![22 Computer Science & Information Technology ( CS & IT )
demanded in broad-band integrated services network (BISDN) and in communication-intensive
applications in parallel and distributed computing systems, such as distributed database updates
and cache coherence protocols. It is projected that multicast will also be increasingly used to
support various other interactive applications such as multimedia, teleconferencing, web servers
and electronic commerce on the Internet. [1] Many of these applications require predictable
communications performance, such as guaranteed multicast latency and bandwidth, called quality
of service (QoS) in addition to multicast capability. The QoS guarantees and the non-uniform
nature of multicast traffic make the problem of the analysis of multicast communication is very
challenging.
However, to provide a quantitative basis for the network designers, determining an optimum
value of the system parameter k (the fold number) is essential. Keeping this view in mind, we
have, in this research work, developed an analytical model to determine the suitable value of k
under different traffic loads for a k-fold multicast network under Poisson traffic with finite buffers
or queue at each node. [2] We have derived stationary distribution for the network states and then
derived expressions for the network throughput and the blocking probability of the network.
Moreover, in this paper we have shown the adjustable parameter K with finite users using
Markovian model M/M/n/n+q/N.
2. LITERATURE REVIEW
A. Background
Multicasting is a technical term which is used as a networking technique of delivering messages
and information to a group simultaneously from the source. A typical multicasting service is
shown in the Figure 1.
Figure 1: Basic multicast service
In a K-fold multicast network, fold number indicates the number of request coming from different
sources to a particular destination. On the other hand, finite queue is a data set shared by program
processes which acts as a buffer for data in multicast network. In this research work, we have
developed a model to help network engineers to design an effective multicast network. To do this,
it is necessary to get the optimum value of k in a K-fold multicast network. By implementing the](https://image.slidesharecdn.com/csit2403-180206051820/85/Achieving-Optimum-Value-of-k-in-a-K-fold-Multicast-Network-with-Buffer-using-M-M-n-n-q-N-Traffic-Model-2-320.jpg)
![Computer Science & Information Technology ( CS & IT ) 23
optimum value of the system parameter k in K-fold multicast network, we improved the network
performance by optimizing network throughput, where throughput is the number of messages
successfully delivered per unit time. In this paper, the term “throughput” has been used to
measure from the arrival of the first bit of data at the receiver. [3]
As mentioned earlier, the primary target of this research work was to evaluate the performance of
K-fold multicast network by using traffic model. For this reason, throughout this paper, we make
following assumptions on the multicast traffic we consider.
• The probability of a destination node being involved in an incoming multicast connection
request is independent of other destination nodes.
• Multicast connection requests at different source nodes are independent to each other.
• Holding time of each multicast connection is exponentially distributed with parameter
and is independent to each other.
• Multicast connection requests arrive at each source node according to a Poisson process
with intensity and are independent to each other. [4]
B. Previous Researches
This research work is basically the extension of previous works of Zhenghao Zhang et.al [3] [5]
who evaluated the performance of k fold network but they did not use buffers. After that Asfara
R. Towfiq et.al [6] again checked the performance of K-fold network with a new look. They
show the optimization of K-fold multicast network with buffers but for infinite users. Here, we
have used finite users to evaluate performance of k-fold multicast network by using Markovian
model.
For this approach, a destination node may be simultaneously involved in two multicast
connections. Such connections will be blocked in a network which is designed to be nonblocking
or rearrange able for only multicast assignments. Specifically, the network can realize multiple
multicast assignments in a single pass with a guaranteed latency.
3. TRAFFIC THEORYAND K-FOLD MULTICAST NETWORK
C. Basic Traffic theory and Markov Chain
Traffic Theory describes the key models of traffic flow and associated traffic phenomena such as
conflicts in traffic, congestion control and effective management of traffic.
