geocast: time--stable geocast for ad hoc networks," in Proceedings of the 2nd ACM international workshop on Vehicular ad hoc networks, 2005, pp. 20-29.
[67] M. Kihl, M. Sichitiu, T. Ekeroth, and M. Rozenberg, "Reliable geographical multicast routing in vehicular ad-hoc networks," Wired/Wireless Internet Communications, pp. 315-325, 2007.
[68] J. Bronsted and L. M. Kristensen, "Specification and performance evaluation of two zone dissemination protocols for vehicular ad-hoc networks," in Proceedings of the 39th annual Symposium on Simulation, 2006, pp. 68-79.
[69] Y. S. Chen, Y. W. Lin, and S. L. Lee, "A mobicast routing protocol in vehicular ad-hoc networks," Mobile Networks and Applications, vol. 15, pp. 20-35, 2010.
[70] Y. W. Lin, Y.-S. Chen, and S.-L. Lee, "Routing protocols in vehicular ad hoc networks: A survey and future perspectives," Journal of Information Science and Engineering, vol. 26, pp. 913-932, 2010.
[71] L. Junhai, Y. Danxia, X. Liu, and F. Mingyu, "A survey of multicast routing protocols for mobile ad-hoc networks," Communications Surveys & Tutorials, IEEE, vol. 11, pp. 78-91, 2009.
[72] E. M. Royer and C. E. Perkins, "Multicast operation of the ad-hoc on-demand distance vector routing protocol," in Proceedings of the 5th annual ACM/IEEE international conference on Mobile computing and networking, 1999, pp. 207-218.
[73] J. G. Jetcheva and D. B. Johnson, "Adaptive demand-driven multicast routing in multi-hop wireless ad hoc networks," in Proceedings of the 2nd ACM international symposium on Mobile ad hoc networking & computing, 2001, pp. 33-44.
[74] A. B. Souza, J. Celestino, F. A. Xavier, F. D. Oliveira, A. Patel, and M. Latifi, "Stable multicast trees based on Ant Colony optimization for vehicular Ad Hoc networks," in Information Networking (ICOIN), 2013 International Conference on, 2013, pp. 101-106.
[75] A. Laouiti, P. Jacquet, P. Minet, L. Viennot, T. Clausen, and C. Adjih, "Multicast optimized link state routing," IETF Internet Draft, draft-ieft-manet-olsr-molsr-01.txt, 2003.
[76] S. J. Lee, W. Su, and M. Gerla, "On-demand multicast routing protocol in multihop wireless mobile networks," Mobile Networks and Applications, vol. 7, pp. 441-453, 2002.
[77] Y. Yan, K. Tian, K. Huang, B. Zhang, and J. Zheng, "D-ODMRP: a destination-driven on-demand multicast routing protocol for mobile ad hoc networks," Communications, IET, vol. 6, pp. 1025-1031, 2012.
[78] M. Dorigo, V. Maniezzo, and A. Colorni, "Ant system: optimization by a colony of cooperating agents," Systems, Man, and Cybernetics, Part B: Cybernetics, IEEE Transactions on, vol. 26, pp. 29-41, 1996.
[79] M. Durresi, A. Durresi, and L. Barolli, "Emergency broadcast protocol for inter-vehicle communications," in Parallel and Distributed Systems, 2005. Proceedings. 11th International Conference on, 2005, pp. 402-406.
[80] O. Tonguz, N. Wisitpongphan, F. Bai, P. Mudalige, and V. Sadekar, "Broadcasting in VANET," in 2007 Mobile Networking for Vehicular Environments, 2007, pp. 7-12.
[81] M. Nekovee and B. B. Bogason, "Reliable and effcient information dissemination in intermittently connected vehicular adhoc networks," in Vehicular Technology Conference, 2007. VTC2007-Spring. IEEE 65th, 2007, pp. 2486-2490.
