The Impact of Speed on the Performance of Dynamic Source Routing in Mobile Ad-Hoc Networks

Document Sample
The Impact of Speed on the Performance of Dynamic Source Routing in Mobile Ad-Hoc Networks Powered By Docstoc
					                                                              (IJCSIS) International Journal of Computer Science and Information Security,
                                                                                                             Vol. 8, No. 8, November 2010


The Impact of Speed on the Performance of Dynamic Source Routing in Mobile Ad-
                                 Hoc Networks


                    Naseer Ali Husieen, Osman B Ghazali, Suhaidi Hassan, Mohammed M. Kadhum
                                           Internetworks Research Group
                                              College of Arts and Sciences
                                                University Utara Malaysia
                                           06010 UUM Sintok, Malaysia
                         naseerali@internetworks.my|{osman|suhaidi|kadhum}@uum.edu.my


Abstract— Ad-hoc networks are characterized by                      Reactive Protocols: In this type of protocols such as
multihop wireless connectivity, frequently changing                 DSR, AODV, the routes are created when it’s required to
network topology and the need for efficient dynamic                 send data packets from the source to the destination [1].
routing protocols plays an important role. Due to                   We have determined the impact of four factors on the
mobility in Ad-hoc network, the topology of the                     performance of DSR by using Random waypoint
network may change rapidly. The mobility models                     mobility model in our previous paper [2] in press. These
represent the moving behavior of each mobile node in                factors pause time, network size, number of traffic
the MANET that should be realistic. This paper                      sources and routing protocol. We examine the impact of
concerns performance of mobile Ad-hoc network                       these factors on four performance metrics: packet
(MANET) routing protocol with respect to the effects                delivery ratio, average end-to-end delay, normalized
of mobility model on the performance of DSR                         routing load and protocol overhead. In this paper, we use
protocol for the purpose of finding the optimal                     Random waypoint as mobility model on DSR protocol to
settings of node speed. In this paper, we evaluate the              study the effect of node speed with other factors in order
performance of DSR protocol using Random                            to find the optimal setting for the node speed parameter
Waypoint Mobility Model in terms of node speed,                     with different scenarios. For this performance study, we
number of connections, and number of nodes.                         use Network Simulator 2 (ns-2) version 2.34.

Keywords-MANET, Mobility Models, Routing Protocol,
DSR Protocol.                                                                II.   DYNAMIC SOURCE ROUTING ( DSR )
                                                                         DSR is reactive and efficient protocol. It determines
                   I.     INTRODUCTION                               the correct path only when a packet wants to be
    With existing advances in technology, wireless                   forwarded. The node broadcast the network with a route
networks are growing in popularity. Wireless networks                request and builds the essential path from the responses
permit users the freedom to move from one position to                it receives. DSR allows the network to be fully self
another without break of their computing services. Ad-               configuring with no need for any existing network
hoc networks is one of the subset of wireless network                infrastructure or administration. The DSR protocol is
that dynamically forming a temporary network without                 composed of two main mechanisms that work together to
using any existing network infrastructure or centralized             allow the discovery and maintenance of source routes in
administration. A major problem in ad hoc network is                 the ad-hoc network. All aspects of protocol operate
how to send data packets among mobile nodes efficiently              entirely on demand allowing routing packet overhead of
without fixed topology or centralized control, which is              DSR to scale up automatically [3] [4].
the most important goal of ad hoc routing protocols.
Therefore, it is necessary a high-quality routing protocol           Route Discovery: The example for route discovery
in order to establish the link between the nodes, since the          shown in Figure 1. When a source node 1 wants to send
mobile node can vary their topology regularly. In ad-hoc             data packets to the destination node 8, node 1 will
network the routing protocol is one of the important                 broadcast Route Request Packet (RREQ) to all the
issue and most challenging research area, since mobile               neighbor nodes 2, 3, 4. After intermediate nodes receive
ad-hoc network vary their topology frequently.                       theses packets will rebroadcast these packets to the
Generally, the major task of routing in a network is to              destination if there is no route in the route cache. When
detect and keep the best path to send data packets                   the destination node 8 will receive RREQ, node 8 will
between source and target through intermediate nodes.                inform the source node 1 by sending the Route Reply
There are two categories of routing protocols in ad hoc              Packet (RREP). The source node will start sending the
networks: Protocols: In this type of protocols such as               data packets to the destination through the intermediate
DSDV, OLSR, consistent and up to-date routing                        nodes. This process mechanism called route discovery.
information to all nodes is maintain at each node. ii.