In this paper, we have derived stationary distribution of the K- fold network from which we can
obtain network throughput and the blocking probability. We assume the Markovian
M/M/n/n+q/N model which is shown in the Figure 2.](https://image.slidesharecdn.com/csit2403-180206051820/85/Achieving-Optimum-Value-of-k-in-a-K-fold-Multicast-Network-with-Buffer-using-M-M-n-n-q-N-Traffic-Model-3-320.jpg)
![24 Computer Science & Information Technology ( CS & IT )
Figure 2: Markov Chain for k-fold network.
The above figure shows a glance of Markov Chain and its impact on Finite State space.
D. K-fold Network
It is defined as a mapping from a subset of network source nodes to a subset of network
destination nodes, with up to K-fold overlapping allowed among the destinations of different
sources. It is an adjustable parameter. In other words, any destination node can be involved in
multicast connections from up to K different sources at a time. [7]
Why A k-fold network?
• A cost-effective solution to provide better quality-of-service functions in supporting
real-world multicast applications.
• Predictable communications performance, such as guaranteed multicast latency and
bandwidth.
• Highly demanded in communication-intensive applications in parallel and distributed
computing systems, such as distributed database updates. [8]
E. Kendal’s notation of queuing system
In 1953 D.G Kendall introduces special notation for queuing models. A complete notation for the
paper is:
M/M/n/K/N
where,
M: Markov or memory less which follows exponential distribution
N: Number of servers/channels](https://image.slidesharecdn.com/csit2403-180206051820/85/Achieving-Optimum-Value-of-k-in-a-K-fold-Multicast-Network-with-Buffer-using-M-M-n-n-q-N-Traffic-Model-4-320.jpg)
![Computer Science & Information Technology ( CS & IT ) 25
n: Number of users
K=n+q: Sum of channels and queue
q: Length of queue
4. MATHEMATICALANALYSIS
Let us consider that there are j multicast connection requests, and let ),(deg mjp be the
probability that a destination node is the destination of exactly m of the multicast connection
requests; or we can say that a destination node is of degree m under these j multicast connection
requests. The probability that any multicast connection request chooses this destination node is θ
and is independent of other multicast connections. Thus, we have
{ }jm
m
j
mjp mjm
.,..,1,0,)1(),(deg ∈−
= −
θθ (1)
which is a binomial random variable. We assume that each destination node has the same
distribution given by (1).
Furthermore, we assume that whether a destination node is chosen by a multicast connection is
independent of other destination nodes. Thus, in addition to having the same distributions, the
degrees of the destination nodes are also independent of each other. That is why, they are a group
of independent, identically distributed (i.i.d.) random variables. [9]
F. Mathematical Analysis of network throughput
Let )(mc jP be the probability that j multicast connection requests are mutually compatible (m.c)
in a k-fold multicast network. We note that a set of multicast connection requests are m.c. when
none of the destination nodes has a degree more than k when realized simultaneously in the
network. From (1), it is obvious that the probability of a destination node having a degree less
than or equal to k is ∑ =
k
m
mjp
0 deg ),( for kj > , and 1 for kj ≤ , because when kj ≤ , no
destination node can have a degree more than k. Since the degrees of destination nodes are
independent of each other, we have
>
= ∑ =
.otherwise,1
,),(
)( 0 deg
mc
kjmjp
jP
n
k
m (2)
Now, let us consider that a new multicast connection request arrives when j multicast connections
are already in the network. If this new connection can be realized along with those ongoing
connections, we say that it can join the ongoing connections. Let )(jn jP be the probability that a
new multicast connection can join j ongoing connections. It can be shown that
.
