Performance Evaluation for Scalable Recursive Multicast Protocol using ns2 simulator

Document Sample
Performance Evaluation for Scalable Recursive Multicast Protocol using ns2 simulator Powered By Docstoc
					                               Performance Evaluation for Scalable Recursive
                                       Multicast Protocol using ns2 simulator applications
                                                             important for internet                          are still
1                                 1
              2                                                                                    1
Jafar Ababneh, Firas E.Albalas, Nidhal Kamel Taha El-Omari, Abdel Rahman A.Alkarabsheh, Abd Alsalam Obiadat, 3[5],
                                                                        open such as scalability [4], billing Mahmood Baklizi
                                                                        address allocation [6] and security [7],
                     Faculty of science and information technology, The World Islamic Sciences and Education  much
                                                                        where the scalability has drawn
                                    (W.I.S.E.) University, Amman, 11947, P.O. Box 1101, Jordan 
                                                                        attention among these issues.
                                      { jafar.ababneh, nidhal.omari, Ar.karabsheh}
                                                                        The main issue that causes the scalability
                   Faculty of science and information technology, Jadara University, Amman – Irbid main Road,  the
                                                                        problem is the increasing in the size of
                                                   21110   ,  P.O. Box 733, Jordan
                                                                        Multicast Forwarding Table (MFT) because
                                                                        of the increase in multicast group members
             National advanced IPV6 center (NAV6) university sains Malaysia,11800 USM, penang, malaysia

                                                                                or the increase in the number of multicast
                                                                                groups. So there are two main aspects that
                                                                                can be used to evaluate the scalability in
            In multicast routing the scalability issue                          multicast protocols: scalability with regard
            should be considered, this issue comes                              to the number of group members and to the
            because the increasing in the size of the                           number of multicast groups in the network.
            Multicast Forwarding Table (MFT) because                            Recently, there are a number of proposed
            of the increase in multicast group members                          mechanisms to solve the scalability issue in
            or the increase in the number of multicast                          multicast protocols, which can be
            groups. SReM[1] is a multicast routing                              categorized into tunnelling techniques,
            protocol that addressed this issue by                               forwarding state reduction and explicit
            explicitly encode the building the multicast                        multicast.
            An extensive evaluation performance is                              The main aim for this paper is to perform an
            considered for this protocol in this paper. As                      extensive performance evaluation of a novel
            a result, this protocol gave an improvement                         explicit multicast protocol; Scalable
            in the scalability issue by minimizing the                          Recursive Multicast Protocol (SReM)
            header size and gave an improvement in the                          [8];and compare this protocol with most
            packet delivery ration and the end to end                           well-known explicit multicast protocols.
                                                                                                SREM Overview
                                                                                Due to the scarce resources in the Internet,
            Due to the scarce of resources in the
                                                                                multicast provides an efficient solution to
            internet, multicast provides an efficient
                                                                                use these recourses fully and efficiently.
            solution to use these recourses fully and
                                                                                Because of the explosive increasing in
            efficiently. Also because of the explosive
                                                                                traffic over the Internet, multicasting
            increasing in traffic over the Internet,
                                                                                became an important issue in routing
            multicasting became an important issue in
                                                                                protocols [2, 3]. Some issues are still open
            routing protocols [2, 3]. Some issues
                                                                                such as scalability [4], billing [5], address
                                                                                allocation [6] and security [7], where the
                                                                                scalability has drawn much attention for
                                                                                Internet application.
                                                                                In this paper a protocol called SReM[1] is
                                                                                discussed and an extensive performance

            Corresponding author. E-mail address:
                                                                                                       ISSN 1947-5500
evaluation using ns2 simulator [9]is                                selection of Xcast+ is done for the following
considered. The basic idea behind SReM is                           reasons:-
to forward data between dynamically                                 • Xcast+ is widely referenced in the area of
selected nodes called Branching Node                                   explicit multicast protocols.
Routers (BNRs). Its goal is to reduce the                           • This protocol (Xcast+) is relatively close
state information kept for the multicast                               to the proposed protocol (SReM), so the
group members and to reduce the routing                                comparison will give good indication of
overhead by providing a local join/leave and                           the performance of SReM.
tree maintenance procedures using fixed size                                     PERFORMANCE METRICS
messages. Hence, it is achieving higher
degree of scalability. A detailed description                          The following metrics are used to
and the detailed SReM process in joining                            evaluate the performance of proposed
and leaving nodes can be found in [1]and a                          work:-
detailed cost analysis for this protocol can
                                                                    • Average Packet Header Size represents the
be found in [1].
                                                                      size of the header included in the data
    PERFORMANCEEVALUATION                                             packet in bytes. It represents the size of
                                                                      each data packet header in order to deliver
    Performance evolution is an idea used to                          this packet to all destinations.
measure and evaluate the performance of                             • Average End To End Delay: is the average
routing protocols. The simulation runs on                             time that takes the data packet sent by
different scenarios and evaluating different                          source node to reach its destination node.
metrics. The results obtained from this part                        Total end to end delay = SUM (time_received(pkt) – 
of evaluation are also compared with other                          time_sent(pkt)) for all data packets 
protocols in the same area of SReM.
                                                                    Avg. End To End Delay (AED) =   Total end to end 
             SIMULATION ENVIROMENT                                  delay / no. of data packets received 