                                                              228                               http://sites.google.com/site/ijcsis/
                                                                                                ISSN 1947-5500
                                                              (IJCSIS) International Journal of Computer Science and Information Security,
                                                                                                             Vol. 8, No. 8, November 2010

  Route Maintenance: This mechanism contains two
 packets; Route Error Packet (RERR) and ACKs packets.
 Route error packet generated when there is changing in
 the network or node out of the transmission range which
 causes link failure. Intermediate node will send RERR
 to the source node. Source will check if there is route in
 the route cache to send the packets to the destination, if
 there is no alternative route. Source node will reinitiate
 route discovery process again. These processes will take
 long delay to re-establish again in order to send the data
 packets to the destination [5].




                                                                       Fig.2. Traveling pattern of an MN using the Random Waypoint
                                                                                               Mobility Model


                                                                                         IV.    RELATED WORK
                                                                       In the recent years, several works has been done by
                                                                     Perkins, Hughes and Owen [7] shows that some
                                                                     parameters such as pause time, node speed, increasing
                                                                     the number of nodes, and increase the number of
                                                                     sources can have an effect on the routing protocols
                   Fig.1. DSR Route Discovery
                                                                     performance. In their work Random waypoint model
                                                                     has been used, but employed Global Mobile System
            III.    RANDOM WAY POITN MODEL                           Simulator (GloMoSim) rather than ns-2.
     The Random waypoint model is widely and simply                    Azzedine Baoukerche has comparing four routing
used to evaluate the performance of ad hoc routing                   protocols such as (AODV, PAODV, DSR, and CBRP)
protocols. The implementation of this model in the                   [8]. The simulation parameters were tested in his paper
network simulator (ns-2) is as follows: each mobile node             with maximum number of nodes 25 and low traffic with
arbitrarily selects one position in the simulation field as          maximum speed 20 m/s. However, our work tested with
the target, then moves towards this target with fix                  various numbers of nodes (10, 20, 40, and 80) with
velocity selected uniformly and randomly from [0,                    different source connections (4, 8, 30, and 40) which
Vmax], where the parameter Vmax is the maximum                       can make high traffic and various speed 20, 40, 60, and
velocity for each mobile node [6]. The velocity and path             80.
of the nodes are selected separately from of other nodes.               Yogesh, Yudhvir, and Manish, they have compared
When will reaching the target, the node stops for a                  and analysis two reactive routing protocols such as
period of time defined by the ‘pause time’.                          AODV, DSR [9]. The main objective in their paper to
   In the Random waypoint model, velocity and pause                  evaluate the performance of these two protocols based
time are two key parameters that determine the mobility              on the packet delivery fraction, end –to-end delay, and
performance of nodes. When the pause time is long and                normalized routing load. The simulation parameters
velocity is small the topology of ad-hoc network                     were tested increasing number on nodes and various
becomes stable. On the other hand, when the mobile                   pause times with fixed maximum speed (0-25 m/s
node moves fast and the pause time is small; the                     only). While our work with various maximum speeds in
topology is likely to be highly dynamic. Random                      order to select the optimal setting for maximum speed.
waypoint model can create different mobility scenarios
with different levels of node speed. In Figure 2, shows                               V.       SIMULATION SETUP
that node movement in the Random waypoint.
                                                                       The MANET network simulations are implemented
                                                                    using Random waypoint model which can generate by
                                                                    using movement tool (setdest) in ns-2 simulator. The
                                                                    simulation period for each scenario is 200 seconds and
                                                                    the simulated mobility network area is 1000 m x 500 m
                                                                    rectangle. Simulation runs are made with the number of
                                                                    random traffic Constant Bit Rate (CBR) which can
                                                                    generate by using (cbgen.tcl). Figure 2, it shows that the