)(
)1(
)(
mc
mc
jn
jP
jP
jP
+
= (3)](https://image.slidesharecdn.com/csit2403-180206051820/85/Achieving-Optimum-Value-of-k-in-a-K-fold-Multicast-Network-with-Buffer-using-M-M-n-n-q-N-Traffic-Model-5-320.jpg)
![26 Computer Science & Information Technology ( CS & IT )
By solving the Markov chain of Figure 2, the stationary states are found to have the probabilities
nrPxPP
r
x
jn
N
r
r ≤≤
= ∏
−
=
0,)( 0
1
0
τ
ρ
Total number of times the network departs from state i due to the arrival of a successful
connection request is,
)()( jPjqnTP jnj λ−+
This is also the total numbers of successful connection requests among at the network when the
network is in state j(j∈{0,1….,N}) during [0,T] Therefore the total number of successful
connection requests carried by the network during [0,T] is obtained by summing average,
)()(
0
jPjqnPTN jn
qn
j
jsucc −+= ∑
+
=
λ
Therefore the network Throughput is,
∑
+
=
−+==
qn
j
jnj
succ
H jqnjpp
T
N
T
0
))((λ (4)
G. Mathematical Analysis of blocking probability
The total number of connection requests arriving at the network during [0, T]
TqnNtotal λ)( +=
Thus the Blocking Probability,
−+
+
−== ∑
+
=
qn
j
jjn
total
b
B PjPjqn
qnN
N
P
0
1
)()(
1
1
(5)
H. Mathematical Analysis of Probability of delay
The Probability of Delay is,
∑=
+=
q
s
snD PP
1
0
1
01
)(
!
)!( Plp
n
P
n
P
sn
sn
l
jn
N
sn
sq
s
n
∏∑
−+
=+=
+=
(6)
5. RESULTS AND DISCUSSIONS
For numerical appreciation of our results, we have plotted in Figs. (3), (4) and (5), the throughput
and the blocking probability as a function of the fold number k. [10]](https://image.slidesharecdn.com/csit2403-180206051820/85/Achieving-Optimum-Value-of-k-in-a-K-fold-Multicast-Network-with-Buffer-using-M-M-n-n-q-N-Traffic-Model-6-320.jpg)
![Computer Science & Information Technology ( CS & IT ) 27
It is seen from Figure 3 that if the fold of the network is increased, network throughput increases
very fast in the lower values of the system parameter k, in our study up to k=5; beyond this value
of k, the network throughput is almost constant with respect to the system parameter k for
particular offered traffic. We also observe that as the offered traffic is increased, the throughput
also increases. [11]
Figure 3: Network throughput as a function of the fold number under different offered traffic
(N=50, n=14, q= 5, )31.=θ
Figure 4 shows the variation of the blocking probability with respect to the fold k. It is seen from
this figure that as the system parameter k increases, the blocking probability decreases. However,
after an optimum value of k, in our present study it is ~ 5, the blocking probability remains
constant for particular value of the offered traffic.
Figure 4: Blocking Probability as a function of the fold number under different offered traffic
(N=50, n=14, q=5, )31.=θ
Figure 5 shows, the variation of the probability of delayed service with respect to the fold number
k. It is observed that the probability of delay is almost negligible for lower values of the fold
number k, whereas, it is suddenly increases as the fold number approaches the optimum value k ~
5. However, after a certain value of k, the probability of delay becomes constant. [12]](https://image.slidesharecdn.com/csit2403-180206051820/85/Achieving-Optimum-Value-of-k-in-a-K-fold-Multicast-Network-with-Buffer-using-M-M-n-n-q-N-Traffic-Model-7-320.jpg)
![Computer Science & Information Technology ( CS & IT ) 29
ACKNOWLEDGMENT
Authors of this paper would like to thank Prof. Dr. Md. Imdadul Islam, Department of Computer
Science and Engineering, Jahangirnagar University, Savar, Dhaka for his helpful guidance in this
research work.
REFERENCES
[1] Md. Syeful Islam, Md. Rezaur Rahman, Anupam Roy, Md. Imdadul Islam, and M. R. Amin,
"Performance Evaluation of a Finite Queue Switching Under Two-Dimensional M/G/1(m) Traffic",
International Journal of Imaging Systems (IJIS), (Korea), vol. 7, no. 4, pp. 679-690, 2011.
[2] Anupam Roy, Md. Imdadul Islam, and M. R. Amin, "MMPP+M/D/1 Traffic Model in Video-Data
Integrated Service under ATM System", International J. Engineering and Technology (Singapore),,
vol. 3, no. 6, pp. 615-620, 2011
[3] Zhenghao Zhang and Y. Yang, “Performance analysis of k-fold multicast networks”, IEEE Trans.