The proposed protocol is implemented using                                        SIMULATION RESULTS
ns2 simulator [9] (version 2.29). The
                                                                    The simulation results will be discussed in
simulation environment consists of 60
                                                                    this section, this discussion is organized
nodes, these nodes represent the number of
                                                                    with regard to the metrics used for protocol
LMRs, and each LMR can carry any number
                                                                    evaluation. At each part, the results are
of receivers depending on its configuration.
                                                                    presented in figure form and then a
In our simulation, each LMR can connect up
                                                                    discussion and explanation for these figures
to 10 receivers so the maximum number of
                                                                    is mentioned.
receivers is 600 nodes. The number of                               Packet header size
multicast group members (LMRs) varies
from 5, 10, 15, 20, 25, 30, 35, 40, and 45                          Figure 1 shows the results obtained for the
nodes; these nodes represent LMRs in                                first part of packet header size evaluation. It
SReM and Designated Router (DR) for                                 can be noticed that SReM has got static
xcast+ protocol.                                                    header size even when the group size is
Traffic generation considered at this                               increasing. This result comes because
simulation is CBR traffic with payload size                         whatever the group size SReM will only
512 bytes. Data packets are generated at                            include the next Branching Node Router
source at a rate of 8 packets per second; this                      (BNR) addresses in the header of each data
will introduce 4096 byte per second. Each                           packet. In Xcast+ the results show that an
simulation runs for 200 second.                                     exponential increase of header size when the
In this protocol evaluation, SReM is                                group size increases. In conclusion, SReM
compared with Xcast+ protocol, the                                  improve the scalability feature because the

Corresponding author. E-mail address:
                                                                                            ISSN 1947-5500
header size is constant even when the group                                                                                                                                                   Delay
size increases.                                                                                                                                      8


                                     Header size for data packets initialized in the source                                                          6

                                                                                                                                         D elay(s)
  Header Size (Bytes)

                            100                                                                                                                      1
                             50                                                                                                                          5/50    10/100   15/150   20/200   25/250   30/300   35/350   40/400   45/450

                                                                                                                                                                          Group size(LM Rs/Receivers)
                                  5/50   10/100    15/150    20/200    25/250    30/300    35/350    40/400     45/450
                                                  Group Size (LM Rs/Receivers)
                                                                                                                                                                          Figure(3)  End to End Delay. 
Figure (1)  Extra packet header size as a function of group 
size                                                                                                                                    In this paper a performance analysis study is
The results for the second part of evaluation                                                                                           done for SReM[1]. SReM is a scalable
is shown in Figure 2, the size of packet                                                                                                multicast protocol for fixed networks. The
header for SReM increases slightly but in                                                                                               scalability issue is an important feature for
Xcast+ a high increase of packet header is                                                                                              wireless networks.
happens again. This result proofs that SReM                                                                                             The performance analysis shows that SReM
improves the scalability feature in wired                                                                                               scales well when the multicast group size
networks.                                                                                                                               becomes large. The results show that SReM
                                                                                                                                        performs a fixed header size in data packets
                                          Header size for each received data packet
                                                                                                                                        where the header size for Xcast+ protocol
                                                                                                                                        increases exponentially when the group size
      Header size (Bytes)

                            150                                                                                              xcas t++
                                                                                                                                        increases. These results show that SReM
                            100                                                                                              SReM
                                                                                                                                        improves the scalability feature in networks.
                                                                                                                                        Other results obtained shows that SReM
                                                                                                                                        performs less end to end delay and overhead
                                  5/50   10/100     15/150    20/200    25/250    30/300    35/350     40/400     45/450

