![Throughput [bps].](https://www.researchgate.net/profile/Dejan_Tepsic/publication/277307281/figure/fig3/AS:613955986792483@1523389922575/Throughput-bps_Q320.jpg)
Content uploaded by Dejan Tepšić
Author content
All content in this area was uploaded by Dejan Tepšić on May 28, 2015
Content may be subject to copyright.
239
Ad Hoc & Sensor Wireless Networks, Vol. 27, pp. 239–261
Reprints available directly from the publisher
Photocopying permitted by license only
©2015 Old City Publishing, Inc.
Published by license under the OCP Science imprint,
a member of the Old City Publishing Group
A Novel Proactive Routing Protocol in
Mobile Ad Hoc Networks*
Dejan Tepsic1, MlaDen VeinoVic1, Dejan ZiVkoVic1 anD naDja ilic2
1Department of Postgraduate Studies and International Cooperation, Singidunum University,
Danijelova 32, 11000 Belgrade, Serbia, phone: +381 64 1848280, tepsicd@gmail.com
2School of Electrical Engineering, University of Belgrade,
Bulevar kralja Aleksandra 73, 11020 Belgrade, Serbia
Received: September 26, 2013. Accepted: October 31, 2013.
Mobile ad hoc networks (MANETs) are self-configuring networks of nodes
connected via wireless without the use of any existing network infrastructure.
In MANETs, each node acts both as a host and as a router, thus it must be
capable of forwarding data packets to other nodes. Due to the mobility of
nodes, topologies of these networks change frequently. In order to solve these
problems, a functional and efficient unicast routing protocol is needed.
Many different unicast routing protocols have been developed for
MANETs. They can be classified into three main categories: proactive,
reactive and hybrid. By studying advantages and disadvantages of each
one, a new routing protocol is proposed. The new scheme called OLSR-
BF is proactive routing protocol mostly based on the existing OLSR rout-
ing protocol. Modification introduced to the new routing protocol is based
primarily on the use of Bellman-Ford algorithm for determining the
shortest distance between nodes within a MANET network.
In order to validate the design and performance of the new routing pro-
tocol it is qualitatively compared with existing AODV, DSDV and OLSR
routing protocols. The final evaluation is presented at the end of this paper.
Keywords: AODV, DSDV, MANET, mobile ad hoc networks, OLSR, routing
protocols.
I. INTRODUCTION
In recent years there has been an expansion in the study of mobile ad hoc
networks (MANETs). MANETs are infrastructure-less and characterized by
* This paper is supported by the Ministry of Education and Science of the Republic of Serbia
through the projects TR32054, ON174008 and III44006.
240 D. Tepsic et al.
mobility of nodes which are constantly moving. The nodes can communicate
directly with each other, provided that they enter into mutually overlapping
communication range. Nodes can forward packets, i.e. serve as relays. There-
fore, the message is relayed through the mobile ad hoc network from one
node to another, until it reaches its destination. As nodes are moving, this
becomes a challenging task, since the topology of MANETs is in constant
change. How to find a destination, how to route to that destination, and how
to provide robust communications due to constant changes in the network
topology are the main challenges in MANETs. Routing in mobile ad hoc
networks is a topic of constant study.
In order to perform this study, experiments were carried out in OPNET
Modeler and NS-3 network simulators. The proposed OLSR-BF routing pro-
tocol is compared with existing AODV, DSDV and OLSR routing protocols.
Data obtained in these experiments quantify and compare the performances,
such as throughput, delay and network load.
II. RELATED WORK
The motivation behind this study is the need for higher performances and
optimization of routing protocols in mobile ad hoc networks. Existing papers
in this field have partly described this [1-4].
In [1], analyzed are throughput performances in mobile ad hoc networks.
The performance of MANETs largely depends on the number of users and
the traffic load. High number of users or traffic increases collisions, which
results in larger wastage of the wireless medium and lowers overall through-
put. The authors have conducted experimental studies under the OPNET
Modeler network simulator in order to compare the throughput of MANETs
in scenarios with different number of users. By analyzing the experimental
results it was concluded that when the number of users is increased beyond
the certain limit, throughput decreases.
In [2], the performances of AODV, DSR and OLSR routing protocols are
compared. AODV is a reactive routing protocol that starts searching for a
destination node whenever it needs to send any information to it. It also
needs a periodic route advertisement and neighbor detection. DSR is also a
reactive routing protocol, but unlike AODV, it is a beacon-less routing pro-
tocol and it does not need periodic route advertisement and neighbor detec-
tion. OLSR is a proactive routing protocol, and hence, each node periodically
broadcasts its routing table allowing each node to build a global view of the
network topology.
Authors investigated the performance of routing protocols in different
scenarios with different traffic loads including end-to-end delay, throughput,
video packet delay variation, and routing traffic overhead. The results of the
simulation shows that in the case of file transfer with TCP connection, OLSR
proacTiVe rouTing 241
acts better in terms of end-to-end delay and upload response time, but with
high routing overhead. In the case of video transfer with UDP connection,
AODV routing protocol acts better in terms of throughput, delay variation,
end-to-end delay, and has low routing traffic.