Commun., vol. 53, pp. 308-314, Feb. 2005.
[4] Abu Sayed Rahaman, Md. Imdadul Islam and M. R. Amin, "Performance Evaluation of Two-Hop
Wireless Link Under Rayleigh and Nakagami-m Fading Channel", International Journal of
Engineering and Technology (IJET), (Singapore), vol. 3, no. 5, pp. 454-459, 2011
[5] Y. Yang and G. M. Masson, “Nonblocking broadcast switching networks”, IEEE Trans. Comput.,
vol. 40, pp. 1005-1015, Sep. 1991.
[6] Asfara R. Rafiq, N.A. Siddiky, Md. Imdadul Islam, M.R.Amin,” Optimization of k- fold wireless
network using M/M/n/n+q Traffic Model” ICECE 2008, Buet, Dhaka, Paper ID ICECE 125 2008.
[7] MM Hasan, Most Tajmary Mahfuz, Woakil Uddin Ahamed, Syed Akhter Hossain, “ Requirement
Analysis for Context-Awareness to Create a Digital Persona in eHealth based Pervasive Intelligent
Environment”, Journal of Information and Communication Technologies (ISSN 2047-3168), vol. 2,
issue 3, pp.11-15, United Kingdom, 2012.
[8] P. Feldman, J. Friedman, and N. Pippenger, “Wide-sense nonblocking networks”, SIAM J. Discr.
Math., vol. 1, no. 2, pp. 158-173, May 1988.
[9] N. McKeown, A. Mekkittijul, V. Anantharam, and J. Walrand, “Achieving 100% throughput in an
input-queued switch”, IEEE Trans. Commun. , vol. 47, pp. 1260-1267, Oct. 1999.
[10] M. Andrews, S. Khanna, and K. Kumaran, “Integrated scheduling of unicast and multicast traffic in
an input-queued switch”, in Proc. IEEE INFOCOM, 1999, pp. 1144-1151.
[11] Y. Yang and J. Wang, “On blocking probability of multicast networks”, IEEE Trans. Commun., vol.
46, pp. 957-968, Jul. 1998.
[12] Y. Yang and J. Wang, “Nonblocking k-fold multicast networks”, IEEE Trans. Parallel Distrib. Syst.,
vol. 14, pp. 131-141, Feb. 2003.](https://image.slidesharecdn.com/csit2403-180206051820/85/Achieving-Optimum-Value-of-k-in-a-K-fold-Multicast-Network-with-Buffer-using-M-M-n-n-q-N-Traffic-Model-9-320.jpg)
Achieving Optimum Value of k in a K-fold Multicast Network with Buffer using M/M/n/n+q/N Traffic Model
- 1. Sundarapandian et al. (Eds): CoNeCo,WiMo, NLP, CRYPSIS, ICAIT, ICDIP, ITCSE, CS & IT 07, pp. 21–30, 2012. © CS & IT-CSCP 2012 DOI : 10.5121/csit.2012.2403 Achieving Optimum Value of k in a K-fold Multicast Network with Buffer using M/M/n/n+q/N Traffic Model Md. Mahmudul Hasan1 , Most. Tajmary Mahfuz2 , Mohamed Ruhul Amin3 1 Department of CSE/CIS/CS, Daffodil International University, Dhaka, Bangladesh 1 mhasan@daffodilvarsity.edu.bd 2 Department of Natural Science, Daffodil International University, Dhaka, Bangladesh 2 tajmary@daffodilvarsity.edu.bd 3 Department of Electronics and Communications Engineering, East West University, Dhaka, Bangladesh 3 ramin@ewubd.edu Abstract Multicast network is widely used for effective communication, transmission and performance optimizations of a network. In this paper, a new model has been developed to determine a suitable value of the fold k of a k-fold multicast network under different traffic loads under Poisson traffic with finite queue at each node. We have derived stationary distribution for the network states and then derived expressions for the network throughput and the blocking probability of the network. It has been found in this research work that the network throughput increases very fast as we increase the fold number. However, at a certain value of the fold, the blocking probability ceases to increase and it remains constant. We have also observed that as the offered traffic is increased, the throughput also increases. Moreover, the system parameter k is increased, the blocking probability decreases. However, after an optimum value of k, the blocking probability remains constant for a particular value of the offered traffic. In fact, in this paper, by evaluating the performance of a k-fold multicast network, our developed model improves the performance of a multicast network. Keywords K-fold network, Kendal’s notation, Markov chain, Multicasting, Throughput, Traffic Theory 1. INTRODUCTION Multicast involves transmitting information from a single source to multiple destinations. This is an important requirement for high-performance communications networks. Multicast communication is one of the most important collective communication operations and is highly