                                                              229                                  http://sites.google.com/site/ijcsis/
                                                                                                   ISSN 1947-5500
                                                                (IJCSIS) International Journal of Computer Science and Information Security,
                                                                                                               Vol. 8, No. 8, November 2010

simulation methodology for our implementation. The                    Average end to end delay: this is defined as the average
rest of simulation parameters shown in the Table below.               delay in transmission of a packet between two nodes and
                                                                      is calculated as follows:
           TABLE I. SIMULATION PARAMETERS
                                                                                 n
                                                                                       Time packet receivedi – time packet senti
                                                                       AED = ∑
  Parameters                    Value                                            i=0     Total number of packets received
  Simulation Time               200 s
  No. of Nodes                  10, 20, 40, 80                                                   Equation.2 AED
  No. of connections            4, 8, 30, 40
  Pause Time                    40 s                                   A higher value of end-to-end delay means that the
  Simulation Area               1000 x 500 m                          network is congested and hence the routing protocol
  Traffic Type                  Constant Bit Rate (CBR)               does not perform well. The upper bound on the values of
  Maximum Speed                 20 ,40,60,80 m/s                      end-to-end delay is determined by the application.
  Mobility Model                Random Waypoint
  Routing Protocol              DSR                                   Protocol Control Overhead: This is the ratio of the
                                                                      number of protocol control packets transmitted to the
  MAC Type                      802.11                                number of data packets received.
                                                                      Normalized routing load: this is calculated as the ratio
                                                                      between the numbers of routing packets transmitted to
                                                                      the number of packets actually received (thus accounting
                                                                      for any dropped packets):




                                                                                                Equation.3 NRL
                                                                       This metric gives an analysis of routing protocol
                                                                       efficiency, since the number of routing packets sent per
                                                                       data packet gives an idea of how well the protocol
                                                                       maintains the routing information updated. The lower
                                                                       of NRL, the lower the overhead of routing protocol and
                                                                       consequently the higher the efficiency of the protocol.


                  Fig.3. Simulation Methodology                                           VII. RESULTS AND DISCUSSION
                                                                       In this section, details of the simulation results in term
               VI.    PERFORMANCE METRICS                              of packet-delivery fraction, average end to end delay,
                                                                       protocol overhead, and normalized routing load. All the
We have consider packet delivery ratio, end to end delay,              results were obtained by averaging 5 times over the
protocol control overhead and normalized routing load                  simulation for every scenario in order to select the
as a metrics during our simulation in order to evaluate                optimal setting for the maximum speed.
the performance of the DSR protocol.
                                                                       i. Packet Delivery Fraction (PDF)
Packet Delivery Fractions (PDF): the packet delivery
ratio is calculated by dividing the number of packets                  This metric with high packet delivery ratio, routing
received by the destination through the number of                      protocol will be more efficient. Figure 4, shows that
originates packets by the application layer of the initiator.          packed delivery ratio is decreased in first scenario
PDF is specifying the loss packets rate, which limits the              whenever increasing node speed. As above stated, it
maximum throughput over the entire network. The more                   has been taken four main scenarios and each of this
complete and correct routing protocol, the better packet               main contains four sub-scenarios which means 4x4 =
delivery ratio.                                                        16 scenarios have been taken in this experiment. In
                                                                       each of these experiments of sub-scenarios, the node
                                                                       speed increases while other parameters are constant. In
                                                                       Figure 4, shows that the optimal setting is 20 speeds
                                                                       among all of the four scenarios in term of packet
                                                                       delivery ratio.
              Equation.1 PDF




                                                                230                                        http://sites.google.com/site/ijcsis/
                                                                                                           ISSN 1947-5500
                                                                                                        (IJCSIS) International Journal of Computer Science and Information Security,
                                                                                                                                                       Vol. 8, No. 8, November 2010