In [3], a performance analysis of AODV, DSR and TORA routing proto-
cols is investigated in case of self similar traffic. Experiments were performed
through NS-3 network simulator. During simulation runs, data packet deliv-
ery ratio, normalized protocol overhead, throughput and average end-to-end
delay were recorded. Based on these results authors have concluded that
throughput is decreased in all the three protocols as the speed of the mobile
node increases, whereas DSR offers better throughput as compared to AODV
and OLSR. Beside throughput, average end-to-end delay of DSR and OLSR
increases with an increase in speed, whereas AODV have merely effect on the
overall average delay and seems to be constant along simulation run. Also, it
was shown that OLSR could not be suitable enough for higher speed mobility
networks.
In [4], compared are relative performances of DSDV, AODV and OLSR
ad hoc routing protocols. The protocols are tested based on two scenarios,
namely tactical networks for ships and sensor-based network nodes. Four
performance metrics were measured by varying the maximum speed of
mobile hosts, network size and traffic load, to assess the routing capability
and protocol efficiency. The simulation results indicate that AODV per-
forms better than OLSR and DSDV in the first scenario. Although OLSR
also performed relatively well, the associated high routing overhead is the
dominant reason for not choosing it. On the other hand, OLSR emerged
as the protocol of choice for sensor networks, where the high routing
overhead is counteracted by consistently better performance in all other
metrics.
III. MANET ROUTING PROTOCOLS
Routing is the most basic network component for communication establish-
ment and packet transmission. In order to make MANETs practically usable,
functional and efficient unicast routing protocol is needed. Therefore, devel-
oped are numerously different unicast routing protocols, which by mecha-
nism of updating routing information can be divided into proactive, reactive
and hybrid.
Proactive (periodic) routing protocols are based on the parameters from
the table, because each node maintains one or more tables in order to main-
tain network topology and routing information. The main characteristic of
proactive routing protocols is that each node maintains routes to all other
nodes in the network. Nodes periodically update this information, regardless
the given path is used or not.
242 D. Tepsic et al.
Reactive routing protocols, as opposed to proactive ones, do not maintain
information about the network topology and the path to every node in the
network. Routes are established when it is needed (on demand) by using a
process of discovering routes. This type of routing protocol more efficiently
uses throughput of wireless links and limited resources of nodes.
The main problem of proactive approach is high level of utilization of
resources and flow when it is not necessary. The main problem of reactive
approach is delay that occurs when establishing new routes. Both proactive
and reactive approaches have the problem of scalability. In order to mitigate
these problems new kind of hybrid unicast routing protocols were created.
Hybrid routing protocols combine both proactive and reactive approaches in
order to achieve better performances.
A. AODV
“Ad hoc On-demand Distance Vector” (AODV) [5] is a hop-by-hop reactive
routing protocol, meaning that it establishes a route to a destination on
demand, i.e. whenever it is required to transmit data packets from the source
node. This is useful in MANETs, since knowledge of the complete routing
table on each node involves consumption of large amounts of memory and
link capacity. AODV is, as the name indicates, a distance-vector routing pro-
tocol. AODV avoids the counting-to-infinity problem of other distance-vector
protocols by using sequence numbers on route updates. AODV is capable of
both unicast and multicast routing. AODV routing mechanism consists of the
processes of route discovery and route maintenance.
B. DSDV
“Destination-Sequenced Distance Vector” (DSDV) [6] routing protocol is a
hop-by-hop distance vector routing protocol requiring each node to periodically
broadcast routing updates based on the idea of classical Bellman-Ford algo-
rithm. Each node maintains a routing table listing the next hop for each reach-
able destination, number of hops to reach destination and the sequence number
assigned by destination node. The sequence number is used to distinguish stale
routes from new ones and thus avoid loop formation. The stations periodically
transmit their routing tables to their immediate neighbors. A station also trans-
mits its routing table if a significant change has occurred in its table from the last
update sent. So, the update is both time-driven and event-driven. The routing
table updates can be sent as a “full dump” or an “incremental” update. A full
dump sends the full routing table to the neighbors and could span many packets
whereas in an incremental update only those entries from the routing table are
sent that has a metric change since the last update and it must fit in a packet.
C. OLSR
The “Optimized Link State Routing protocol” (OLSR) is an IP routing proto-
col optimized for mobile ad hoc networks [7]. The OLSR protocol is an
proacTiVe rouTing 243
improvement over the older and less effective proactive routing protocol
DSDV. It uses a different routing technique designed to adapt to a network
which is dense and where data transmission is assumed to occur frequently
between large numbers of nodes.
OLSR is a proactive link-state routing protocol, which uses hello and
topology control (TC) messages to discover and then disseminate link-state
information throughout the mobile ad hoc network. Individual nodes use this
topology information to compute next hop destinations for all nodes in the
network, using shortest hop forwarding paths.
Using hello messages the OLSR protocol at each node discovers 2-hop
neighbor information and performs a distributed election of a set of multi-
point relays (MPRs). Nodes select MPRs such that there exists a path to each
of its 2-hop neighbors via a node selected as an MPR. These MPR nodes then
source and forward TC messages that contain the MPR selectors. This func-
tioning of MPRs makes OLSR unique from other link-state routing protocols.
The main goal of using the MPR nodes is to reduce the number of broadcasts
in the same region. Each node selects a set of neighboring nodes within a one
hop (MPR set of nodes). Adjacent nodes of a given node that are not selected
as MPR nodes handle packets, but do not forward them. Only MPR nodes
forward packets.