- 2. 22 Computer Science & Information Technology ( CS & IT ) demanded in broad-band integrated services network (BISDN) and in communication-intensive applications in parallel and distributed computing systems, such as distributed database updates and cache coherence protocols. It is projected that multicast will also be increasingly used to support various other interactive applications such as multimedia, teleconferencing, web servers and electronic commerce on the Internet. [1] Many of these applications require predictable communications performance, such as guaranteed multicast latency and bandwidth, called quality of service (QoS) in addition to multicast capability. The QoS guarantees and the non-uniform nature of multicast traffic make the problem of the analysis of multicast communication is very challenging. However, to provide a quantitative basis for the network designers, determining an optimum value of the system parameter k (the fold number) is essential. Keeping this view in mind, we have, in this research work, developed an analytical model to determine the suitable value of k under different traffic loads for a k-fold multicast network under Poisson traffic with finite buffers or queue at each node. [2] We have derived stationary distribution for the network states and then derived expressions for the network throughput and the blocking probability of the network. Moreover, in this paper we have shown the adjustable parameter K with finite users using Markovian model M/M/n/n+q/N. 2. LITERATURE REVIEW A. Background Multicasting is a technical term which is used as a networking technique of delivering messages and information to a group simultaneously from the source. A typical multicasting service is shown in the Figure 1. Figure 1: Basic multicast service In a K-fold multicast network, fold number indicates the number of request coming from different sources to a particular destination. On the other hand, finite queue is a data set shared by program processes which acts as a buffer for data in multicast network. In this research work, we have developed a model to help network engineers to design an effective multicast network. To do this, it is necessary to get the optimum value of k in a K-fold multicast network. By implementing the
- 3. Computer Science & Information Technology ( CS & IT ) 23 optimum value of the system parameter k in K-fold multicast network, we improved the network performance by optimizing network throughput, where throughput is the number of messages successfully delivered per unit time. In this paper, the term “throughput” has been used to measure from the arrival of the first bit of data at the receiver. [3] As mentioned earlier, the primary target of this research work was to evaluate the performance of K-fold multicast network by using traffic model. For this reason, throughout this paper, we make following assumptions on the multicast traffic we consider. • The probability of a destination node being involved in an incoming multicast connection request is independent of other destination nodes. • Multicast connection requests at different source nodes are independent to each other. • Holding time of each multicast connection is exponentially distributed with parameter and is independent to each other. • Multicast connection requests arrive at each source node according to a Poisson process with intensity and are independent to each other. [4] B. Previous Researches This research work is basically the extension of previous works of Zhenghao Zhang et.al [3] [5] who evaluated the performance of k fold network but they did not use buffers. After that Asfara R. Towfiq et.al [6] again checked the performance of K-fold network with a new look. They show the optimization of K-fold multicast network with buffers but for infinite users. Here, we have used finite users to evaluate performance of k-fold multicast network by using Markovian model. For this approach, a destination node may be simultaneously involved in two multicast connections. Such connections will be blocked in a network which is designed to be nonblocking or rearrange able for only multicast assignments. Specifically, the network can realize multiple multicast assignments in a single pass with a guaranteed latency. 3. TRAFFIC THEORYAND K-FOLD MULTICAST NETWORK C. Basic Traffic theory and Markov Chain Traffic Theory describes the key models of traffic flow and associated traffic phenomena such as conflicts in traffic, congestion control and effective management of traffic. In this paper, we have derived stationary distribution of the K- fold network from which we can obtain network throughput and the blocking probability. We assume the Markovian M/M/n/n+q/N model which is shown in the Figure 2.