                                                   Group size (LM Rs/Receivers)                                                         in addition to scalability feature.
                             Figure (2) Average header size as function of group 
size                                                                                                                                                                                 REFERENCES

End To End Delay                                                                                                                        1.                      Cao, Y. and K. Al‐Begain. SREM: A Novel 
                                                                                                                                                                Multicast       Routing         Algorithm‐
Figure 3 shows the average end to end                                                                                                                           Comprehensive  Cost  Analysis.  in5th 
delays as a function of group size with error                                                                                                                   World  Wireless  Congress  (WWC  04). 
rate because of random generation for                                                                                                                           2004. San Francisco, USA. 
simulator. The bigger value of delay the less                                                                                           2.                      Ballardie,  T.,  P.  Francis,  and  J. 
efficient the protocol. In reality, SReM                                                                                                                        Crowcroft,  Core  Based  Trees  (CBT)  An 
shows lower delay in comparison with                                                                                                                            Architecture  for  Scalable  Inter‐Domain 
Xcast+. This is because scalability feature in                                                                                                                  Multicast       Routing.        Computer 
SReM make the intermediate nodes to                                                                                                                             Communication review, 1993. 23: p. 85‐
forward the data packets faster where in                                                                                                                        85. 
Xcast+ the data packets will take high                                                                                                  3.                      Waitzman,  D.,  C.  Partridge,  and  S.E. 
processing time in the intermediate nodes                                                                                                                       Deering,  Distance  Vector  Multicast 
which lastly will increase the delay for the                                                                                                                    Routing  Protocol.RFC    Editor  1075  , 
data packets arrival at the end nodes.                                                                                                                          United States, 1988. 
                                                                                                                                        4.                      El‐Marakby,  R.  and  D.  Hutchison. 
                                                                                                                                                                Scalability  Improvement  of  the  Real‐

Corresponding author. E-mail address:
                                                                                                                                                                                             ISSN 1947-5500
         time Control Protocol (RTCP) Leading to                    avoidance at network routers using discrete and 
         Management  Facilities  in  the  Internet.                 continuous  time,  also  his  research  interests 
         inComputer Communications. 2005.                           includes  computer  networks  design  and 
5.       Dondeti,  L.,  et  al.  MBA:  a  tool  for 
                                                                    architecture, wire and wireless communication, 
         multicast  billing  and  accounting. 
         inProceedings  Sixth  IEEE  Symposium  on                  artificial  intelligence  and  expert  system, 
         Computers and Communications. 2001.                        knowledge base systems, security systems, data 
6.       Bhattacharyya,  S.,  D.  Towsley,  and  J.                 mining and information. 
         Kurose.  The  loss  path  multiplicity 
         problem  in  multicast  congestion                          
         control.in18th  Annual  Joint  Conference 
         of     the      IEEE       Computer      and 
         Communications Societies INFOCOM'99. 
7.       Judge, P. and M. Ammar, Security issues 
         and  solutions  in  multicast  content 
         distribution:  a  survey.  IEEE  Network, 
         2003. 17(1): p. 30‐36. 
8.       Al‐Begain,  K.,  Y.  Cao,  and  K.  Alameh,  A 
         DBT‐Based  Mobile  Multicast  Protocol. 
         Systems  Communications.,  2005:  p. 
9.       The  network  simulator  ‐  NS2  in 
Authors Information

                   Dr. jafar Ababneh  is  an 
                   assistant professor. He received 
                   his  PhD  degree  from  Arab 
                   Academy  for  Banking  & 
                   Financial  Sciences  (Jordan)  in 
                   2009.  He  received  his  M.Sc 
degree in computer engineering from University 
of the Yarmouk (Jordan) in 2005. He earned his 
B.Sc  in  Telecommunication  engineering  from 
University of Mu'ta (Jordan) in 1991. In 2009, he 
joined  The  World  Islamic  Sciences  and 
Education  (WISE)  University  as  a  head  of  the 
departments  of  computer  information  systems 
and  network  systems  in  the  school  of 
information technology(CISN). He has published 
many  research  papers  and  book  chapters  in 
different  fields  of  science  in  refereed  journal 
and  international  conference  proceedings.
         His  field  research  lies  in  development 
and  performance  evaluation  of  multi‐queue 
nodes  queuing  systems  for  congestion 

Corresponding author. E-mail address:
                                                                                          ISSN 1947-5500