D. A Novel Proactive Routing Protocol
The routing protocol for MANETs introduced in this paper is called “Opti-
mized Link State Routing Protocol - Bellman-Ford” (OLSR-BF). Original
OLSR routing protocol [7] was utilized as the basis for the development of a
new routing protocol, because it is very popular, widely used and performs
well compared with other proactive routing protocols.
Similar to its predecessor OLSR-BF is also a proactive routing protocol,
so the routes are always available. The route discovery process is based on the
emission of hello and topology control (TC) messages. Besides, the OLSR-
BF is based on the multipoint relay (MPR) concept.
In OLSR-BF, each node selects a group of neighbours as MPR. Hence, to
each node MPR is associated a group of neighbours called MPR selectors.
Only the nodes selected as MPR forward the control information in the net-
work. Thanks to that, it is possible to employ only a few transmissions and
minimize the network load. The control information is updated periodically
by transmission of the control messages sent out by the MPR. Operation of
OLSR-BF does not depend on any central entity.
The parameters used by OLSR-BF to control the protocol overheads are
Hello-interval, TC-interval, MPR coverage and TC-redundancy. Contrary to
classic link-state algorithm DSDV, instead of all links, only small subsets of
links are declared.
The predecessor of a newly proposed OLSR-BF routing protocol, OLSR,
is considered as a good protocol for dense and big mobile networks, thanks to
244 D. Tepsic et al.
the optimization of the MPR nodes. Much like its predecessor, the OLSR-BF
is also a hop-by-hop routing protocol. Thus, each node uses its local informa-
tion to forward the packets to the destination.
The newly proposed OLSR-BF routing protocol is largely based on the
existing OLSR routing protocol [6], with the exception of using Bellman-Ford
algorithm instead of Dijkstra algorithm when calculating the shortest distances
from nodes relative to the parent node. Bellman-Ford is a distance vector algo-
rithm which uses dynamic programming. The main advantage of Bellman-
Ford algorithm, with respect to Dijkstra, lies in the possibility of using edges
with negative weights, and the possibility of using unidirectional connections
between nodes. For instance, MANET can be represented as a directed
weighted graph, where the edges are directed from one to another neighboring
node (asymmetric, unidirectional links). If there is an edge between the
observed nodes directed in the opposite direction, the connection between the
observed nodes is considered to be symmetrical (bidirectional link).
The main reason why the creators of the original version of OLSR routing
protocol [7] decided to use Dijkstra algorithm is in the fact that Dijkstra algo-
rithm has smaller execution time when calculating the shortest distance
between nodes relative to the parent node. However, although slightly more
complex and demanding in terms of resources, Bellman-Ford algorithm
brings significant enhancements in a newly proposed OLSR-BF routing pro-
tocol in mobile ad hoc networks.
Negative weight of edges enables the creation of asymmetric links between
nodes. Dijkstra algorithm does not support graphs with negative weights of
edges. Using the Bellman-Ford algorithm in OLSR-BF routing protocol des-
ignates usage of asymmetric links in order to take advantage of unidirectional
links between neighboring nodes, which would otherwise have been rejected
in the traditional OLSR routing protocol.
Concept of Bellman-Ford algorithm was originally used in traditional
DSDV routing protocol. However, the way on which OLSR-BF routing pro-
tocol functions differs from DSDV routing protocol. Both DSDV and OLSR-
BF are regarded as proactive routing protocols that utilize a table-driven
technique by recording all routes they find between all source-destination
pairs regardless of the use or need of such route. However, the OLSR-BF
routing protocol uses concept of multipoint relays (MPRs). The use of MPRs
is to minimize routing overhead by reducing duplicated retransmissions of
routing information in the same region. It is an optimization over a pure link-
state protocol and henceforth expected to perform better in large and dense ad
hoc networks.
Same as its predecessor OLSR, OLSR-BF is capable of both unicast and
multicast routing. This is not true for the DSDV which is only capable of
unicast routing.
Design of a newly proposed OLSR-BF routing protocol was performed
within OPNET Modeler network simulator. Namely, OPNET Modeler
proacTiVe rouTing 245
within the installation directory on the hard disk stores files of all supported
models of nodes and protocols. Specifically, within the directory \\OPNET\
models\std\manet\ are stored files related to mobile ad hoc networks.
Among them is the original model of OLSR routing protocol, whose origi-
nal algorithm for the calculation of routing tables is stored within the sepa-
rate file.
By using Microsoft Visual Studio programming tool and C/C++ editor,
Dijkstra algorithm was replaced with Bellman-Ford algorithm in the \\
OPNET\models\std\manet\olsr_rte.pr.c file. The original version of the Bell-
man-Ford algorithm was slightly adapted for usage in the OPNET Modeler
network simulator.
Routing Table Calculation
Each node maintains a routing table. Routing table is calculated with the
information obtained from the neighbour tables and from the network topol-
ogy. The nodes that receive a TC message store pairs of linked nodes (previ-
ous hop, node), where the “nodes” are the addresses stored in the TC messages
list. To find a route from source node S to a remote destination node D, a pair
(Previous_hop, D) must be found within the routing table. Once it has been
found, this previous hop becomes an intermediate destination (Destination_
inter) and now it looks for a pair (Previous_hop, Destination_inter). This
process is realized successively until it finds a Previous_hop node in the
neighbours set of the node that looks for the route. In Figure 1 the calculation
of the complete route from the source node S to the destination node D is
shown.