- 4. 24 Computer Science & Information Technology ( CS & IT ) Figure 2: Markov Chain for k-fold network. The above figure shows a glance of Markov Chain and its impact on Finite State space. D. K-fold Network It is defined as a mapping from a subset of network source nodes to a subset of network destination nodes, with up to K-fold overlapping allowed among the destinations of different sources. It is an adjustable parameter. In other words, any destination node can be involved in multicast connections from up to K different sources at a time. [7] Why A k-fold network? • A cost-effective solution to provide better quality-of-service functions in supporting real-world multicast applications. • Predictable communications performance, such as guaranteed multicast latency and bandwidth. • Highly demanded in communication-intensive applications in parallel and distributed computing systems, such as distributed database updates. [8] E. Kendal’s notation of queuing system In 1953 D.G Kendall introduces special notation for queuing models. A complete notation for the paper is: M/M/n/K/N where, M: Markov or memory less which follows exponential distribution N: Number of servers/channels
- 5. Computer Science & Information Technology ( CS & IT ) 25 n: Number of users K=n+q: Sum of channels and queue q: Length of queue 4. MATHEMATICALANALYSIS Let us consider that there are j multicast connection requests, and let ),(deg mjp be the probability that a destination node is the destination of exactly m of the multicast connection requests; or we can say that a destination node is of degree m under these j multicast connection requests. The probability that any multicast connection request chooses this destination node is θ and is independent of other multicast connections. Thus, we have { }jm m j mjp mjm .,..,1,0,)1(),(deg ∈− = − θθ (1) which is a binomial random variable. We assume that each destination node has the same distribution given by (1). Furthermore, we assume that whether a destination node is chosen by a multicast connection is independent of other destination nodes. Thus, in addition to having the same distributions, the degrees of the destination nodes are also independent of each other. That is why, they are a group of independent, identically distributed (i.i.d.) random variables. [9] F. Mathematical Analysis of network throughput Let )(mc jP be the probability that j multicast connection requests are mutually compatible (m.c) in a k-fold multicast network. We note that a set of multicast connection requests are m.c. when none of the destination nodes has a degree more than k when realized simultaneously in the network. From (1), it is obvious that the probability of a destination node having a degree less than or equal to k is ∑ = k m mjp 0 deg ),( for kj > , and 1 for kj ≤ , because when kj ≤ , no destination node can have a degree more than k. Since the degrees of destination nodes are independent of each other, we have > = ∑ = .otherwise,1 ,),( )( 0 deg mc kjmjp jP n k m (2) Now, let us consider that a new multicast connection request arrives when j multicast connections are already in the network. If this new connection can be realized along with those ongoing connections, we say that it can join the ongoing connections. Let )(jn jP be the probability that a new multicast connection can join j ongoing connections. It can be shown that . )( )1( )( mc mc jn jP jP jP + = (3)
- 6. 26 Computer Science & Information Technology ( CS & IT ) By solving the Markov chain of Figure 2, the stationary states are found to have the probabilities nrPxPP r x jn N r r ≤≤ = ∏ − = 0,)( 0 1 0 τ ρ Total number of times the network departs from state i due to the arrival of a successful connection request is, )()( jPjqnTP jnj λ−+ This is also the total numbers of successful connection requests among at the network when the network is in state j(j∈{0,1….,N}) during [0,T] Therefore the total number of successful connection requests carried by the network during [0,T] is obtained by summing average, )()( 0 jPjqnPTN jn qn j jsucc −+= ∑ + = λ Therefore the network Throughput is, ∑ + = −+== qn j jnj succ H jqnjpp T N T 0 ))((λ (4) G. Mathematical Analysis of blocking probability The total number of connection requests arriving at the network during [0, T] TqnNtotal λ)( += Thus the Blocking Probability, −+ + −== ∑ + = qn j jjn total b B PjPjqn qnN N P 0 1 )()( 1 1 (5) H. Mathematical Analysis of Probability of delay The Probability of Delay is, ∑= += q s snD PP 1 0 1 01 )( ! )!