                                                            Speed vs Packet Delivery Fraction
                                          120                                                                 iii. Routing Protocol Overhead
           Packet Delivery Fraction (%)   100
                                                                                                              In Figure 6, it shows the overhead result which
                                          80
                                                                                                              generated by the routing protocols to achieve this level
                                          60                                                                  of data packet delivery. Figure 6, shows that overhead is
                                          40                                                                  direct proportional to the number of sending packets, in
                                          20
                                                                                                              the first scenario with low mobility overhead increased
                                                                                                              whenever node speed increased. With normal speed 20
                                           0
                                                                                                              m/s in first scenario overhead decreased. The rest of
                                                      20        40         60        80
                                                                                                              scenarios with high mobility overhead is expect to
                                                                  Speed (m/s)
                                                                                                              increase and decrease because there are more
                                                    Scenario1: 10 nodes           Scenario2: 20 nodes
                                                                                                              destinations to which the network must maintain
                                                    Scenario3: 40 nodes           Scenario4: 80 nodes
                                                                                                              working routes.
                                                    Fig.4. Speed vs. Packet Delivery Fraction

                                                                                                                                                                                Speed vs Routing overhead
ii. Average End-to-End Delay                                                                                                                                 4000
The average end-to-end delay is affected when the                                                                                                            3500




                                                                                                                 Routing overhead
traffic Constant Bit Rate (CBR) is high rate of                                                                                                              3000
packets as well. The buffers become filled much                                                                                                              2500
                                                                                                                                                             2000
faster, so the packets have to wait in the buffers a
                                                                                                                                                             1500
much longer period of time before they are sent.
                                                                                                                                                             1000
Figure 5, shows that effect of node speed on the end
                                                                                                                                                              500
to end delay. In the first scenario 10 nodes with 4
                                                                                                                                                                0
connections CBR, there is less delay with 20 speeds
                                                                                                                                                                        20         40       60              80
comparing with others sub –scenario 40, 60, and 80                                                                                                                                  Speed (m/s)
speeds. In addition when the number of mobile nodes                                                                                                                 Scenario1: 10 nodes               Scenario2: 20 nodes
(MNs) is increased to 80 nodes and 40 CBR                                                                                                                           Scenario3: 40 nodes               Scenario4: 80 nodes
connections with respect increase node speed up to
80 m/s the end to end delay increases because of the                                                                                                                  Fig.6.Speed vs. Routing overhead
time consumed for route discovery and the increasing
number of packets in the buffer. However, when the                                                                       iv. Normalized Routing Load (NRL)
pause time is 40 s and speed 20 m/s, the network is
stable and the end to end delay decreases. With                                                                In figure 6, the value of normalized routing load versus
normal speed, the end to end delay is low because the                                                          node speeds are plotted. From the figure 7, this is clear
network is not congested.                                                                                      that DSR protocol performs well in the first scenario
                                                                                                               with node speed 20 m/s. Because of the NRL direct
                                                                                                               proportional with overhead and sending packets.NRL
                                                       Speed vs Average end to end Delay
                                                                                                               represents the number of routing packets transmitted per
                                          7000
 Average end to end Delay




                                                                                                               data packet delivered at the destination. This metrics
                                          6000
                                                                                                               checks the efficiency of the DSR protocols, meaning
                                          5000                                                                 that with low NRL, DSR perform well.
                                          4000
                                          3000
                                                                                                                                                                            Speed vs Normalized routing load
                                                                                                                            Normalized routing load (NRL )




                                          2000                                                                                                                3.5

                                          1000                                                                                                                 3

                                                0                                                                                                             2.5

                                                       20        40      60          80                                                                        2
                                                                   Speed (m/s)                                                                                1.5
                                                     Scenario1: 10 nodes        Scenario2: 20 nodes                                                            1

                                                     Scenario3: 40 nodes        Scenario4: 80 nodes                                                           0.5
                                                                                                                                                               0
                                            Fig.5.Speed vs. Average end to end delay                                                                                   20         40          60          80
                                                                                                                                                                                       Speed (m/s)