The routing table is calculated again each time a change in the topology is
detected. In such a way, the updating of the routing table is done when there
is a change in:
FIGURE 1
Calculation of route from source node S to the destination node D.
246 D. Tepsic et al.
•The set of neighbouring nodes, or
•The set of neighbouring nodes within two hops.
The topology set is calculated again when a new neighbour appears or an old
one is lost.
IV. RESEARCH METHODOLOGY
The following assumptions, which have been widely adopted in the literature
[8]–[10] have been used throughout this research:
•Each mobile node has sufficient power to function throughout the sim-
ulation time. At no time does a mobile node run out of power or mal-
function because of lack of power. In addition, the wireless transceivers
are active at all times.
•The number of nodes in a given topology remains fixed throughout
the simulation time. There are no nodes joining or leaving the net-
work area during a simulation. However, network partitioning may
still be evident during simulation and so the network may not be con-
nected at all times.
•Transmissions may interfere with each other (i.e. affect each other if
they occur in close proximity). However, a node will always success-
fully decode a transmission provided it is within transmission range of
the source and there is no interfering transmission. All the transmis-
sions that exist are produced within a given network.
•All mobile nodes are homogeneous, i.e. all nodes are equipped with
IEEE 802.11 transceivers with the same nominal transmission range.
•All nodes participate fully in the routing protocol of the network. In
particular, each node participating in the network should also be willing
to forward packets to other nodes in the network.
•A route discovery process can be initiated by any source node which
has a data packet to be transmitted.
V. COMPARATIVE ANALYSIS OF ROUTING PROTOCOLS
This section presents a comparative discussion of four routing protocols
namely, AODV, DSDV, OLSR and OLSR-BF:
•DSDV, OLSR and OLSR-BF are proactive routing protocols which
maintains routes to each and every node in the network, whereas AODV
is a reactive routing protocol which finds the path on demand, or when-
ever the route is required.
proacTiVe rouTing 247
•Broadcasting in DSDV, OLSR and OLSR-BF is done periodically to
maintain routing updates and in AODV, only hello messages are propa-
gated to its neighbors to maintain local connectivity.
•DSDV, OLSR and OLSR-BF routing protocols maintains a sequence
number concept for updating the latest information for a route. Even
the same concept is adapted by AODV routing protocol.
•Due to the periodic updates being broadcasted in DSDV, OLSR and
OLSR-BF routing protocols, throughput is wasted when the nodes are
stationary. But, this is not the case with AODV, as it propagates only
hello messages to its neighbours.
•For sending data to a particular destination, there is no need to find a
route as DSDV, OLSR and OLSR-BF routing protocols maintain all the
routes in the routing tables for each node. Contrary to that, AODV has
to find a route before sending the data, which introduces possible long
delays in transmission.
•Overhead in DSDV, OLSR and OLSR-BF routing protocols is higher
when the network is large and it becomes hard to maintain the routing
tables at every node. But, in AODV overhead is less as it maintains
small tables to maintain local connectivity.
•Proactive routing protocols DSDV, OLSR and OLSR-BF cannot handle
mobility at high speeds due to lack of alternative routes, hence routes in
routing table is stale, whereas in AODV this is the other way, as it find
the routes on demand.
•Throughput decreases comparatively in DSDV, OLSR and OLSR-BF
routing protocols as it needs to advertise periodic updates and event-
driven updates. If the node mobility is high then occurrence of event-
driven updates are more. But, in AODV it does not advertise any routing
updates and hence the throughput is stable.
A comparison of the characteristics of the above ad hoc routing protocols
AODV, DSDV, OLSR and OLSR-BF is given in Table I.
VI. SIMULATION
The aim of the experiments is to measure network performances of the pro-
posed OLSR-BF routing protocol over the existing AODV, DSDV and OLSR
routing protocols in different network topologies and various conditions
within the network. The survey was conducted using discrete event network
simulators OPNET Modeler [11] and NS-3 [12]. OPNET Modeler has pre-
defined models for traditional AODV and OLSR routing protocols. A new
model for OLSR-BF routing protocol was introduced for the purposes of
these experiments. At the moment when this study was conducted simulation
248 D. Tepsic et al.
model for traditional DSDV routing protocol was not available within OPNET
Modeler. Therefore, for purposes of simulation of DSDV routing protocol
NS-3 simulator was used. In addition, simulation results of DSDV routing
protocol were imported into OPNET Modeler simulator in order to compare
with other routing protocols.
Simulation is focused on the metrics of MANET routing protocols, namely
throughput, delay, and network load. For this purpose two simulation sce-
narios were created with different number of mobile nodes. The nodes are
randomly placed in the area of 1000 m x 1000 m. In the first scenario MANET
consists of 20 mobile nodes, whereas in the second scenario the number of
mobile nodes is increased to 50. These values of network size combined with
the amount of generated traffic and control over the number of source nodes
within the MANET provides testing at different network conditions.