( Plp n P n P sn sn l jn N sn sq s n ∏∑ −+ =+= += (6) 5. RESULTS AND DISCUSSIONS For numerical appreciation of our results, we have plotted in Figs. (3), (4) and (5), the throughput and the blocking probability as a function of the fold number k. [10]
- 7. Computer Science & Information Technology ( CS & IT ) 27 It is seen from Figure 3 that if the fold of the network is increased, network throughput increases very fast in the lower values of the system parameter k, in our study up to k=5; beyond this value of k, the network throughput is almost constant with respect to the system parameter k for particular offered traffic. We also observe that as the offered traffic is increased, the throughput also increases. [11] Figure 3: Network throughput as a function of the fold number under different offered traffic (N=50, n=14, q= 5, )31.=θ Figure 4 shows the variation of the blocking probability with respect to the fold k. It is seen from this figure that as the system parameter k increases, the blocking probability decreases. However, after an optimum value of k, in our present study it is ~ 5, the blocking probability remains constant for particular value of the offered traffic. Figure 4: Blocking Probability as a function of the fold number under different offered traffic (N=50, n=14, q=5, )31.=θ Figure 5 shows, the variation of the probability of delayed service with respect to the fold number k. It is observed that the probability of delay is almost negligible for lower values of the fold number k, whereas, it is suddenly increases as the fold number approaches the optimum value k ~ 5. However, after a certain value of k, the probability of delay becomes constant. [12]
- 8. 28 Computer Science & Information Technology ( CS & IT ) Figure 5: Probability of delay as a function of the fold number under different offered traffic (N=50, n=14, q= 5, )31.=θ 6. CONCLUSIONS Determining the system parameter k (the fold number) and finding its optimum value is must to design an effective network. To keep this view in mind, we have developed a systematical model to determine an optimum value for a k-fold multicast network under Poisson traffic with finite queue at each node. We have derived stationary distribution for the network states and then derived expressions for the network throughput and the blocking probability of the network. It has been found in this study that the network throughput increases very fast as we increase the fold number. However, at a certain value of the fold, the blocking probability ceases to increase and it remains constant. We have also observed that as the offered traffic is increased, the throughput also increases. In addition, it has been observed that the blocking probability decreased proportionately based on the traffic when the system parameter k in a K-fold multicast network is increased. However, after a suitable value of k, which is approximately 5 based on our research in this paper, the blocking probability remains constant for specific value of the offered traffic in multicast network. Note that although K-fold multicast assignments can be realized by simply stacking k copies of one fold network together, the k–fold network designed in has a much lower hardware cost. In fact, the cost of the former is about 3-k times of a k-fold network for any k. Thus, a k-fold network is a cost effective choice to provide better QoS functions in supporting arbitrary multicast communication. To sum up, this model could help to find out more suitable and appropriate value of the system parameter k in a K-fold multicast network to increase throughput of the network. Moreover, this model can be used for designing an effective ubiquitous network in future.