                                                                                                                                                                     Scenario1: 10 nodes             Scenario2: 20 nodes

                                                                                                                                                                     Scenario3: 40 nodes             Scenario4: 80 nodes



                                                                                                                                                                    Fig.7.Speed vs. Normalized routing load




                                                                                                        231                                                                                http://sites.google.com/site/ijcsis/
                                                                                                                                                                                           ISSN 1947-5500
                                                                             (IJCSIS) International Journal of Computer Science and Information Security,
                                                                                                                            Vol. 8, No. 8, November 2010




              VIII.   CONCLUTION AND FUTURE WORK                                    AUTHORS PROFILE
        The most important parameter of a mobility model
      is a node speed, either in the form of a constant value                                                    Naseer      Ali     Husieen,
      or in the form of a certain distribution. In this paper,                                                   received his B.Sc. degree in
      we have present our work on evaluating performance                                                         Computer Science from Al-
      of DSR protocol under widely and simplest mobility                                                         Rafedain University, Iraq
      model which called Random Waypoint with respect                                                            and, his M.Sc. degree in
      to the effect of node speed for different scenarios.                                                       Computer Science focusing
      Simulation result shows that node speed has                                                                on Computer Network and
      significant on the performance of DSR protocol                                                             Communications         from
      because mobility models are characterized by the                                                           Hamdard University, Delhi
      movement of their constituents. The nature of                                                              (Faculty    of      Computer
      movement its speed, direction, and rate of change                                                          Science),    India.   Naseer
      can have a dramatic effect on protocols and systems                           currently attached to the InterNetWorks Research Group,
      designed to support mobility. The Mobile nodes                                College of Arts and Sciences at the Utara University
      (MNs) randomly select the next destination in the                             Malaysia. He is currently pursuing his PhD research in
      simulation area and choose a speed uniformly                                  Ad-hoc Mobile networking as a doctoral researcher. His
      distributed between the minimum and maximum                                   current research interest is on Ad-hoc mobile network
      speed and travels with a random speed which is                                routing protocol.
      chosen uniformly. In addition, results shows that
      average optimal setting for our scenarios is when
      node speed is 20 m/s. The experimentation suggests                                                        Osman Ghazali, Ph.D. a
      that several parameters such as maximum speed,                                                            Senior Lecturer in the
      node density pause time and traffic source of nodes                                                       Department of Computer
      also affect the routing performance and need to be                                                        Science for Postgraduate
      investigated with various mobility models.                                                                Studies, Northern University
                                                                                                                of Malaysia (Universiti
                                                                                                                Utara     Malaysia).     He
                               REFERENCE                                                                        received his BIT, Master
[1]    Elizabeth Belding –Royer,” Routing approaches inn mobile ad                                              and PhD in Information
       hoc networks”, in: S.Basagni, M.Conti, S.Giordano,                                                       Technology from Northern
       I.Stojemenvoic (Eds), Ad Hoc Networking, IEEE Press Wiley,
       New York, 2003.                                                                                          University of Malaysia in
[2]     Naseer,Osman, Suhaidi, Kadhum “Effect of Pause Time on the                                              1994, 1996, and 2008. He
       Performance of Mobile Ad Hoc Network Routing Protocols”. in                  published a    number     of papers in international
       IEEE 4International Conference on Inteligent Information                     conferences.
       Technology Application (IITA 2010), China, 2010, in press
[3]    D.B. Johnson, D.A. Maltz, Y Hu, Dynamic Source Routing
       Protocol    for      Mobile      Ad-hoc        Networks      (DSR),
       http://www.ietf.org/internet-drafts/draft-ietf-manet-dsr-10.txt,
       July 2004.
[4]    David B. Johnson David A. Maltz Josh Brooch, “DSR: the                                                  Associate Professor Suhaidi
       Dynamic Source Routing Protocol for Multi-Hop Wireless                                                  Hassan is currently the
       Ad Hoc Networks”, http://www.monarch.cs.cmu.edu/.
                                                                                                               Assistant Vice Chancellor of
[5]    The Dynamic Source Routing Protocol for Mobile Ad-hoc
       Networks, “http://www.ietf.org/internet-drafts/draft-ietf-manet-                                        the College of Arts and
       dsr-03.txt, IETF Internet draft, Oct. 1999.                                                             Sciences, Universiti Utara
[6]    L. Breslau, D. Estrin, K. Fall, S. Floyd, J. Heidemann, A. Helmy,                                       Malaysia (UUM). He is an
       P. Huang, S. McCanne, K. Varadhan, Y. Xu, and              H. Yu,                                       associate     professor    in
       Advances in network simulation, in IEEE Computer, vol. 33, no.                                          Computer      Systems    and
       5, May 2000, pp. 59--67.
                                                                                                               Communication       Networks
[7]    D. Perkins, H. D. Hughes, and C. B. Owen, “Factors affecting the
       performance of ad-hoc networks,” in Proceedings of the IEEE                                             and the former Dean of the
       International Conference on Communications (ICC),Electronic                                             Faculty    of     Information
       Publication: Digital Object Identifiers (DOIs),2000.
                                                                                   Technology, Universiti Utara Malaysia. Dr. Suhaidi
[8]    Azzedine Boukerche,” Performance Evaluation of Routing
       Protocols for Ad Hoc Wireless Networks”, Kluwer Academic                    Hassan received his B.Sc. degree in Computer Science
       Publishers. Manufactured in the Netherlands Mobile Networks                 from Binghamton University, New York (USA) and his
       and Applications 9, 333–342, 2004.                                          MS degree in Information Science (concentration in
[9]    Yogesh, Yudhvir, and Manish,” Simulation based Performance                  Telecommunications and Networks) from the University
       Analysis of On-Demand Routing Protocols in MANETs, Second
       international Conference on Computer Modeling and Simulation.               of Pittsburgh, Pennsylvania (USA). He received his PhD