[82] G. Maia, A. L. L. Aquino, A. Viana, A. Boukerche, and A. A. F. Loureiro, "HyDi: a hybrid data dissemination protocol for highway scenarios in vehicular ad hoc networks," in Proceedings of the second ACM international symposium on Design and analysis of intelligent vehicular networks and applications, 2012, pp. 115-122.
[83] N. N. Nakorn and K. Rojviboonchai, "DECA: Density-aware reliable broadcasting in vehicular ad hoc networks," in Electrical Engineering/Electronics Computer Telecommunications and Information Technology (ECTI-CON), 2010 International Conference on, 2010, pp. 598-602.
[84] W. L. M. A. C. M. a. P. L. P. S. h. s. i. o. The Institute of Electrical and Electronics Engineers (IEEE), ANSI/IEEE Std.802.11, 1999. (a.k.a. ISO/IEC 8802-11:1999(E)).
[85] S. M. Bilal, C. J. Bernardos, and C. Guerrero, "Position Based Routing in Vehicular Networks: A Survey," Journal of Network and Computer Applications, vol. 36, pp. 685–697, 2012.
[86] C. Campolo, H. Cozzetti, A. Molinaro, and R. Scopigno, "Augmenting Vehicle-to-Roadside connectivity in multi-channel vehicular Ad Hoc Networks," Journal of Network and Computer Applications, 2012. http://dx.doi.org/10.1016/j.jnca.2012.04.001.
[87] Y. Ding, C. Wang, and L. Xiao, "A static-node assisted adaptive routing protocol in vehicular networks," in Proceedings of the fourth ACM international workshop on Vehicular ad hoc networks, 2007, pp. 59-68.
[88] Y. Peng, Z. Abichar, and J. M. Chang, "Roadside-aided routing (RAR) in vehicular networks," in Communications, 2006. ICC'06. IEEE International Conference on, 2006, pp. 3602-3607.
[89] R. K. Shrestha, M. Sangman, C. Ilyong, and C. Dongmin, "Vertex-Based Multihop Vehicle-to-Infrastructure Routing for Vehicular Ad Hoc Networks," in System Sciences (HICSS), 2010 43rd Hawaii International Conference on, 2010, pp. 1-7.
[90] D. Borsetti and J. Gozalvez, "Infrastructure-assisted geo-routing for cooperative vehicular networks," in Vehicular Networking Conference (VNC), 2010 IEEE, 2010, pp. 255-262.
[91] J. LeBrun, C. N. Chuah, D. Ghosal, and M. Zhang, "Knowledge-based opportunistic forwarding in vehicular wireless ad hoc networks," in Vehicular Technology Conference, 2005. VTC 2005-Spring. 2005 IEEE 61st, 2005, pp. 2289-2293.
[92] W. Zhao, Y. Chen, M. Ammar, M. Corner, B. Levine, and E. Zegura, "Capacity enhancement using throwboxes in DTNs," in Mobile Adhoc and Sensor Systems (MASS), 2006 IEEE International Conference on, 2006, pp. 31-40.
[93] V. Namboodiri and L. Gao, "Prediction-based routing for vehicular ad hoc networks," IEEE Transactions on Vehicular Technology, vol. 56, pp. 2332-2345, 2007.
[94] H. Y. Pan, R. H. Jan, A. A. K. Jeng, C. Chen, and H. R. Tseng, "Mobile-Gateway Routing for Vehicular Networks," IEEE VTSI APWCS, 2010.
[95] J. Luo, X. Gu, T. Zhao, and W. Yan, "A mobile infrastructure based VANET routing protocol in the urban environment," in Communications and Mobile Computing (CMC), 2010 International Conference on, 2010, pp. 432-437.
[96] M. Fiore, J. Harri, F. Filali, and C. Bonnet, "Vehicular mobility simulation for VANETs," in Simulation Symposium, 2007. ANSS'07. 40th Annual, 2007, pp. 301-309.