- 9. Computer Science & Information Technology ( CS & IT ) 29 ACKNOWLEDGMENT Authors of this paper would like to thank Prof. Dr. Md. Imdadul Islam, Department of Computer Science and Engineering, Jahangirnagar University, Savar, Dhaka for his helpful guidance in this research work. REFERENCES [1] Md. Syeful Islam, Md. Rezaur Rahman, Anupam Roy, Md. Imdadul Islam, and M. R. Amin, "Performance Evaluation of a Finite Queue Switching Under Two-Dimensional M/G/1(m) Traffic", International Journal of Imaging Systems (IJIS), (Korea), vol. 7, no. 4, pp. 679-690, 2011. [2] Anupam Roy, Md. Imdadul Islam, and M. R. Amin, "MMPP+M/D/1 Traffic Model in Video-Data Integrated Service under ATM System", International J. Engineering and Technology (Singapore),, vol. 3, no. 6, pp. 615-620, 2011 [3] Zhenghao Zhang and Y. Yang, “Performance analysis of k-fold multicast networks”, IEEE Trans. Commun., vol. 53, pp. 308-314, Feb. 2005. [4] Abu Sayed Rahaman, Md. Imdadul Islam and M. R. Amin, "Performance Evaluation of Two-Hop Wireless Link Under Rayleigh and Nakagami-m Fading Channel", International Journal of Engineering and Technology (IJET), (Singapore), vol. 3, no. 5, pp. 454-459, 2011 [5] Y. Yang and G. M. Masson, “Nonblocking broadcast switching networks”, IEEE Trans. Comput., vol. 40, pp. 1005-1015, Sep. 1991. [6] Asfara R. Rafiq, N.A. Siddiky, Md. Imdadul Islam, M.R.Amin,” Optimization of k- fold wireless network using M/M/n/n+q Traffic Model” ICECE 2008, Buet, Dhaka, Paper ID ICECE 125 2008. [7] MM Hasan, Most Tajmary Mahfuz, Woakil Uddin Ahamed, Syed Akhter Hossain, “ Requirement Analysis for Context-Awareness to Create a Digital Persona in eHealth based Pervasive Intelligent Environment”, Journal of Information and Communication Technologies (ISSN 2047-3168), vol. 2, issue 3, pp.11-15, United Kingdom, 2012. [8] P. Feldman, J. Friedman, and N. Pippenger, “Wide-sense nonblocking networks”, SIAM J. Discr. Math., vol. 1, no. 2, pp. 158-173, May 1988. [9] N. McKeown, A. Mekkittijul, V. Anantharam, and J. Walrand, “Achieving 100% throughput in an input-queued switch”, IEEE Trans. Commun. , vol. 47, pp. 1260-1267, Oct. 1999. [10] M. Andrews, S. Khanna, and K. Kumaran, “Integrated scheduling of unicast and multicast traffic in an input-queued switch”, in Proc. IEEE INFOCOM, 1999, pp. 1144-1151. [11] Y. Yang and J. Wang, “On blocking probability of multicast networks”, IEEE Trans. Commun., vol. 46, pp. 957-968, Jul. 1998. [12] Y. Yang and J. Wang, “Nonblocking k-fold multicast networks”, IEEE Trans. Parallel Distrib. Syst., vol. 14, pp. 131-141, Feb. 2003.
- 10. 30 Computer Science & Information Technology ( CS & IT ) Authors Md. Mahmudul Hasan is currently serving himself as a lecturer in Daffodil International University, Bangladesh. He has completed his MSc in Computer Science from University of Essex, UK and worked as a research assistant in International Development Academy, UoE, UK. He has achieved distinction in Pervasive Computing and Ambient Intelligence at MSc level. Email: mhasan@daffodilvarsity.edu.bd Most. Tajmary Mahfuz is working as a lecturer in Daffodil International University and has completed her MS in Statistics from Jahangirnagar University, Dhaka. She has strong background in field level researching and served herself as a Survey Research Analyst in the project of SME foundation under Maxwell Stamp Ltd. She also has experiences on survey based project such as School Mapping of IFC contracted by ADSL. Email: tajmary@daffodilvarsity.edu.bd Dr. Mohamed Ruhul Amin is serving himself as a professor in the department of Electronics and Communications Engineering at East West University, Dhaka. He did his Ph.D. in Plasma Physics from University of St Andrews, Scotland, 1990. He was Dean of Faculty of Science and Engineering at East West University. Email: rami n@ewubd.edu