The network topology is based on the usage of wireless nodes. Each node
in the network was configured to perform identical routing protocol. In each
scenario, AODV, OLSR and OLSR-BF routing protocols are examined with
default values under the OPNET Modeler simulator (Table II and Table III).
Same applies to the DSDV routing protocol simulation within NS-3 simula-
tor (Table IV). Each scenario was run for 900 seconds. For the analysis of
different routing protocols, a constant FTP traffic was generated in the net-
work with multiple FTP sessions of high traffic load.
Mobility model and wireless network parameters were identical for all of
the nodes in each scenario. “Random Waypoint mobility model” (RWP) was
used [13] and defined in the Mobility Config object. All of the nodes were
randomly moving within a defined wireless domain. Velocity of each mobile
node was defined by a uniform distribution between 0 and 10 m/s.
Characteristic AODV DSDV OLSR OLSR-BF
Routing
Technique Reactive Proactive Proactive Proactive
Routing
Algorithm Route discovery Bellman-Ford Dijkstra Bellman-Ford
Distributed Yes Yes Yes Yes
Unidirectional
Link Support No No No Yes
Multicast Yes No Yes Ye s
Periodic
Broadcast Yes (lite) Yes Yes Yes
Route Recovery New route, notify
source, local repair
Periodic
broadcast
Periodic
updates
Periodic
updates
Route Repository Routing table Routing table Routing table Routing table
TABLE 1
Comparison of routing protocols.
proacTiVe rouTing 249
Receiver Group object was used to limit the possible set of receivers based
on the distance between nodes. Distance threshold was set to a value of 250
meters. Author in [14], came to the conclusion that the most common length
of routes in similar areas were up to 4 hops long. On the transmission range
of 250 meters, the minimum distance needed to cover this number of hops
was 1000 meters, so the topology area was selected as 1000 m x 1000 m.
These values were chosen on the basis of a number of examples from the
literature in which performance measurements of MANET routing protocols
were performed [15]–[17], and are presented in Table V.
The wireless LAN parameters of MANET were configured with default
values available in OPNET Modeler simulator (Table VI), and were identical
Attribute Value
Hello Interval Uniform (1, 1.1) seconds
Allowed Hello Loss 2 packets
Network Diameter 35
Node Traversal Time 0.04 seconds
Route Error Rate Limit 10 packets/second
Timeout Buffer 2
TTL Increment 2
TTL Threshold 7
Local Add TTL 2
Packet Queue Size Infinity
Local Repair Enabled
Addressing IPv4
TABLE 2
AODV parameters.
Attribute Value
Willingness Default
Hello Interval 2 seconds
TC Interval 5 seconds
Neighbor Hold Time 6 seconds
Topology Hold Time 15 seconds
Duplicate Message Hold Time 30 seconds
Addressing IPv4
TABLE 3
OLSR and OLSR-BF parameters.
250 D. Tepsic et al.
Attribute Value
Periodic Update Interval 15 seconds
Weighted Settling Time Enabled
Settling Time 6 seconds
Weighted Factor 0.875
Buffering Enabled
Max Queue Length 100 packets
Max Queue Time 30 seconds
Max Queue Packets Per
Destination 5 packets
Holdtimes 3
Route Aggregation Disabled
Addressing IPv4
TABLE 4
DSDV parameters.
Simulation Parameter Value
Simulator OPNET Modeler version 14.5, NS-3 version 3.18
Topology Size 1000 m x 1000 m
Network Size 20 and 50 nodes
Mobility Model Random Waypoint (RWP)
Node mobility speed Uniform (0, 10) m/s
Traffic Type FTP
Simulation Time 900 seconds
Addressing IPv4
Wireless Standard IEEE 802.11b
Data Rate 11 Mbps
Transmission Range 250 m
Routing Protocol AODV, DSDV, OLSR, OLSR-BF
TABLE 5
Simulation parameters.
for all of the nodes during the simulation runs in both network scenarios.
Similar applies to the DSDV routing protocol, but some configuration
changes were applied within NS-3 simulator so all of the values correspond
to the OPNET Modeler. All nodes were equipped with transponders that use
the IEEE 802.11b standard in wireless communication with the data rate of
11 Mbps.
proacTiVe rouTing 251
VII. EXPERIMENTAL RESULTS
For the purposes of this research within OPNET Modeler and NS-3 network
simulators were created two network scenarios of MANETs with different
number of mobile nodes.
VIII. MANET WITH 20 MOBILE NODES
In this network scenario were used 20 mobile nodes and one FTP server
(Figure 2). All mobile nodes performed identical routing protocol, and the
same experiments have been conducted with AODV, DSDV, OLSR and
OLSB-BF routing protocols. During simulation runs following metrics were
measured: throughput, delay and network load.
A. Throughput
Throughput is the ratio of the total amount of data packets that reaches the
receiver in a given time period. Values of throughput for AODV, DSDV,
OLSR and OLSR-BF routing protocols in a MANET with 20 mobile nodes
are shown in Figure 3. It is notable that OLSR-BF shows increased through-
put compared to the original OLSR routing protocol. DSDV also showed
better throughput performances than OLSR, but much less than OLSR-BF.
Also, it is evident that reactive AODV routing protocol achieves the worst
results of throughput.