[97] C. Tuduce and T. Gross, "A mobility model based on wlan traces and its validation," in INFOCOM 2005. 24th Annual Joint Conference of the IEEE Computer and Communications Societies. Proceedings IEEE, 2005, pp. 664-674.
[98] N. C. Romano Jr and J. F. Nunamaker Jr, "Meeting analysis: Findings from research and practice," in System Sciences, 2001. Proceedings of the 34th Annual Hawaii International Conference on, 2001, p. 13.
[99] R. Jain, A. Puri, and R. Sengupta, "Geographical routing using partial information for wireless ad hoc networks," Personal Communications, IEEE, vol. 8, pp. 48-57, 2001.
[100] J. Yoon, B. D. Noble, M. Liu, and M. Kim, "Building realistic mobility models from coarse-grained traces," in Proceedings of the 4th international conference on Mobile systems, applications and services, 2006, pp. 177-190.
[101] (PTV simulation—VISSIM (2010). www.english.ptv.de/cgi-bin/traffic/traf_vissim.pl ). [102] D. Son, A. Helmy, and B. Krishnamachari, "The effect of mobility-induced location errors on geographic routing in ad
hoc networks: Analysis and improvement using mobility prediction," 2004, pp. 189-194. [103] S. Kwon and N. B. Shroff, "Geographic routing in the presence of location errors," Computer Networks, vol. 50, pp.
2902-2917, 2006. [104] B. Peng, A. Kemp, and H. Maheshwari, "Impact of location errors on energy-efficient geographic routing in wireless
sensor networks," 2009, pp. 830-833. [105] A. Sheffer, M. Etzion, A. Rappoport, and M. Bercovier, "Hexahedral mesh generation using the embedded voronoi
graph," in In Proceedings of the 7th International Meshing Roundtable, 1998. [106] M. Fiore, "Mobility models in inter-vehicle communications literature," Technieal Report, Department of Electronics,
Polytechnic Institute of Torino, 2006. [107] W. Chen, R. K. Guha, T. J. Kwon, J. Lee, and Y. Y. Hsu, "A survey and challenges in routing and data dissemination in
vehicular ad hoc networks," Wireless Communications and Mobile Computing, vol. 11, pp. 787-795, 2011. [108] M. M. Artimy, W. Robertson, and W. J. Phillips, "Assignment of dynamic transmission range based on estimation of
vehicle density," in Proceedings of the 2nd ACM international workshop on Vehicular ad hoc networks, 2005, pp. 40-48.
[109] C. Bettstetter, "Mobility modeling in wireless networks: categorization, smooth movement, and border effects," ACM SIGMOBILE Mobile Computing and Communications Review, vol. 5, pp. 55-66, 2001.
[110] G. J. Klir and B. Yuan, Fuzzy sets and fuzzy logic: Prentice Hall New Jersey, 1995. [111] L. A. Zadeh, "Fuzzy sets," Information and control, vol. 8, pp. 338-353, 1965. [112] E. Cox, "Fuzzy fundamentals," Spectrum, IEEE, vol. 29, pp. 58-61, 1992. [113] E. Mamdani and S. Assilian, "An experiment in linguistic synthesis with a fuzzy logic controller," International journal
of human-computer studies, vol. 51, pp. 135-147, 1999. [114] M. Mauve, A. Widmer, and H. Hartenstein, "A survey on position-based routing in mobile ad hoc networks," Network,
IEEE, vol. 15, pp. 30-39, 2001. [115] E. D. Kaplan and C. J. Hegarty, Understanding GPS: principles and applications: Artech House Publishers, 2006. [116] B. Karp and H. T. Kung, "GPSR: Greedy perimeter stateless routing for wireless networks," 2000, pp. 243-254. [117] K. Seada, A. Helmy, and R. Govindan, "Modeling and analyzing the correctness of geographic face routing under
realistic conditions," Ad Hoc Networks, vol. 5, pp. 855-871, 2007. [118] W. T. Zaumen, "Simulations in Drama , SIR ." Network information system centre International, Menlo Park,
California, Jan 1991.