                                                                             232                               http://sites.google.com/site/ijcsis/
                                                                                                               ISSN 1947-5500
                                                             (IJCSIS) International Journal of Computer Science and Information Security,
                                                                                                            Vol. 8, No. 8, November 2010

degree in computing (focusing in Networks Performance
Engineering) from the University of Leeds in the United
Kingdom. In 2006, he established the ITU-UUM Asia
Pacific Centre of Excellence (ASP CoE) for Rural ICT
Development, a human resource development initiative of
the Geneva-based International Telecommunication
Union (ITU) which serves as the focal point for all rural
ICT development initiatives across Asia Pacific region by
providing executive training programs, knowledge
repositories, R &D and consultancy activities. Dr. Suhaidi
Hassan is a senior member of the Institute of Electrical
and Electronic Engineers (IEEE) in which he actively
involved in both the IEEE Communications and IEEE
Computer societies. He has served as the Vice Chair
(2003-2007) of the IEEE Malaysia Computer Society. He
also serves as a technical committee for the Malaysian
Research and Educational Network (MYREN) and as a
Council Member of the Cisco Malaysia Network
Academy.




                               Mohammed M. Kadhum,
                               Ph.D. is an assistant
                               professor in the Graduate
                               Department of Computer
                               Science, Universiti Utara
                               Malaysia (UUM) and is
                               currently attached to the
                               InterNetWorks     Research
                               Group at the UUM College
                               of Arts and Sciences as a
                               research advisor. He had
                               completed his PhD research
 in computer networking at Universiti Utara Malaysia
 (UUM). His research interest is on Internet Congestion
 and QoS. He has been awarded with several medals for
 his outstanding research projects. His professional
 activity includes being positioned as Technical Program
 Chair for NetApps2008 and NetApps2010, a technical
 committee member for various well known journal and
 international conferences, a speaker for conferences, and
 a member of several science and technology societies.
 To date, he has published a number of papers including
 on well-known and influential international journals.




                                                             233                               http://sites.google.com/site/ijcsis/
                                                                                               ISSN 1947-5500