B. Delay
Figure 4 shows the overall delay for AODV, DSDV, OLSR and OLSR-BF
routing protocols in a mobile ad hoc network with 20 mobile nodes. This
statistics reflects the time between the packet creation at the source node and
its reception at the destination node. It is evident that reactive AODV routing
Attribute Value
Physical and MAC Model IEEE 802.11b (HR/DSSS PHY)
Data Rate 11 Mbps
Short Retry Limit 7
Long Retry Limit 4
Max Receive Lifetime 0.5 seconds
Buffer Size 256000 bits
Roaming Capability Disabled
TABLE 6
Wireless lan parameters.
252 D. Tepsic et al.
FIGURE 2
MANET with 20 mobile nodes.
FIGURE 3
Throughput [bps].
protocol achieves a significantly higher value than the delay present in the
proactive routing protocols DSDV, OLSR and OLSR-BF. The reason of this
low value within proactive routing protocols is knowledge of routes to the
destination nodes already before communication request arrival.
proacTiVe rouTing 253
C. Network Load
Figure 5 shows the amount of routing traffic sent into the network with 20
mobile nodes for all four routing protocols. This statistics includes the hello
messages sent, TC messages sent and TC messages forwarded. Basic differ-
ence between the proactive and reactive routing principle is evident. From the
experimental results it is clear that proactive routing protocols DSDV, OLSR
FIGURE 4
Delay [s].
FIGURE 5
Network load [bps].
254 D. Tepsic et al.
and OLSR-BF, builds up the complete routing tables immediately at the
beginning of simulation. On the other hand, reactive routing protocol AODV
route to the destination calculates when there is a request for the transfer of
data packets.
OLSR-BF generates the highest value of network load due to a frequent
updating of the routing tables, and selection of MPR nodes with every change
of the network topology. DSDV have similar characteristic because of the
table updates flooded throughout the entire network.
IX. MANET WITH 50 MOBILE NODES
In this scenario, the number of mobile nodes was increased to 50 (Figure 6),
whereas all other parameters in MANET remained the same. Reason for the
increased number of mobile nodes is a network simulation with higher den-
sity of mobile nodes on the same network topology. Identical experiments
were repeated for each of the routing protocols, and same metrics were mea-
sured.
A. Throughput
Figure 7 compares the values of throughput for AODV, DSDV, OLSR and
OLSR-BF routing protocols in MANET with 50 mobile nodes. It was
FIGURE 6
MANET with 50 mobile nodes.
proacTiVe rouTing 255
observed that throughput increases when the number of nodes increases in
the network, since more nodes are available to generate or route packets to
the destination. Keeping the mobility and packet length constant, OLSR-BF
delivers a higher throughput than traditional proactive routing protocols and
a much higher value than reactive AODV routing protocol.
B. Delay
Figure 8 shows the overall delay of AODV, DSDV, OLSR and OLSR-BF
routing protocols in mobile ad hoc network with 50 mobile nodes. The
increased number of nodes in the network allows more hops and different
paths to the destination, which increases the delay of reactive AODV routing
protocol, due to aggressive emission of control packages for establishing
routes to the destination. On the other hand, proactive routing protocols
DSDV, OLSR and OLSR-BF realize higher value of throughput while main-
taining low delay in the network. Compared to network scenario with fewer
nodes, AODV has increased its throughput, but with a substantial increase in
the overall delay in the transmission of data packets to the destination.
C. Network Load
Figure 9 shows the amount of routing traffic sent into MANET with 50
mobile nodes for all three routing protocols. It was observed that for proac-
tive routing protocols DSDV, OLSR and OLSR-BF, value of network load
begins with a sudden high value which then slowly decreases during the sim-
ulation run. The reason for this behaviour is the fact that the proactive routing
protocols establish entire routing table at each node with paths to all the
nodes within the network, regardless of whether a path is used or not.
FIGURE 7
Throughput [bps].
256 D. Tepsic et al.
FIGURE 8
Delay [s].
FIGURE 9
Network load [bps].
Increasing of network load due to the bigger number of mobile nodes
occurs as a consequence of the fact that the packets have to pass through more
intermediate nodes to their destination. Therefore, due to the increased num-
ber of mobile nodes grows the network load in MANETs.
X. DISCUSSION AND INTERPRETATION OF EXPERIMENTAL
RESULTS
For the purpose of this research wireless mobile ad hoc network environment
was created, in order to measure the performance of the proposed routing
proacTiVe rouTing 257
protocol. To perform the experiments OPNET Modeler and NS-3 network
simulators were used. Within the mobile ad hoc network FTP traffic was
transmitted by using the IPv4 addressing scheme, and values of following
metrics were recorded: throughput, delay and network load. In order to com-
pare the obtained results, identical experiments were conducted for other
existing routing protocols AODV, DSDV and OLSR.
A. Throughput
In the network scenario with 20 mobile nodes proactive routing protocols
DSDV, OLSR and OLSR-BF have high throughput. The reason for the high
throughput compared with reactive AODV routing protocol is in the fact
that routes to the destinations are always available due to their proactive
nature.
In the network scenario with 50 mobile nodes relationship between the
observed routing protocols in terms of throughput is unchanged. All rout-
ing protocols show an increase in the value of the throughput with increas-
ing number of nodes. OLSR and OLSR-BF have similar values of
throughput; DSDV is slightly below them, whereas AODV shows signifi-
cantly less value. The reason for the increase in throughput by OLSR and
OLSR-BF routing protocols is produced by the usage of MRP nodes,
which increase the efficiency and reduce the amount of generated control
traffic, by controlling and localizing flooding in the network, thus making
them more efficient in the process of maintaining the connections without
having all nodes in the network to participate in it. OLSR and OLSR-BF
routing protocols are suitable for use in dense networks with a relatively
large number of nodes, which is evident from the fact that both routing
protocols generate more throughput values in a scenario with a large num-
ber of nodes.
The routing protocol which has a higher throughput generally gives the
best results. The obtained results show that OLSR-BF has a higher through-
put than its predecessor.
B. Delay
In all performed experiments DSDV, OLSR and OLSR-BF routing protocols
achieved a lower value of the delay compared to reactive AODV routing pro-
tocol. DSDV, OLSR and OLSR-BF are proactive routing protocols, meaning
that routes to all of the nodes in MANET are always available, regardless of
whether there are data packets that should be transferred. Periodic routing
updates maintain current paths within the network. The absence of delay in
the process of route discovery explains the relatively low overall delay of
DSDV, OLSR and OLSR-BF routing protocols.
In proactive DSDV, OLSR and OLSR-BF routing protocols node which
wants to find a route to the destination only needs to look inside its routing
table. In reactive AODV routing protocol node must initiate the process of
258 D. Tepsic et al.
route discovery, if a valid path to the destination does not exist. Since search-
ing within existing routing table requires much less time than searching for a
new path with a network, proactive routing protocols behaves better than its
competitor in the networks vulnerable to delays.
In a network scenario with 50 mobile nodes AODV suffers significant
performance degradation due to the high level of delay. One of the main rea-
sons for the degradation of the delay in AODV routing protocol is the process
of route discovery. Routing paths in AODV routing protocol cannot be found
immediately, which leads toward a potentially lengthy delays when sending
packets. In general, AODV routing protocol increases the delay by increasing
number of nodes. Also, it is evident that the value of delay in AODV routing
protocol has large variations during simulation run.
With the increasing number of nodes within the MANET relationship
between the observed routing protocols remained unchanged, with a note that
the results of a reactive routing protocol AODV are significantly worse than
in the network scenario with fewer nodes. Experimental results showed that
the OLSR and OLSR-BF are exhibited to a very low overall delay in all sce-
narios, and that the function of their delay is consistent. DSDV showed deg-
radation of delay performance in comparison with a smaller network, but
remained relatively low. AODV has high overall delay, especially in the net-
work with a large number of mobile nodes. The results showed that in terms
of delay, OLSR-BF routing protocol achieved slightly worse performances
than its predecessor.
C. Network Load
In a MANET with 20 mobile nodes AODV has the lowest value of the net-
work load due to establishment of path to the destination on demand, i.e.
only when there is a need to transfer data packets to the destination. As a
result, AODV does not need to update routing tables periodically. Proactive
routing protocols DSDV, OLSR and OLSR-BF achieve significantly higher
network loads due to the fact that they constantly maintain routing tables in
the nodes.
Similar considerations apply for the network scenario with 50 mobile
nodes, where AODV again has least network load. It is notable that the
OLSR-BF routing protocol generates the largest amount of routing traffic
within the network, closely followed by the DSDV and OLSR routing proto-
cols. DSDV, OLSR and OLSR-BF are proactive routing protocols that con-
stantly broadcast network control traffic in order to maintain valid routing
tables. In MANETs with low capacity links, AODV is more usable than the
proactive routing protocols.
The results showed that AODV routing protocol is stable and efficient in
terms of network load. The worst performance in terms of the amount of
generated routing messages can be seen in OLSR-BF routing protocol.
proacTiVe rouTing 259
X. SUMMARY OF EXPERIMENTAL RESULTS
Experimental results showed that DSDV, OLSR and OLSR-BF routing pro-
tocols outperform reactive AODV routing protocol in terms of the value of
throughput and low overall delay in MANETs of low mobility and high traf-
fic load. On the other hand, AODV is the complete opposite of the proactive
DSDV, OLSR and OLSR-BF routing protocols. AODV generates much less
network load, but failed in all other aspects that favours the usage of proactive
routing protocols. OLSR-BF has the worst performance in terms of the
amount of routing messages generated, and therefore it is not suitable for
usage in MANETs with low capacity links. Results of this study showed that
the choice of an appropriate ad hoc routing protocol should be performed
based on the network topology and the amount of traffic which is transmitted
over a network.
Summary of experimental results for both network scenarios with differ-
ent number of mobile nodes is shown in Table VII.
CONCLUSIONS
A qualitative study of the performance of various routing protocols shows
that OLSR-BF routing protocol dominates in high density networks with low
mobility nodes. However, OLSR-BF routing protocol requires the continued
usage of certain throughput in order to update the network topology and rout-
ing tables. A similar effect is also present in the existing DSDV and OLSR
routing protocols. Generally, the output of a newly proposed OLSR-BF rout-
ing protocol is more efficient than the original OLSR routing protocol in both
observed network scenarios.
Experimental analysis showed that the proposed OLSR-BF routing proto-
col is better in terms of the value of the throughput than its predecessor
Nodes Parameter AODV DSDV OLSR OLSR-BF
20
Throughput Medium High High Highest
Delay High Low Lowest Low
Network Load Low High High Highest
50
Throughput Medium High High Highest
Delay Highest Medium Lowest Low
Network Load Medium High High Highest
TABLE 7
Summary results of experiments.
260 D. Tepsic et al.
OLSR, also better than traditional DSDV routing protocol, and far better than
a reactive AODV routing protocol. However, it is not proven that this is true
for all MANETs, because performance of routing protocols may vary in dif-
ferent network environments.
From this study it can be concluded that there is no single routing protocol
that is superior in terms of all performance metrics. One routing protocol may
be superior in terms of low network load, whereas others may be superior in
terms of high throughput and low oveall delay. The choice of a particular
routing protocol depends on the purpose of the network.
Finally, the fact is that the nature of routing protocol, proactive or reactive,
has far-reaching consequences on the performance of routing protocols in
MANETs. The main differences are reflected primarily in the way of discov-
ering and maintaining routing paths, which further dictates the behavior of
routing protocols. Generally, proactive routing protocols are suitable for
usage in MANETs with high capacity links, whereas reactive routing proto-
cols work better in low-capacity mobile ad hoc networks.
FUTURE WORK
There are several research directions in this field. Future studies of observed
routing protocols may use different performance metrics. An extensive study
can be carried over the same routing protocols using IPv6 addressing scheme.
Extensive studies can be done by considering other MANET routing proto-
cols and assess their performances using various metrics.
REFERENCES
[1] Fazeli, M., Vaziri, H., “Assessment of Throughput Performance under OPNET Modeler
Simulation Tools in Mobile Ad Hoc Networks (MANETs),” International Conference on
Computational Intelligence, Communication Systems and Networks (CICSYN), pp. 328–
331, 2011.
[2] Lally, S., Trajkovic, Lj., “Performance Analysis of Routing Protocols for Wireless Ad-Hoc
Networks,” in Proc. OPNETWORK, Washington, DC, USA, Aug. 2011.
[3] Al-Maashri, A., Khaoua, M. O. , “Performance Analysis of MANET Routing Protocols in
the Presence of Self-Similar Traffic,” in Proc. 31st IEEE conference on Local Computer
Networks, pp. 801–807, Nov. 2006.
[4] Lye, P. G., McEachen, J. C., “A Comparison of Optimized Link State Routing with Tra-
ditional Ad-hoc Routing Protocols,” in Proc. 5th Workshop on the Internet, Telecommuni-
cations and Signal Processing, Hobart, Australia, Dec. 2006.
[5] Ad hoc On-Demand Distance Vector (AODV) Routing, IETF RFC 3561, July 2003.
[6] Perkins, C. E., Bhagwat, P., “Highly Dynamic Destination-Sequenced Distance-Vector
Routing (DSDV) for Mobile Computers,” in Proc. Conference on Communications archi-
tectures, protocols and applications (SIGCOMM), New York, NY, USA, pp. 234–244,
1994.
[7] Optimized Link State Routing Protocol (OLSR), IETF RFC 3626, October 2003.
proacTiVe rouTing 261
[8] Yassein, M. B., Khaoua, M. O., Mackenzie, L. M., Papanastasiou, S., “Performance Anal-
ysis of Adjusted Probabilistic Broadcasting in Mobile Ad Hoc Networks,” International
Journal of Wireless Information Networks, pp. 1–14, 2006.
[9] Peng, W., Lu, X. C., “On the Reduction of Broadcast Redundancy in Mobile Ad Hoc
Networks,” in Proc. of the Annual Workshop on Mobile and Ad Hoc Networking and Com-
puting (MobiHoc), Boston, Massachusetts, USA, pp. 129–130, 2000.
[10] Ni, S., Tseng, Y., Chen, Y., Sheu, J., “The Broadcast Storm Problem in a Mobile Ad Hoc
Network,” in Proc. of the ACM/IEEE International Conference on Mobile Computing and
Networking (MobiCom), pp. 151–162, 1999.
[11] OPNET Modeler Network Simulator, [Online]. Available: http://www.opnet.com
[12] NS-3 Network Simulator, [Online]. Available: http://www.nsnam.org
[13] Bettstetter, C., Wagner, C., “The Spatial Node Distribution of the Random Waypoint
Mobility Model,” in Proc. of the German Workshop on Mobile Ad Hoc Networks (WMAN),
Ulm, Germany, pp. 48–51, 2002.
[14] Papanastasiou, S., Investigating TCP Performance in Mobile Ad Hoc Networks, Saar-
brucken, Germany: VDM Verlag, 2008, pp. 127–130.
[15] Wang, X., Mobile Ad-Hoc Networks: Protocol Design, InTech, 2011, pp. 615–638.
[16] Kumar, S., “Reactive and Proactive Routing Protocols for Wireless Mesh Network using
Multimedia Streaming,” International Conference on Recent Advances and Future Trends
in Information Technology (iRAFIT), no. 7, pp. 13–17, Apr. 2012.
[17] Tafti, V. A., Gandomi, A., “Performance of QoS Parameters in MANET Application Traf-
fics in Large Scale Scenarios,” World Academy of Science, Engineering and Technology,
vol. 72, Feb. 2011.
